[연세대][PAXBD] 한국지방소멸_모델링.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 시군구별 소멸위험지수 프로젝트_모델링"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "목차:\n",
    "\n",
    "**[I. 데이터 불러오기](#one)**\n",
    "\n",
    "**[II. 타겟 변수 시각화](#two)**\n",
    "- [a. folium](#two-a)\n",
    "- [b. Cartogram](#two-b)\n",
    "\n",
    "**[III. EDA & FE](#three)**\n",
    "\n",
    "- [a. 결측치확인](#three-a)\n",
    "- [b. 타겟변수](#three-b)\n",
    "- [c. 지표비교](#three-c)\n",
    "- [d. 파생변수추가](#three-d)\n",
    "\n",
    "**[IV. 모델링](#four)**\n",
    "- [a. K-fold](#four-a)\n",
    "- [b. GradientBoosting](#four-b)\n",
    "- [c. RandomForest](#four-c)\n",
    "- [d. Xgboost](#four-d)\n",
    "- [e. 앙상블](#four-e)\n",
    "\n",
    "**[V. 결과](#five)**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## IV. 모델링  <a id=\"four\"></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "#라이브러리 설치 - 해당 라이브러리가 없다면 아래 코드를 이용해서 설치해주시기 바랍니다.\n",
    "\n",
    "#!pip install xgboost\n",
    "#!pip install lightgbm\n",
    "#!pip install eli5"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    ">버전 정보 python 3.7\n",
    ">\n",
    ">pandas:  1.1.2\n",
    ">\n",
    ">matplotlib:  3.3.2\n",
    ">\n",
    ">seaborn:  0.11.0\n",
    ">\n",
    ">folium:  0.11.0\n",
    ">\n",
    ">plotly:  4.10.0\n",
    ">\n",
    ">sklearn:  0.23.2\n",
    ">\n",
    ">patsy:  0.5.1\n",
    ">\n",
    ">statsmodels:  0.12.0\n",
    ">\n",
    ">xgboost:  1.2.0\n",
    ">\n",
    ">lightgbm:  3.0.0\n",
    ">\n",
    ">eli5:  0.10.1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\", category=FutureWarning)\n",
    "warnings.filterwarnings(\"ignore\", category=DeprecationWarning)\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "from sklearn.metrics import f1_score\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.model_selection import cross_validate \n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.neural_network import MLPClassifier\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.ensemble import GradientBoostingClassifier\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.svm import SVC\n",
    "from xgboost import XGBClassifier\n",
    "from lightgbm import LGBMClassifier\n",
    "\n",
    "from sklearn.model_selection import cross_val_score, StratifiedKFold\n",
    "\n",
    "from sklearn.metrics import confusion_matrix, precision_recall_curve, classification_report, precision_score, recall_score, accuracy_score\n",
    "from sklearn.model_selection import RandomizedSearchCV\n",
    "\n",
    "from sklearn.utils.class_weight import compute_sample_weight\n",
    "from sklearn.utils import class_weight\n",
    "from sklearn.ensemble import VotingClassifier\n",
    "from scipy.stats import randint as sp_randint\n",
    "import xgboost as xgb\n",
    "import eli5 \n",
    "from eli5.sklearn import PermutationImportance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "train = pd.read_csv('모델링_train.csv',encoding='cp949')\n",
    "test = pd.read_csv('모델링_test.csv', encoding='cp949')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "dat1 = train.set_index('Unnamed: 0')\n",
    "test_x1 = test.set_index('Unnamed: 0')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>sido</th>\n",
       "      <th>sigun_nm</th>\n",
       "      <th>sigun_code</th>\n",
       "      <th>병원</th>\n",
       "      <th>보건소</th>\n",
       "      <th>상급종합병원</th>\n",
       "      <th>약국</th>\n",
       "      <th>요양병원</th>\n",
       "      <th>의원</th>\n",
       "      <th>종합병원</th>\n",
       "      <th>...</th>\n",
       "      <th>부동산업_종사자수</th>\n",
       "      <th>전문과학기술서비스업_사업체수</th>\n",
       "      <th>전문과학기술서비스업_종사자수</th>\n",
       "      <th>target</th>\n",
       "      <th>등급</th>\n",
       "      <th>평균연령x인구증가율</th>\n",
       "      <th>의료복지기관</th>\n",
       "      <th>가구수당x합계출산율</th>\n",
       "      <th>가구수당_평균연령</th>\n",
       "      <th>녹지지역비율x토지거래면적</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Unnamed: 0</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>강원도강릉시</th>\n",
       "      <td>강원도</td>\n",
       "      <td>강릉시</td>\n",
       "      <td>42150</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>81</td>\n",
       "      <td>4</td>\n",
       "      <td>103</td>\n",
       "      <td>4</td>\n",
       "      <td>...</td>\n",
       "      <td>1196</td>\n",
       "      <td>277</td>\n",
       "      <td>1677</td>\n",
       "      <td>1.423144</td>\n",
       "      <td>b</td>\n",
       "      <td>-22.36</td>\n",
       "      <td>194</td>\n",
       "      <td>108137.888</td>\n",
       "      <td>0.049069</td>\n",
       "      <td>8009.9701</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>강원도고성군</th>\n",
       "      <td>강원도</td>\n",
       "      <td>고성군</td>\n",
       "      <td>42820</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>13</td>\n",
       "      <td>0</td>\n",
       "      <td>8</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>33</td>\n",
       "      <td>20</td>\n",
       "      <td>94</td>\n",
       "      <td>1.143675</td>\n",
       "      <td>b</td>\n",
       "      <td>-39.10</td>\n",
       "      <td>22</td>\n",
       "      <td>15727.952</td>\n",
       "      <td>0.369686</td>\n",
       "      <td>4793.3755</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>강원도동해시</th>\n",
       "      <td>강원도</td>\n",
       "      <td>동해시</td>\n",
       "      <td>42170</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>41</td>\n",
       "      <td>3</td>\n",
       "      <td>50</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>560</td>\n",
       "      <td>65</td>\n",
       "      <td>378</td>\n",
       "      <td>1.377623</td>\n",
       "      <td>b</td>\n",
       "      <td>-29.24</td>\n",
       "      <td>97</td>\n",
       "      <td>48261.440</td>\n",
       "      <td>0.119391</td>\n",
       "      <td>2261.6460</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>강원도삼척시</th>\n",
       "      <td>강원도</td>\n",
       "      <td>삼척시</td>\n",
       "      <td>42230</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>29</td>\n",
       "      <td>1</td>\n",
       "      <td>26</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>326</td>\n",
       "      <td>59</td>\n",
       "      <td>462</td>\n",
       "      <td>1.191157</td>\n",
       "      <td>b</td>\n",
       "      <td>-84.92</td>\n",
       "      <td>58</td>\n",
       "      <td>37995.797</td>\n",
       "      <td>0.151354</td>\n",
       "      <td>7782.4944</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>강원도속초시</th>\n",
       "      <td>강원도</td>\n",
       "      <td>속초시</td>\n",
       "      <td>42210</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>43</td>\n",
       "      <td>2</td>\n",
       "      <td>58</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>694</td>\n",
       "      <td>81</td>\n",
       "      <td>477</td>\n",
       "      <td>1.438448</td>\n",
       "      <td>b</td>\n",
       "      <td>-11.18</td>\n",
       "      <td>105</td>\n",
       "      <td>49043.944</td>\n",
       "      <td>0.130287</td>\n",
       "      <td>739.1310</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 101 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           sido sigun_nm  sigun_code  병원  보건소  상급종합병원  약국  요양병원   의원  종합병원  \\\n",
       "Unnamed: 0                                                                   \n",
       "강원도강릉시      강원도      강릉시       42150   5    1       0  81     4  103     4   \n",
       "강원도고성군      강원도      고성군       42820   0    1       0  13     0    8     0   \n",
       "강원도동해시      강원도      동해시       42170   2    1       0  41     3   50     1   \n",
       "강원도삼척시      강원도      삼척시       42230   1    1       0  29     1   26     1   \n",
       "강원도속초시      강원도      속초시       42210   1    1       0  43     2   58     2   \n",
       "\n",
       "            ...  부동산업_종사자수  전문과학기술서비스업_사업체수  전문과학기술서비스업_종사자수    target  등급  \\\n",
       "Unnamed: 0  ...                                                              \n",
       "강원도강릉시      ...       1196              277             1677  1.423144   b   \n",
       "강원도고성군      ...         33               20               94  1.143675   b   \n",
       "강원도동해시      ...        560               65              378  1.377623   b   \n",
       "강원도삼척시      ...        326               59              462  1.191157   b   \n",
       "강원도속초시      ...        694               81              477  1.438448   b   \n",
       "\n",
       "            평균연령x인구증가율  의료복지기관  가구수당x합계출산율  가구수당_평균연령  녹지지역비율x토지거래면적  \n",
       "Unnamed: 0                                                            \n",
       "강원도강릉시          -22.36     194  108137.888   0.049069      8009.9701  \n",
       "강원도고성군          -39.10      22   15727.952   0.369686      4793.3755  \n",
       "강원도동해시          -29.24      97   48261.440   0.119391      2261.6460  \n",
       "강원도삼척시          -84.92      58   37995.797   0.151354      7782.4944  \n",
       "강원도속초시          -11.18     105   49043.944   0.130287       739.1310  \n",
       "\n",
       "[5 rows x 101 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dat1.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>sido</th>\n",
       "      <th>sigun_nm</th>\n",
       "      <th>sigun_code</th>\n",
       "      <th>병원</th>\n",
       "      <th>보건소</th>\n",
       "      <th>상급종합병원</th>\n",
       "      <th>약국</th>\n",
       "      <th>요양병원</th>\n",
       "      <th>의원</th>\n",
       "      <th>종합병원</th>\n",
       "      <th>...</th>\n",
       "      <th>부동산업_종사자수</th>\n",
       "      <th>전문과학기술서비스업_사업체수</th>\n",
       "      <th>전문과학기술서비스업_종사자수</th>\n",
       "      <th>target</th>\n",
       "      <th>등급</th>\n",
       "      <th>평균연령x인구증가율</th>\n",
       "      <th>의료복지기관</th>\n",
       "      <th>가구수당x합계출산율</th>\n",
       "      <th>가구수당_평균연령</th>\n",
       "      <th>녹지지역비율x토지거래면적</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Unnamed: 0</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>강원도강릉시</th>\n",
       "      <td>강원도</td>\n",
       "      <td>강릉시</td>\n",
       "      <td>42150</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>89</td>\n",
       "      <td>4</td>\n",
       "      <td>113</td>\n",
       "      <td>4</td>\n",
       "      <td>...</td>\n",
       "      <td>1139</td>\n",
       "      <td>286</td>\n",
       "      <td>2040</td>\n",
       "      <td>1.333658</td>\n",
       "      <td>b</td>\n",
       "      <td>-12.818</td>\n",
       "      <td>210</td>\n",
       "      <td>91743.512</td>\n",
       "      <td>0.048371</td>\n",
       "      <td>8270.9707</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>강원도고성군</th>\n",
       "      <td>강원도</td>\n",
       "      <td>고성군</td>\n",
       "      <td>42820</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>13</td>\n",
       "      <td>0</td>\n",
       "      <td>7</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>43</td>\n",
       "      <td>25</td>\n",
       "      <td>131</td>\n",
       "      <td>0.947702</td>\n",
       "      <td>a</td>\n",
       "      <td>-271.504</td>\n",
       "      <td>21</td>\n",
       "      <td>12358.632</td>\n",
       "      <td>0.392189</td>\n",
       "      <td>7214.2704</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>강원도동해시</th>\n",
       "      <td>강원도</td>\n",
       "      <td>동해시</td>\n",
       "      <td>42170</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>38</td>\n",
       "      <td>3</td>\n",
       "      <td>51</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>478</td>\n",
       "      <td>72</td>\n",
       "      <td>378</td>\n",
       "      <td>1.202411</td>\n",
       "      <td>b</td>\n",
       "      <td>-71.928</td>\n",
       "      <td>95</td>\n",
       "      <td>39848.368</td>\n",
       "      <td>0.122119</td>\n",
       "      <td>1434.4052</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>강원도삼척시</th>\n",
       "      <td>강원도</td>\n",
       "      <td>삼척시</td>\n",
       "      <td>42230</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>27</td>\n",
       "      <td>1</td>\n",
       "      <td>26</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>292</td>\n",
       "      <td>73</td>\n",
       "      <td>716</td>\n",
       "      <td>1.210148</td>\n",
       "      <td>b</td>\n",
       "      <td>-14.656</td>\n",
       "      <td>56</td>\n",
       "      <td>35386.992</td>\n",
       "      <td>0.154146</td>\n",
       "      <td>7937.1764</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>강원도속초시</th>\n",
       "      <td>강원도</td>\n",
       "      <td>속초시</td>\n",
       "      <td>42210</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>41</td>\n",
       "      <td>2</td>\n",
       "      <td>64</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>726</td>\n",
       "      <td>82</td>\n",
       "      <td>462</td>\n",
       "      <td>1.304214</td>\n",
       "      <td>b</td>\n",
       "      <td>-27.714</td>\n",
       "      <td>108</td>\n",
       "      <td>35013.791</td>\n",
       "      <td>0.131622</td>\n",
       "      <td>936.6500</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 101 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           sido sigun_nm  sigun_code  병원  보건소  상급종합병원  약국  요양병원   의원  종합병원  \\\n",
       "Unnamed: 0                                                                   \n",
       "강원도강릉시      강원도      강릉시       42150   3    1       0  89     4  113     4   \n",
       "강원도고성군      강원도      고성군       42820   0    1       0  13     0    7     0   \n",
       "강원도동해시      강원도      동해시       42170   2    1       0  38     3   51     1   \n",
       "강원도삼척시      강원도      삼척시       42230   1    1       0  27     1   26     1   \n",
       "강원도속초시      강원도      속초시       42210   0    1       0  41     2   64     2   \n",
       "\n",
       "            ...  부동산업_종사자수  전문과학기술서비스업_사업체수  전문과학기술서비스업_종사자수    target  등급  \\\n",
       "Unnamed: 0  ...                                                              \n",
       "강원도강릉시      ...       1139              286             2040  1.333658   b   \n",
       "강원도고성군      ...         43               25              131  0.947702   a   \n",
       "강원도동해시      ...        478               72              378  1.202411   b   \n",
       "강원도삼척시      ...        292               73              716  1.210148   b   \n",
       "강원도속초시      ...        726               82              462  1.304214   b   \n",
       "\n",
       "            평균연령x인구증가율  의료복지기관  가구수당x합계출산율  가구수당_평균연령  녹지지역비율x토지거래면적  \n",
       "Unnamed: 0                                                            \n",
       "강원도강릉시         -12.818     210   91743.512   0.048371      8270.9707  \n",
       "강원도고성군        -271.504      21   12358.632   0.392189      7214.2704  \n",
       "강원도동해시         -71.928      95   39848.368   0.122119      1434.4052  \n",
       "강원도삼척시         -14.656      56   35386.992   0.154146      7937.1764  \n",
       "강원도속초시         -27.714     108   35013.791   0.131622       936.6500  \n",
       "\n",
       "[5 rows x 101 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_x1.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['a' 'b' 'c' 'd']\n"
     ]
    }
   ],
   "source": [
    "target = train['등급']\n",
    "LB = LabelEncoder() \n",
    "target = LB.fit_transform(target)\n",
    "test_target = LB.fit_transform(test['등급'])\n",
    "print(LB.classes_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "#모델링 하기위해 불필요 변수들 제거\n",
    "dat1.drop(['sido','sigun_nm','sigun_code','target','등급'],axis=1, inplace=True)\n",
    "test_x1.drop(['sido','sigun_nm','sigun_code','target','등급'],axis =1 , inplace = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "96 96\n"
     ]
    }
   ],
   "source": [
    "print(len(dat1.columns),len(test_x1.columns))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "#from sklearn.model_selection import train_test_split\n",
    "#X_tr, X_vld, y_tr, y_vld = train_test_split(dat1, target, test_size = 0.3, random_state = 2019)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> train_test_split은 과적합에 취약한 약점을 가질수 있다. 데이터가 크지 않을때 train vaild를 7:3 으로 나누는것은 알고리즘의 예측 성능을 판단하는데 적절하지 않을수도있음 그래서 교차검증을 실행"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### a.K-fold <a id=\"four-a\"></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> 데이터의 수가 적은 경우에는 이 데이터 중의 일부인 검증 데이터의 수도 적기 때문에 검증 성능의 신뢰도가 떨어진다. 그렇다고 검증 데이터의 수를 증가시키면 학습용 데이터의 수가 적어지므로 정상적인 학습이 되지 않는다. 이러한 딜레마를 해결하기 위한 검증 방법이 K-폴드(K-fold) 교차검증 방법"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> K 겹 교차 검증(Cross validation) 모든 데이터가 최소 한 번은 테스트셋으로 쓰이도록 하는것\n",
    "> 1. Unseen 데이터에 대한 성능을 예측하기 위해\n",
    ">\n",
    "> 2. 더 좋은 모델을 선택하기 위해 (혹은 Hyperparameter Tuning)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "kfold = StratifiedKFold(n_splits=10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> 그중에서 Stratified K 폴드는 불균형한 분포도를 가진 레이블 데이터 집합을 위한 K-폴드 방식\n",
    "일반적으로 분류에서의 교차검증은 K 폴드가 아니라 Stratified K 폴드로 분할해야해"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 각 분류모델들을 clf에 붙여넣음\n",
    "random_state = 1\n",
    "clf = []\n",
    "\n",
    "clf.append(XGBClassifier(random_state = random_state))\n",
    "clf.append(KNeighborsClassifier())\n",
    "clf.append(RandomForestClassifier(random_state=random_state))\n",
    "clf.append(GradientBoostingClassifier(random_state=random_state))\n",
    "clf.append(DecisionTreeClassifier(random_state=random_state))\n",
    "clf.append(LogisticRegression(random_state = random_state))\n",
    "clf.append(SVC(random_state=random_state))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 각 모델의 f1_weighted score의 결과를 clf_result에 저장\n",
    "clf_results = []\n",
    "for classifier in clf :\n",
    "    clf_results.append(cross_val_score(classifier, dat1, y = target, scoring = \"f1_weighted\", cv = kfold, n_jobs=4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 각 모델의 f1_weighted score의 평균과 표준편차를 저장\n",
    "clf_means = []\n",
    "clf_std = []\n",
    "for clf_result in clf_results:\n",
    "    clf_means.append(clf_result.mean())\n",
    "    clf_std.append(clf_result.std())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>CrossValMeans</th>\n",
       "      <th>CrossValerrors</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>XGB</th>\n",
       "      <td>0.882298</td>\n",
       "      <td>0.036967</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>KNeighbors</th>\n",
       "      <td>0.522065</td>\n",
       "      <td>0.078578</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>RF</th>\n",
       "      <td>0.887933</td>\n",
       "      <td>0.029035</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>GBC</th>\n",
       "      <td>0.885359</td>\n",
       "      <td>0.034012</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>DT</th>\n",
       "      <td>0.813907</td>\n",
       "      <td>0.047284</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Logist</th>\n",
       "      <td>0.282414</td>\n",
       "      <td>0.067128</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SVC</th>\n",
       "      <td>0.359241</td>\n",
       "      <td>0.089164</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            CrossValMeans  CrossValerrors\n",
       "XGB              0.882298        0.036967\n",
       "KNeighbors       0.522065        0.078578\n",
       "RF               0.887933        0.029035\n",
       "GBC              0.885359        0.034012\n",
       "DT               0.813907        0.047284\n",
       "Logist           0.282414        0.067128\n",
       "SVC              0.359241        0.089164"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 각 모델의 f1 결과값들을 데이터프레임화\n",
    "clf_re = pd.DataFrame({\"CrossValMeans\":clf_means,\"CrossValerrors\": clf_std},\n",
    "                     index=['XGB','KNeighbors','RF','GBC','DT','Logist','SVC'])\n",
    "clf_re"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> 모든 변수를 포함한 데이터셋을 학습시켰을때의 결과"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<BarContainer object of 7 artists>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#모델의 결과를 그래프화\n",
    "fig, ax1 = plt.subplots(figsize = (10,8))\n",
    "ax1.plot(clf_re['CrossValerrors'], color = 'r')\n",
    "\n",
    "ax2 = ax1.twinx()\n",
    "ax2.bar(clf_re.index, clf_re['CrossValMeans'], alpha = 0.3,color = 'teal')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    ">데이터 세트를 10개의 그룹으로 분할한 다음에 10개의 모델에 대해서 기본 모델을 구축한 뒤 RandomizedSearchCV를 통해서 변수들을 최적화 시켜줍니다. kfold 확인 결과 3개의 모델을 이용해서 앙상블 하려고 합니다"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### b.GradientBoosting <a id=\"four-b\"></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    ">부스팅은 머신러닝 앙상블 기법 중 하나로 sequential한 weak learner들을 여러개 결합하여 예측 혹은 분류 성능을 높이는 알고리즘"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    ">- Sequential한 weak learner들을 residual을 줄이는 방향으로 결합하여 object function과의 loss를 줄여나가는 아이디어. 여기서 정의되는 residual이 negative gradient와 같은 의미를 지니게 되므로 gradient 부스팅이라는 이름이 붙었다"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "#gbc 파라미터 튜닝\n",
    "parametros_gb = {\n",
    "    \"loss\":[\"deviance\"],\n",
    "    \"learning_rate\": [0.01, 0.025, 0.005,0.5, 0.075, 0.1, 0.15, 0.2,0.3,0.8,0.9],\n",
    "    \"min_samples_split\": [0.01, 0.025, 0.005,0.4,0.5, 0.075, 0.1, 0.15, 0.2,0.3,0.8,0.9],\n",
    "    \"min_samples_leaf\": [1,2,3,5,8,10,15,20,40,50,55,60,65,70,80,85,90,100],\n",
    "    \"max_depth\":[3,5,8,10,15,20,25,30,40,50],\n",
    "    \"max_features\":[\"log2\",\"sqrt\"],\n",
    "    \"criterion\": [\"friedman_mse\",  \"mae\"],\n",
    "    \"subsample\":[0.5, 0.618, 0.8, 0.85, 0.9, 0.95, 1.0],\n",
    "    \"n_estimators\":range(1,1000)\n",
    "    }\n",
    "\n",
    "model_gb= GradientBoostingClassifier(random_state=16)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 10 folds for each of 100 candidates, totalling 1000 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done  25 tasks      | elapsed:  1.1min\n",
      "[Parallel(n_jobs=-1)]: Done 146 tasks      | elapsed: 10.3min\n",
      "[Parallel(n_jobs=-1)]: Done 349 tasks      | elapsed: 24.8min\n",
      "[Parallel(n_jobs=-1)]: Done 632 tasks      | elapsed: 43.1min\n",
      "[Parallel(n_jobs=-1)]: Done 1000 out of 1000 | elapsed: 100.5min finished\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'subsample': 1.0,\n",
       " 'n_estimators': 773,\n",
       " 'min_samples_split': 0.1,\n",
       " 'min_samples_leaf': 15,\n",
       " 'max_features': 'log2',\n",
       " 'max_depth': 30,\n",
       " 'loss': 'deviance',\n",
       " 'learning_rate': 0.2,\n",
       " 'criterion': 'friedman_mse'}"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# RandomizedSearchCV를 통해서 파라미터 최적화 여기서 stratifed kfold 사용\n",
    "# 파라미터 튜닝 후 학습 결과 105분이 소요됨\n",
    "# 최적화된 파라미터들 \n",
    "gb_random = RandomizedSearchCV(estimator = model_gb, param_distributions = parametros_gb, n_iter = 100, cv = kfold, \n",
    "                               verbose=2, random_state=16,n_jobs = -1, scoring='f1_weighted')\n",
    "gb_random.fit(dat1,target)\n",
    "\n",
    "gb_random.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GradientBoostingClassifier(learning_rate=0.2, max_depth=30, max_features='log2',\n",
       "                           min_samples_leaf=15, min_samples_split=0.1,\n",
       "                           n_estimators=773)"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 최적화된 파라미터를 가지고 GBC모델을 만듬\n",
    "model_gbc = GradientBoostingClassifier(**gb_random.best_params_)\n",
    "model_gbc.fit(dat1, target)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 모델에서 test를 예측하고 f1_score점수화\n",
    "Predicted_gbc = model_gbc.predict(test_x1)\n",
    "gbc_f1 = cross_val_score(model_gbc, dat1, target, cv=kfold, scoring='f1_weighted').mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "#4가지 f1_score 비교 함수\n",
    "def four_f1(model_f1,model):\n",
    "    print(\"F1 Cross_validate\",model_f1)\n",
    "    print(\"F1 Macro:\",f1_score(test_target, model, average='macro'))\n",
    "    print(\"F1 Micro:\",f1_score(test_target, model, average='micro'))  \n",
    "    print(\"F1 Weighted:\",f1_score(test_target, model, average='weighted'))\n",
    "    print(\"\\nMatrix of confusion\")\n",
    "    return confusion_matrix(test_target, model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "F1 Cross_validate 0.8977127168887284\n",
      "F1 Macro: 0.8927282395024331\n",
      "F1 Micro: 0.8947368421052632\n",
      "F1 Weighted: 0.8957137938462217\n",
      "\n",
      "Matrix of confusion\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([[65,  9,  0,  0],\n",
       "       [ 1, 44,  4,  0],\n",
       "       [ 0,  2, 55,  7],\n",
       "       [ 0,  0,  1, 40]], dtype=int64)"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#각각의 f1_score점수 \n",
    "four_f1(gbc_f1,Predicted_gbc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# confusion_matrix 시각화 함수\n",
    "def cnf_matrix_model(model):\n",
    "    cnf_matrix_gbc = confusion_matrix(test_target, model)\n",
    "    g = sns.heatmap(pd.DataFrame(cnf_matrix_gbc), annot=True, cmap=\"BuGn\", fmt='g')\n",
    "    buttom , top = g.get_ylim()\n",
    "    g.set_ylim(buttom+0.5, top-0.5)\n",
    "    plt.ylabel('Actual Label')\n",
    "    plt.xlabel('Predicted Label')\n",
    "    return g"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='Predicted Label', ylabel='Actual Label'>"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cnf_matrix_model(Predicted_gbc)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### gbc voting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "#다중 분류기들을 Voting하는 방법을 사용하기 위해 GBC 모델의 random_state(난수값)을 변경해가면서 10개의 GBC 분류기를 만들어냄\n",
    "#파라미터는 위에서 정의해준 최적 파라미터 사용 \n",
    "ets_gbc = []    \n",
    "for i in range(10):\n",
    "    gbc = GradientBoostingClassifier(random_state=217+i,**gb_random.best_params_)\n",
    "    ets_gbc.append(('gbc{}'.format(i), gbc))   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# target 불균형에 대해 class_weigt 라이브러리 사용\n",
    "# 클래스 불균형 문제를 해소하기 위해 compute_sample_weight 사용\n",
    "\n",
    "y_train_weights = class_weight.compute_sample_weight('balanced', target, indices=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# VotingClassifier 정의 \n",
    "# 위에서 정의해준 10개의 GBC 분류기를 사용\n",
    "# 앙상블의 소프트 보팅\n",
    "vc2 = VotingClassifier(ets_gbc, voting='soft')    \n",
    "_ = vc2.fit(dat1, target,sample_weight=y_train_weights)    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "gbc_clf_final = vc2.estimators_[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 단일 모델, soft voting, hard voting 3가지 방법에 대한 점수\n",
    "gbc_global_score = f1_score(test_target, gbc_clf_final.predict(test_x1), average='weighted')\n",
    "vc2.voting = 'hard'\n",
    "gbc_global_score_hard = f1_score(test_target, vc2.predict(test_x1), average='weighted')\n",
    "vc2.voting = 'soft'\n",
    "gbc_global_score_soft = f1_score(test_target, vc2.predict(test_x1),average='weighted')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Validation score of a single XGB Classifier: 0.9042\n",
      "Validation score of a VotingClassifier on 3 XGB with hard voting strategy: 0.9042\n",
      "Validation score of a VotingClassifier on 3 XGB with soft voting strategy: 0.9042\n"
     ]
    }
   ],
   "source": [
    "# Soft Voting 방법이 가장 높은 점수를 보임 \n",
    "print('Validation score of a single XGB Classifier: {:.4f}'.format(gbc_global_score))\n",
    "print('Validation score of a VotingClassifier on 3 XGB with hard voting strategy: {:.4f}'.format(gbc_global_score_hard))\n",
    "print('Validation score of a VotingClassifier on 3 XGB with soft voting strategy: {:.4f}'.format(gbc_global_score_soft))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> soft 보팅의 경우 f1_score 0.9042"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "f1  0.9041742397858935\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.99      0.89      0.94        74\n",
      "           1       0.81      0.90      0.85        49\n",
      "           2       0.93      0.86      0.89        64\n",
      "           3       0.85      1.00      0.92        41\n",
      "\n",
      "    accuracy                           0.90       228\n",
      "   macro avg       0.90      0.91      0.90       228\n",
      "weighted avg       0.91      0.90      0.90       228\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cnf_matrix_vgbc = confusion_matrix(test_target, vc2.predict(test_x1))\n",
    "g = sns.heatmap(pd.DataFrame(cnf_matrix_vgbc), annot=True, cmap=\"BuGn\", fmt='g')\n",
    "buttom , top = g.get_ylim()\n",
    "g.set_ylim(buttom+0.5, top-0.5)\n",
    "plt.ylabel('Actual Label')\n",
    "plt.xlabel('Predicted Label')\n",
    "\n",
    "print('f1 ',f1_score(y_true=test_target, y_pred=vc2.predict(test_x1), average='weighted'))\n",
    "print(classification_report(test_target, vc2.predict(test_x1)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### c.RandomForest <a id=\"four-c\"></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "# RandomForest 파라미터 튜닝\n",
    "parametros_rf = {\"max_depth\": [3, 5,10,20,30,40,None],\n",
    "            'max_features': ['auto', 'sqrt'],\n",
    "              \"max_features\": sp_randint(1, 11),\n",
    "              \"min_samples_split\": sp_randint(2, 11),\n",
    "              \"min_samples_leaf\": sp_randint(1, 11),\n",
    "             'n_estimators': [10,100,200,500,800,1000],\n",
    "              \"bootstrap\": [True, False]\n",
    "             }\n",
    "model_rf= RandomForestClassifier(random_state=16)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 10 folds for each of 100 candidates, totalling 1000 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done  25 tasks      | elapsed:   11.3s\n",
      "[Parallel(n_jobs=-1)]: Done 146 tasks      | elapsed:   55.9s\n",
      "[Parallel(n_jobs=-1)]: Done 349 tasks      | elapsed:  1.9min\n",
      "[Parallel(n_jobs=-1)]: Done 632 tasks      | elapsed:  4.0min\n",
      "[Parallel(n_jobs=-1)]: Done 1000 out of 1000 | elapsed:  6.6min finished\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'bootstrap': False,\n",
       " 'max_depth': 10,\n",
       " 'max_features': 7,\n",
       " 'min_samples_leaf': 1,\n",
       " 'min_samples_split': 3,\n",
       " 'n_estimators': 100}"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 파라미터 최적화 cv = kfold(stratifed kfold 이용)\n",
    "# 파라미터 최적화 시간 7분 \n",
    "# 최적의 파라미터들\n",
    "rf_random = RandomizedSearchCV(estimator = model_rf, param_distributions = parametros_rf, n_iter = 100, cv = kfold, \n",
    "                               verbose=2, random_state=16,n_jobs = -1, scoring='f1_weighted')\n",
    "rf_random.fit(dat1,target)\n",
    "rf_random.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap=False, max_depth=10, max_features=7,\n",
       "                       min_samples_split=3)"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 최적화된 파라미터를 가지고 RF모델 만듬\n",
    "model_rf = RandomForestClassifier(**rf_random.best_params_)\n",
    "model_rf.fit(dat1,target)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "# RF모델로 test 데이터 예측과 점수\n",
    "Predicted_rf = model_rf.predict(test_x1)\n",
    "rf_f1 = cross_val_score(model_rf, dat1,target, cv=kfold, scoring='f1_weighted').mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "F1 Cross_validate 0.8880985857693998\n",
      "F1 Macro: 0.884410087053297\n",
      "F1 Micro: 0.8859649122807017\n",
      "F1 Weighted: 0.8874800550123053\n",
      "\n",
      "Matrix of confusion\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([[64, 10,  0,  0],\n",
       "       [ 0, 42,  7,  0],\n",
       "       [ 0,  2, 57,  5],\n",
       "       [ 0,  0,  2, 39]], dtype=int64)"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# RF 모델의 f1 score\n",
    "four_f1(rf_f1,Predicted_rf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='Predicted Label', ylabel='Actual Label'>"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# confusion_matrix 시각화\n",
    "cnf_matrix_model(Predicted_rf)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### rf voting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 다중 분류기들을 Voting하는 방법을 사용하기 위해 RF 모델의 random_state(난수값)을 변경해가면서 10개의 RF 분류기를 만들어냄\n",
    "# 파라미터는 위에서 정의해준 최적 파라미터 사용 \n",
    "\n",
    "rf_ets = []    \n",
    "for i in range(10):\n",
    "    rf = RandomForestClassifier(random_state=217+i,**rf_random.best_params_)\n",
    "    rf_ets.append(('rf{}'.format(i), rf))   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 위에서 정의해준 10개의 rf분류기를 사용\n",
    "# 앙상블의 소프트 보팅\n",
    "vc3 = VotingClassifier(rf_ets, voting='soft')    \n",
    "_ = vc3.fit(dat1,target,sample_weight=y_train_weights)    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "rf_clf_final = vc3.estimators_[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 단일 모델, soft voting, hard voting 3가지 방법에 대한 점수\n",
    "rf_global_score = f1_score(test_target, rf_clf_final.predict(test_x1), average='weighted')\n",
    "vc3.voting = 'hard'\n",
    "rf_global_score_hard = f1_score(test_target, vc3.predict(test_x1),average='weighted')\n",
    "vc3.voting = 'soft'\n",
    "rf_global_score_soft = f1_score(test_target, vc3.predict(test_x1), average='weighted')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Validation score of a singl rf Classifier: 0.8829\n",
      "Validation score of a VotingClassifier on 3 rf with soft voting strategy: 0.8828\n",
      "Validation score of a VotingClassifier on 3 rf with hard voting strategy: 0.8828\n"
     ]
    }
   ],
   "source": [
    "# soft Voting 방법이 가장 높은 점수를 보임 \n",
    "print('Validation score of a singl rf Classifier: {:.4f}'.format(rf_global_score))\n",
    "print('Validation score of a VotingClassifier on 3 rf with soft voting strategy: {:.4f}'.format(rf_global_score_soft))\n",
    "print('Validation score of a VotingClassifier on 3 rf with hard voting strategy: {:.4f}'.format(rf_global_score_hard))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "f1  0.8828186319722502\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.98      0.86      0.92        74\n",
      "           1       0.77      0.84      0.80        49\n",
      "           2       0.86      0.89      0.88        64\n",
      "           3       0.89      0.95      0.92        41\n",
      "\n",
      "    accuracy                           0.88       228\n",
      "   macro avg       0.88      0.89      0.88       228\n",
      "weighted avg       0.89      0.88      0.88       228\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cnf_matrix_vrf = confusion_matrix(test_target, vc3.predict(test_x1))\n",
    "g = sns.heatmap(pd.DataFrame(cnf_matrix_vrf), annot=True, cmap=\"BuGn\", fmt='g')\n",
    "buttom , top = g.get_ylim()\n",
    "g.set_ylim(buttom+0.5, top-0.5)\n",
    "plt.ylabel('Actual Label')\n",
    "plt.xlabel('Predicted Label')\n",
    "\n",
    "print('f1 ',f1_score(y_true=test_target, y_pred=vc3.predict(test_x1), average='weighted'))\n",
    "print(classification_report(test_target, vc3.predict(test_x1)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "soft voting의 경우 0.8828186319722502"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### d.xgboost <a id=\"four-d\"></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> - GBM은 residaul을 줄이는 방향으로 weak learner를 결합해 강력한 성능을 자랑하지만, 해당 train data에 residual을 계속 줄이니까 overfitting 되기 쉽다는 문제점이 있다. 이를 해결하기 위해 XGBoost는 GBM에 regularization term을 추가한 알고리즘이다. 또한 다양한 loss function을 지원해 task에 따른 유연한 튜닝이 가능하다는 장점이 있다"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> XGBoost = GBM + Regularization+ 다양한 Loss function 지원"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[02:47:22] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { cv } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "f1  0.9003906654932211\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       1.00      0.88      0.94        74\n",
      "           1       0.81      0.94      0.87        49\n",
      "           2       0.88      0.91      0.89        64\n",
      "           3       0.90      0.88      0.89        41\n",
      "\n",
      "    accuracy                           0.90       228\n",
      "   macro avg       0.90      0.90      0.90       228\n",
      "weighted avg       0.91      0.90      0.90       228\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#xgboost 모델 생성 후 학습\n",
    "#모델 생성\n",
    "model = xgb.XGBClassifier(n_estimators = 5000, max_depth = 30, cv= kfold,learning_rate = 0.01,random_state = 16)\n",
    "#모델 학습\n",
    "model.fit(dat1, target) \n",
    "#예측 분류\n",
    "y_pred = model.predict(test_x1)\n",
    "\n",
    "#오분류표 코드 \n",
    "cnf_matrix = confusion_matrix(test_target, y_pred)\n",
    "g = sns.heatmap(pd.DataFrame(cnf_matrix), annot=True, cmap=\"BuGn\", fmt='g')\n",
    "buttom , top = g.get_ylim()\n",
    "g.set_ylim(buttom+0.5, top-0.5)\n",
    "plt.ylabel('Actual Label')\n",
    "plt.xlabel('Predicted Label')\n",
    "\n",
    "#f1 score와 오분류표 보여줌\n",
    "print('f1 ',f1_score(y_true=test_target, y_pred=y_pred, average='weighted')) #불균현 문제를 위하 가중치 weighted \n",
    "print(classification_report(test_target, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> accuracy의 경우 데이터가 균형있을때 효과적이고 f1_score의 경우 데이터가 불균형일때 효과적\n",
    "지방 소멸위험의 경우 소멸위험이적은 D등급이 36개로 불균형으로 f1_score를 채택한다"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> 실제 a등급(0)을 a등급으로 제대로 예측한 데이터 65개 \n",
    ">\n",
    "> f1_score 90%이며  confusion_matrix확인결과 데이터가 잘 분류된것을 확인할수있음"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "#변수 중요도 코드\n",
    "def plot_feature_importance(model, X_train, figsize=(12, 6)):\n",
    "    sns.set_style('darkgrid')\n",
    "    \n",
    "    # Plot feature importance\n",
    "    feature_importance = model.feature_importances_\n",
    "    feature_importance = 100.0 * (feature_importance / feature_importance.max())\n",
    "    sorted_idx = np.argsort(feature_importance)\n",
    "    pos = np.arange(sorted_idx.shape[0]) + .5\n",
    "\n",
    "    plt.figure(figsize=figsize)\n",
    "    plt.rc(\"font\",family=\"Malgun Gothic\")\n",
    "    plt.rc(\"axes\",unicode_minus=False)\n",
    "    \n",
    "    plt.barh(pos, feature_importance[sorted_idx], align='center',color = 'teal' )\n",
    "    plt.yticks(pos, X_train.columns[sorted_idx])\n",
    "    plt.xlabel('Relative Importance')\n",
    "    plt.title('Variable Importance')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x2520 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 변수 중요도 확인 \n",
    "plot_feature_importance(model, dat1, (10, 35))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "    <style>\n",
       "    table.eli5-weights tr:hover {\n",
       "        filter: brightness(85%);\n",
       "    }\n",
       "</style>\n",
       "\n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "        <table class=\"eli5-weights eli5-feature-importances\" style=\"border-collapse: collapse; border: none; margin-top: 0em; table-layout: auto;\">\n",
       "    <thead>\n",
       "    <tr style=\"border: none;\">\n",
       "        <th style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">Weight</th>\n",
       "        <th style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">Feature</th>\n",
       "    </tr>\n",
       "    </thead>\n",
       "    <tbody>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 80.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.3049\n",
       "                \n",
       "                    &plusmn; 0.0232\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                평균연령\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 94.69%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0459\n",
       "                \n",
       "                    &plusmn; 0.0078\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                십만명당_문화기반시설수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 95.29%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0386\n",
       "                \n",
       "                    &plusmn; 0.0092\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                인구증가율\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 96.98%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0205\n",
       "                \n",
       "                    &plusmn; 0.0018\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                노인천명당_노인여가복지시설수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 99.34%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0023\n",
       "                \n",
       "                    &plusmn; 0.0023\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                혼인건수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 99.52%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0015\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                광업_사업체수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 99.85%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0003\n",
       "                \n",
       "                    &plusmn; 0.0012\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                조혼인율\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                교통안전영역\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                주민등록인구60세이상\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                교원1인당_학생수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                재적학생수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                교원수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                교통문화지수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                운전행태영역\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                남녀성비\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                보행행태영역\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                녹지지역면적\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                녹지지역면적비율\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                문화기반시설수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                유치원교원수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                유치원수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                유치원아수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                천명당_사설학원수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                사설학원수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                노인여가복지시설수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                녹지지역비율x토지거래면적\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                주관적_건강수준인지율\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                당해년총인구\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                보건소\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                상급종합병원\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                약국\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                요양병원\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                의원\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                종합병원\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                치과병원\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                한방병원\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                인구십만명당_자살률_\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                합계출산율\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                출생아수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                도시지역면적\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                주택수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                가구수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                하수도보급률\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                건강보험_적용인구_현황\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                상수도보급률\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                고위험음주율\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                비만율\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                EQ_5D\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                전년총인구\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                가구수당_평균연령\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                정보통신업_사업체수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                수도하수폐기물원료재생업_종사자수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                건설업_사업체수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                건설업_종사자수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                도매소매업_사업체수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                도매소매업_종사자수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                운수창고업_사업체수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                운수창고업_종사자수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                숙박음식점업_사업체수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                숙박음식점업_종사자수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                정보통신업_종사자수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                전기가스증기및공기조절공급업_종사자수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                금융보험업_사업체수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                금융보험업_종사자수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                부동산업_사업체수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                부동산업_종사자수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                전문과학기술서비스업_사업체수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                전문과학기술서비스업_종사자수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                평균연령x인구증가율\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                의료복지기관\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                가구수당x합계출산율\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                수도하수폐기물원료재생업_사업체수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                전기가스증기및공기조절공급업_사업체수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                인당_자동차등록대수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                초등교원수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                사회복지예산비중\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                사회복지분야예산액\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                보건분야예산액\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                일반공공행정예산비중\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                일반공공행정분야예산액\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                천명당_외국인수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                주민등록인구\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                천명당_의료기관병상수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                총병상수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                초등학생수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                제조업_종사자수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                토지거래면적\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                유치원학급당_학생수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                초등학교학급당_학생수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                중학교학급당_학생수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                고등학교학급당_학생수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                농업임업및어업_사업체수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                농업임업및어업_종사자수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                광업_종사자수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                제조업_사업체수\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "                    &plusmn; 0.0000\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                병원\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "    \n",
       "    </tbody>\n",
       "</table>\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Permutation Importance 확인 \n",
    "perm = PermutationImportance(model, scoring = \"f1_weighted\",random_state = 42).fit(dat1, target)\n",
    "eli5.show_weights(perm, top = 100, feature_names = dat1.columns.tolist())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "#트리모델 시각화 색하고 모형결정 \n",
    "node_params = {'shape': 'box',\n",
    "              'style':'filled, rounded',\n",
    "              'fillcolor': 'LightBlue'}\n",
    "leaf_params = {'shape':'box',\n",
    "              'style':'filled',\n",
    "              'fillcolor':'GreenYellow'}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/svg+xml": [
       "<?xml version=\"1.0\" encoding=\"UTF-8\" standalone=\"no\"?>\r\n",
       "<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\r\n",
       " \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\r\n",
       "<!-- Generated by graphviz version 2.38.0 (20140413.2041)\r\n",
       " -->\r\n",
       "<!-- Title: %3 Pages: 1 -->\r\n",
       "<svg width=\"1231pt\" height=\"653pt\"\r\n",
       " viewBox=\"0.00 0.00 1231.00 653.00\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\r\n",
       "<g id=\"graph0\" class=\"graph\" transform=\"scale(1 1) rotate(0) translate(4 649)\">\r\n",
       "<title>%3</title>\r\n",
       "<polygon fill=\"white\" stroke=\"none\" points=\"-4,4 -4,-649 1227,-649 1227,4 -4,4\"/>\r\n",
       "<!-- 0 -->\r\n",
       "<g id=\"node1\" class=\"node\"><title>0</title>\r\n",
       "<path fill=\"LightBlue\" stroke=\"black\" d=\"M668.5,-645C668.5,-645 552.5,-645 552.5,-645 546.5,-645 540.5,-639 540.5,-633 540.5,-633 540.5,-621 540.5,-621 540.5,-615 546.5,-609 552.5,-609 552.5,-609 668.5,-609 668.5,-609 674.5,-609 680.5,-615 680.5,-621 680.5,-621 680.5,-633 680.5,-633 680.5,-639 674.5,-645 668.5,-645\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"610.5\" y=\"-623.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">평균연령&lt;47.5999985</text>\r\n",
       "</g>\r\n",
       "<!-- 1 -->\r\n",
       "<g id=\"node2\" class=\"node\"><title>1</title>\r\n",
       "<path fill=\"LightBlue\" stroke=\"black\" d=\"M575,-558C575,-558 324,-558 324,-558 318,-558 312,-552 312,-546 312,-546 312,-534 312,-534 312,-528 318,-522 324,-522 324,-522 575,-522 575,-522 581,-522 587,-528 587,-534 587,-534 587,-546 587,-546 587,-552 581,-558 575,-558\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"449.5\" y=\"-536.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">노인천명당_노인여가복지시설수&lt;18.1500015</text>\r\n",
       "</g>\r\n",
       "<!-- 0&#45;&gt;1 -->\r\n",
       "<g id=\"edge1\" class=\"edge\"><title>0&#45;&gt;1</title>\r\n",
       "<path fill=\"none\" stroke=\"#0000ff\" d=\"M577.919,-608.799C552.741,-595.506 517.729,-577.021 490.621,-562.71\"/>\r\n",
       "<polygon fill=\"#0000ff\" stroke=\"#0000ff\" points=\"492.234,-559.604 481.757,-558.03 488.966,-565.794 492.234,-559.604\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"575\" y=\"-579.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">yes, missing</text>\r\n",
       "</g>\r\n",
       "<!-- 2 -->\r\n",
       "<g id=\"node3\" class=\"node\"><title>2</title>\r\n",
       "<path fill=\"LightBlue\" stroke=\"black\" d=\"M793.5,-558C793.5,-558 657.5,-558 657.5,-558 651.5,-558 645.5,-552 645.5,-546 645.5,-546 645.5,-534 645.5,-534 645.5,-528 651.5,-522 657.5,-522 657.5,-522 793.5,-522 793.5,-522 799.5,-522 805.5,-528 805.5,-534 805.5,-534 805.5,-546 805.5,-546 805.5,-552 799.5,-558 793.5,-558\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"725.5\" y=\"-536.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">인구증가율&lt;0.805000007</text>\r\n",
       "</g>\r\n",
       "<!-- 0&#45;&gt;2 -->\r\n",
       "<g id=\"edge2\" class=\"edge\"><title>0&#45;&gt;2</title>\r\n",
       "<path fill=\"none\" stroke=\"#ff0000\" d=\"M633.772,-608.799C651.152,-595.953 675.092,-578.258 694.156,-564.167\"/>\r\n",
       "<polygon fill=\"#ff0000\" stroke=\"#ff0000\" points=\"696.301,-566.934 702.263,-558.175 692.14,-561.305 696.301,-566.934\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"683\" y=\"-579.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">no</text>\r\n",
       "</g>\r\n",
       "<!-- 3 -->\r\n",
       "<g id=\"node4\" class=\"node\"><title>3</title>\r\n",
       "<path fill=\"LightBlue\" stroke=\"black\" d=\"M285.5,-471C285.5,-471 175.5,-471 175.5,-471 169.5,-471 163.5,-465 163.5,-459 163.5,-459 163.5,-447 163.5,-447 163.5,-441 169.5,-435 175.5,-435 175.5,-435 285.5,-435 285.5,-435 291.5,-435 297.5,-441 297.5,-447 297.5,-447 297.5,-459 297.5,-459 297.5,-465 291.5,-471 285.5,-471\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"230.5\" y=\"-449.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">조혼인율&lt;2.9000001</text>\r\n",
       "</g>\r\n",
       "<!-- 1&#45;&gt;3 -->\r\n",
       "<g id=\"edge3\" class=\"edge\"><title>1&#45;&gt;3</title>\r\n",
       "<path fill=\"none\" stroke=\"#0000ff\" d=\"M405.446,-521.901C370.474,-508.328 321.436,-489.295 284.173,-474.832\"/>\r\n",
       "<polygon fill=\"#0000ff\" stroke=\"#0000ff\" points=\"285.073,-471.427 274.484,-471.071 282.54,-477.953 285.073,-471.427\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"388\" y=\"-492.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">yes, missing</text>\r\n",
       "</g>\r\n",
       "<!-- 4 -->\r\n",
       "<g id=\"node5\" class=\"node\"><title>4</title>\r\n",
       "<path fill=\"LightBlue\" stroke=\"black\" d=\"M519.5,-471C519.5,-471 379.5,-471 379.5,-471 373.5,-471 367.5,-465 367.5,-459 367.5,-459 367.5,-447 367.5,-447 367.5,-441 373.5,-435 379.5,-435 379.5,-435 519.5,-435 519.5,-435 525.5,-435 531.5,-441 531.5,-447 531.5,-447 531.5,-459 531.5,-459 531.5,-465 525.5,-471 519.5,-471\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"449.5\" y=\"-449.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">인구증가율&lt;&#45;0.344999999</text>\r\n",
       "</g>\r\n",
       "<!-- 1&#45;&gt;4 -->\r\n",
       "<g id=\"edge4\" class=\"edge\"><title>1&#45;&gt;4</title>\r\n",
       "<path fill=\"none\" stroke=\"#ff0000\" d=\"M449.5,-521.799C449.5,-510.163 449.5,-494.548 449.5,-481.237\"/>\r\n",
       "<polygon fill=\"#ff0000\" stroke=\"#ff0000\" points=\"453,-481.175 449.5,-471.175 446,-481.175 453,-481.175\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"457\" y=\"-492.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">no</text>\r\n",
       "</g>\r\n",
       "<!-- 5 -->\r\n",
       "<g id=\"node18\" class=\"node\"><title>5</title>\r\n",
       "<path fill=\"LightBlue\" stroke=\"black\" d=\"M851,-471C851,-471 600,-471 600,-471 594,-471 588,-465 588,-459 588,-459 588,-447 588,-447 588,-441 594,-435 600,-435 600,-435 851,-435 851,-435 857,-435 863,-441 863,-447 863,-447 863,-459 863,-459 863,-465 857,-471 851,-471\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"725.5\" y=\"-449.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">노인천명당_노인여가복지시설수&lt;12.2999992</text>\r\n",
       "</g>\r\n",
       "<!-- 2&#45;&gt;5 -->\r\n",
       "<g id=\"edge17\" class=\"edge\"><title>2&#45;&gt;5</title>\r\n",
       "<path fill=\"none\" stroke=\"#0000ff\" d=\"M725.5,-521.799C725.5,-510.163 725.5,-494.548 725.5,-481.237\"/>\r\n",
       "<polygon fill=\"#0000ff\" stroke=\"#0000ff\" points=\"729,-481.175 725.5,-471.175 722,-481.175 729,-481.175\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"760\" y=\"-492.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">yes, missing</text>\r\n",
       "</g>\r\n",
       "<!-- 6 -->\r\n",
       "<g id=\"node19\" class=\"node\"><title>6</title>\r\n",
       "<path fill=\"LightBlue\" stroke=\"black\" d=\"M1079,-471C1079,-471 906,-471 906,-471 900,-471 894,-465 894,-459 894,-459 894,-447 894,-447 894,-441 900,-435 906,-435 906,-435 1079,-435 1079,-435 1085,-435 1091,-441 1091,-447 1091,-447 1091,-459 1091,-459 1091,-465 1085,-471 1079,-471\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"992.5\" y=\"-449.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">농업임업및어업_사업체수&lt;22.5</text>\r\n",
       "</g>\r\n",
       "<!-- 2&#45;&gt;6 -->\r\n",
       "<g id=\"edge18\" class=\"edge\"><title>2&#45;&gt;6</title>\r\n",
       "<path fill=\"none\" stroke=\"#ff0000\" d=\"M779.21,-521.901C822.481,-508.126 883.417,-488.727 929.083,-474.189\"/>\r\n",
       "<polygon fill=\"#ff0000\" stroke=\"#ff0000\" points=\"930.409,-477.44 938.876,-471.071 928.285,-470.77 930.409,-477.44\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"883\" y=\"-492.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">no</text>\r\n",
       "</g>\r\n",
       "<!-- 7 -->\r\n",
       "<g id=\"node6\" class=\"node\"><title>7</title>\r\n",
       "<polygon fill=\"GreenYellow\" stroke=\"black\" points=\"129,-384 0,-384 0,-348 129,-348 129,-384\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"64.5\" y=\"-362.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">leaf=0.00397102255</text>\r\n",
       "</g>\r\n",
       "<!-- 3&#45;&gt;7 -->\r\n",
       "<g id=\"edge5\" class=\"edge\"><title>3&#45;&gt;7</title>\r\n",
       "<path fill=\"none\" stroke=\"#0000ff\" d=\"M193.949,-434.882C182.249,-429.294 169.278,-422.986 157.5,-417 139.85,-408.03 120.541,-397.73 104.113,-388.818\"/>\r\n",
       "<polygon fill=\"#0000ff\" stroke=\"#0000ff\" points=\"105.768,-385.734 95.3116,-384.027 102.421,-391.882 105.768,-385.734\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"192\" y=\"-405.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">yes, missing</text>\r\n",
       "</g>\r\n",
       "<!-- 8 -->\r\n",
       "<g id=\"node7\" class=\"node\"><title>8</title>\r\n",
       "<path fill=\"LightBlue\" stroke=\"black\" d=\"M301.5,-384C301.5,-384 159.5,-384 159.5,-384 153.5,-384 147.5,-378 147.5,-372 147.5,-372 147.5,-360 147.5,-360 147.5,-354 153.5,-348 159.5,-348 159.5,-348 301.5,-348 301.5,-348 307.5,-348 313.5,-354 313.5,-360 313.5,-360 313.5,-372 313.5,-372 313.5,-378 307.5,-384 301.5,-384\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"230.5\" y=\"-362.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">운전행태영역&lt;26.6599998</text>\r\n",
       "</g>\r\n",
       "<!-- 3&#45;&gt;8 -->\r\n",
       "<g id=\"edge6\" class=\"edge\"><title>3&#45;&gt;8</title>\r\n",
       "<path fill=\"none\" stroke=\"#ff0000\" d=\"M230.5,-434.799C230.5,-423.163 230.5,-407.548 230.5,-394.237\"/>\r\n",
       "<polygon fill=\"#ff0000\" stroke=\"#ff0000\" points=\"234,-394.175 230.5,-384.175 227,-394.175 234,-394.175\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"238\" y=\"-405.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">no</text>\r\n",
       "</g>\r\n",
       "<!-- 9 -->\r\n",
       "<g id=\"node16\" class=\"node\"><title>9</title>\r\n",
       "<polygon fill=\"GreenYellow\" stroke=\"black\" points=\"453.5,-384 331.5,-384 331.5,-348 453.5,-348 453.5,-384\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"392.5\" y=\"-362.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">leaf=0.0161221139</text>\r\n",
       "</g>\r\n",
       "<!-- 4&#45;&gt;9 -->\r\n",
       "<g id=\"edge15\" class=\"edge\"><title>4&#45;&gt;9</title>\r\n",
       "<path fill=\"none\" stroke=\"#0000ff\" d=\"M437.965,-434.799C429.853,-422.702 418.857,-406.305 409.712,-392.667\"/>\r\n",
       "<polygon fill=\"#0000ff\" stroke=\"#0000ff\" points=\"412.494,-390.532 404.018,-384.175 406.68,-394.43 412.494,-390.532\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"459\" y=\"-405.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">yes, missing</text>\r\n",
       "</g>\r\n",
       "<!-- 10 -->\r\n",
       "<g id=\"node17\" class=\"node\"><title>10</title>\r\n",
       "<polygon fill=\"GreenYellow\" stroke=\"black\" points=\"605.5,-384 471.5,-384 471.5,-348 605.5,-348 605.5,-384\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"538.5\" y=\"-362.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">leaf=&#45;0.00525286002</text>\r\n",
       "</g>\r\n",
       "<!-- 4&#45;&gt;10 -->\r\n",
       "<g id=\"edge16\" class=\"edge\"><title>4&#45;&gt;10</title>\r\n",
       "<path fill=\"none\" stroke=\"#ff0000\" d=\"M475.363,-434.977C482.817,-429.593 490.753,-423.374 497.5,-417 505.456,-409.483 513.299,-400.44 519.961,-392.122\"/>\r\n",
       "<polygon fill=\"#ff0000\" stroke=\"#ff0000\" points=\"522.803,-394.17 526.19,-384.131 517.282,-389.866 522.803,-394.17\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"519\" y=\"-405.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">no</text>\r\n",
       "</g>\r\n",
       "<!-- 15 -->\r\n",
       "<g id=\"node8\" class=\"node\"><title>15</title>\r\n",
       "<polygon fill=\"GreenYellow\" stroke=\"black\" points=\"220,-297 91,-297 91,-261 220,-261 220,-297\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"155.5\" y=\"-275.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">leaf=0.00119367288</text>\r\n",
       "</g>\r\n",
       "<!-- 8&#45;&gt;15 -->\r\n",
       "<g id=\"edge7\" class=\"edge\"><title>8&#45;&gt;15</title>\r\n",
       "<path fill=\"none\" stroke=\"#0000ff\" d=\"M211.092,-347.765C205.241,-342.276 198.934,-336.054 193.5,-330 186.581,-322.292 179.564,-313.419 173.475,-305.308\"/>\r\n",
       "<polygon fill=\"#0000ff\" stroke=\"#0000ff\" points=\"176.258,-303.185 167.508,-297.211 170.623,-307.338 176.258,-303.185\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"228\" y=\"-318.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">yes, missing</text>\r\n",
       "</g>\r\n",
       "<!-- 16 -->\r\n",
       "<g id=\"node9\" class=\"node\"><title>16</title>\r\n",
       "<path fill=\"LightBlue\" stroke=\"black\" d=\"M392.5,-297C392.5,-297 250.5,-297 250.5,-297 244.5,-297 238.5,-291 238.5,-285 238.5,-285 238.5,-273 238.5,-273 238.5,-267 244.5,-261 250.5,-261 250.5,-261 392.5,-261 392.5,-261 398.5,-261 404.5,-267 404.5,-273 404.5,-273 404.5,-285 404.5,-285 404.5,-291 398.5,-297 392.5,-297\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"321.5\" y=\"-275.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">보행행태영역&lt;14.9499998</text>\r\n",
       "</g>\r\n",
       "<!-- 8&#45;&gt;16 -->\r\n",
       "<g id=\"edge8\" class=\"edge\"><title>8&#45;&gt;16</title>\r\n",
       "<path fill=\"none\" stroke=\"#ff0000\" d=\"M248.915,-347.799C262.36,-335.241 280.765,-318.049 295.677,-304.12\"/>\r\n",
       "<polygon fill=\"#ff0000\" stroke=\"#ff0000\" points=\"298.193,-306.559 303.112,-297.175 293.415,-301.444 298.193,-306.559\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"289\" y=\"-318.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">no</text>\r\n",
       "</g>\r\n",
       "<!-- 21 -->\r\n",
       "<g id=\"node10\" class=\"node\"><title>21</title>\r\n",
       "<path fill=\"LightBlue\" stroke=\"black\" d=\"M299,-210C299,-210 194,-210 194,-210 188,-210 182,-204 182,-198 182,-198 182,-186 182,-186 182,-180 188,-174 194,-174 194,-174 299,-174 299,-174 305,-174 311,-180 311,-186 311,-186 311,-198 311,-198 311,-204 305,-210 299,-210\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"246.5\" y=\"-188.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">광업_종사자수&lt;830</text>\r\n",
       "</g>\r\n",
       "<!-- 16&#45;&gt;21 -->\r\n",
       "<g id=\"edge9\" class=\"edge\"><title>16&#45;&gt;21</title>\r\n",
       "<path fill=\"none\" stroke=\"#0000ff\" d=\"M299.852,-260.733C293.637,-255.342 287.047,-249.181 281.5,-243 274.848,-235.587 268.362,-226.876 262.817,-218.821\"/>\r\n",
       "<polygon fill=\"#0000ff\" stroke=\"#0000ff\" points=\"265.674,-216.797 257.206,-210.429 259.855,-220.688 265.674,-216.797\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"316\" y=\"-231.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">yes, missing</text>\r\n",
       "</g>\r\n",
       "<!-- 22 -->\r\n",
       "<g id=\"node11\" class=\"node\"><title>22</title>\r\n",
       "<polygon fill=\"GreenYellow\" stroke=\"black\" points=\"462,-210 329,-210 329,-174 462,-174 462,-210\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"395.5\" y=\"-188.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">leaf=4.47986577e&#45;06</text>\r\n",
       "</g>\r\n",
       "<!-- 16&#45;&gt;22 -->\r\n",
       "<g id=\"edge10\" class=\"edge\"><title>16&#45;&gt;22</title>\r\n",
       "<path fill=\"none\" stroke=\"#ff0000\" d=\"M336.475,-260.799C347.207,-248.471 361.827,-231.679 373.831,-217.89\"/>\r\n",
       "<polygon fill=\"#ff0000\" stroke=\"#ff0000\" points=\"376.621,-220.016 380.547,-210.175 371.341,-215.42 376.621,-220.016\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"371\" y=\"-231.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">no</text>\r\n",
       "</g>\r\n",
       "<!-- 23 -->\r\n",
       "<g id=\"node12\" class=\"node\"><title>23</title>\r\n",
       "<polygon fill=\"GreenYellow\" stroke=\"black\" points=\"267.5,-123 133.5,-123 133.5,-87 267.5,-87 267.5,-123\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"200.5\" y=\"-101.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">leaf=&#45;0.00661460264</text>\r\n",
       "</g>\r\n",
       "<!-- 21&#45;&gt;23 -->\r\n",
       "<g id=\"edge11\" class=\"edge\"><title>21&#45;&gt;23</title>\r\n",
       "<path fill=\"none\" stroke=\"#0000ff\" d=\"M237.191,-173.799C230.707,-161.817 221.94,-145.617 214.601,-132.057\"/>\r\n",
       "<polygon fill=\"#0000ff\" stroke=\"#0000ff\" points=\"217.633,-130.304 209.795,-123.175 211.476,-133.636 217.633,-130.304\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"261\" y=\"-144.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">yes, missing</text>\r\n",
       "</g>\r\n",
       "<!-- 24 -->\r\n",
       "<g id=\"node13\" class=\"node\"><title>24</title>\r\n",
       "<path fill=\"LightBlue\" stroke=\"black\" d=\"M457.5,-123C457.5,-123 297.5,-123 297.5,-123 291.5,-123 285.5,-117 285.5,-111 285.5,-111 285.5,-99 285.5,-99 285.5,-93 291.5,-87 297.5,-87 297.5,-87 457.5,-87 457.5,-87 463.5,-87 469.5,-93 469.5,-99 469.5,-99 469.5,-111 469.5,-111 469.5,-117 463.5,-123 457.5,-123\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"377.5\" y=\"-101.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">주관적_건강수준인지율&lt;45.5</text>\r\n",
       "</g>\r\n",
       "<!-- 21&#45;&gt;24 -->\r\n",
       "<g id=\"edge12\" class=\"edge\"><title>21&#45;&gt;24</title>\r\n",
       "<path fill=\"none\" stroke=\"#ff0000\" d=\"M273.01,-173.799C292.986,-160.838 320.568,-142.941 342.378,-128.789\"/>\r\n",
       "<polygon fill=\"#ff0000\" stroke=\"#ff0000\" points=\"344.546,-131.555 351.03,-123.175 340.736,-125.683 344.546,-131.555\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"328\" y=\"-144.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">no</text>\r\n",
       "</g>\r\n",
       "<!-- 25 -->\r\n",
       "<g id=\"node14\" class=\"node\"><title>25</title>\r\n",
       "<polygon fill=\"GreenYellow\" stroke=\"black\" points=\"369.5,-36 235.5,-36 235.5,-0 369.5,-0 369.5,-36\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"302.5\" y=\"-14.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">leaf=&#45;0.00352568133</text>\r\n",
       "</g>\r\n",
       "<!-- 24&#45;&gt;25 -->\r\n",
       "<g id=\"edge13\" class=\"edge\"><title>24&#45;&gt;25</title>\r\n",
       "<path fill=\"none\" stroke=\"#0000ff\" d=\"M362.323,-86.799C351.445,-74.4713 336.628,-57.6787 324.462,-43.8902\"/>\r\n",
       "<polygon fill=\"#0000ff\" stroke=\"#0000ff\" points=\"326.895,-41.3582 317.655,-36.1754 321.647,-45.9895 326.895,-41.3582\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"379\" y=\"-57.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">yes, missing</text>\r\n",
       "</g>\r\n",
       "<!-- 26 -->\r\n",
       "<g id=\"node15\" class=\"node\"><title>26</title>\r\n",
       "<polygon fill=\"GreenYellow\" stroke=\"black\" points=\"517,-36 388,-36 388,-0 517,-0 517,-36\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"452.5\" y=\"-14.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">leaf=0.00118756271</text>\r\n",
       "</g>\r\n",
       "<!-- 24&#45;&gt;26 -->\r\n",
       "<g id=\"edge14\" class=\"edge\"><title>24&#45;&gt;26</title>\r\n",
       "<path fill=\"none\" stroke=\"#ff0000\" d=\"M399.148,-86.7329C405.363,-81.3419 411.953,-75.1811 417.5,-69 424.152,-61.5869 430.638,-52.8764 436.183,-44.8208\"/>\r\n",
       "<polygon fill=\"#ff0000\" stroke=\"#ff0000\" points=\"439.145,-46.6876 441.794,-36.4291 433.326,-42.7968 439.145,-46.6876\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"437\" y=\"-57.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">no</text>\r\n",
       "</g>\r\n",
       "<!-- 11 -->\r\n",
       "<g id=\"node20\" class=\"node\"><title>11</title>\r\n",
       "<path fill=\"LightBlue\" stroke=\"black\" d=\"M739.5,-384C739.5,-384 635.5,-384 635.5,-384 629.5,-384 623.5,-378 623.5,-372 623.5,-372 623.5,-360 623.5,-360 623.5,-354 629.5,-348 635.5,-348 635.5,-348 739.5,-348 739.5,-348 745.5,-348 751.5,-354 751.5,-360 751.5,-360 751.5,-372 751.5,-372 751.5,-378 745.5,-384 739.5,-384\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"687.5\" y=\"-362.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">비만율&lt;30.3999996</text>\r\n",
       "</g>\r\n",
       "<!-- 5&#45;&gt;11 -->\r\n",
       "<g id=\"edge19\" class=\"edge\"><title>5&#45;&gt;11</title>\r\n",
       "<path fill=\"none\" stroke=\"#0000ff\" d=\"M717.81,-434.799C712.505,-422.932 705.35,-406.928 699.324,-393.449\"/>\r\n",
       "<polygon fill=\"#0000ff\" stroke=\"#0000ff\" points=\"702.455,-391.876 695.178,-384.175 696.065,-394.733 702.455,-391.876\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"744\" y=\"-405.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">yes, missing</text>\r\n",
       "</g>\r\n",
       "<!-- 12 -->\r\n",
       "<g id=\"node21\" class=\"node\"><title>12</title>\r\n",
       "<polygon fill=\"GreenYellow\" stroke=\"black\" points=\"891.5,-384 769.5,-384 769.5,-348 891.5,-348 891.5,-384\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"830.5\" y=\"-362.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">leaf=0.0191237871</text>\r\n",
       "</g>\r\n",
       "<!-- 5&#45;&gt;12 -->\r\n",
       "<g id=\"edge20\" class=\"edge\"><title>5&#45;&gt;12</title>\r\n",
       "<path fill=\"none\" stroke=\"#ff0000\" d=\"M756.644,-434.848C765.321,-429.551 774.543,-423.412 782.5,-417 791.826,-409.486 801.156,-400.256 809.077,-391.781\"/>\r\n",
       "<polygon fill=\"#ff0000\" stroke=\"#ff0000\" points=\"811.764,-394.029 815.915,-384.281 806.591,-389.313 811.764,-394.029\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"806\" y=\"-405.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">no</text>\r\n",
       "</g>\r\n",
       "<!-- 13 -->\r\n",
       "<g id=\"node24\" class=\"node\"><title>13</title>\r\n",
       "<path fill=\"LightBlue\" stroke=\"black\" d=\"M1063.5,-384C1063.5,-384 921.5,-384 921.5,-384 915.5,-384 909.5,-378 909.5,-372 909.5,-372 909.5,-360 909.5,-360 909.5,-354 915.5,-348 921.5,-348 921.5,-348 1063.5,-348 1063.5,-348 1069.5,-348 1075.5,-354 1075.5,-360 1075.5,-360 1075.5,-372 1075.5,-372 1075.5,-378 1069.5,-384 1063.5,-384\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"992.5\" y=\"-362.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">상수도보급률&lt;69.8499985</text>\r\n",
       "</g>\r\n",
       "<!-- 6&#45;&gt;13 -->\r\n",
       "<g id=\"edge23\" class=\"edge\"><title>6&#45;&gt;13</title>\r\n",
       "<path fill=\"none\" stroke=\"#0000ff\" d=\"M992.5,-434.799C992.5,-423.163 992.5,-407.548 992.5,-394.237\"/>\r\n",
       "<polygon fill=\"#0000ff\" stroke=\"#0000ff\" points=\"996,-394.175 992.5,-384.175 989,-394.175 996,-394.175\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"1027\" y=\"-405.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">yes, missing</text>\r\n",
       "</g>\r\n",
       "<!-- 14 -->\r\n",
       "<g id=\"node25\" class=\"node\"><title>14</title>\r\n",
       "<polygon fill=\"GreenYellow\" stroke=\"black\" points=\"1223,-384 1094,-384 1094,-348 1223,-348 1223,-384\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"1158.5\" y=\"-362.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">leaf=0.00947291311</text>\r\n",
       "</g>\r\n",
       "<!-- 6&#45;&gt;14 -->\r\n",
       "<g id=\"edge24\" class=\"edge\"><title>6&#45;&gt;14</title>\r\n",
       "<path fill=\"none\" stroke=\"#ff0000\" d=\"M1029.05,-434.882C1040.75,-429.294 1053.72,-422.986 1065.5,-417 1083.15,-408.03 1102.46,-397.73 1118.89,-388.818\"/>\r\n",
       "<polygon fill=\"#ff0000\" stroke=\"#ff0000\" points=\"1120.58,-391.882 1127.69,-384.027 1117.23,-385.734 1120.58,-391.882\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"1102\" y=\"-405.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">no</text>\r\n",
       "</g>\r\n",
       "<!-- 17 -->\r\n",
       "<g id=\"node22\" class=\"node\"><title>17</title>\r\n",
       "<polygon fill=\"GreenYellow\" stroke=\"black\" points=\"617.5,-297 483.5,-297 483.5,-261 617.5,-261 617.5,-297\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"550.5\" y=\"-275.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">leaf=&#45;0.00400540652</text>\r\n",
       "</g>\r\n",
       "<!-- 11&#45;&gt;17 -->\r\n",
       "<g id=\"edge21\" class=\"edge\"><title>11&#45;&gt;17</title>\r\n",
       "<path fill=\"none\" stroke=\"#0000ff\" d=\"M651.075,-347.929C640.38,-342.544 628.795,-336.337 618.5,-330 605.446,-321.965 591.706,-312.148 580.02,-303.347\"/>\r\n",
       "<polygon fill=\"#0000ff\" stroke=\"#0000ff\" points=\"581.986,-300.444 571.913,-297.158 577.738,-306.008 581.986,-300.444\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"653\" y=\"-318.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">yes, missing</text>\r\n",
       "</g>\r\n",
       "<!-- 18 -->\r\n",
       "<g id=\"node23\" class=\"node\"><title>18</title>\r\n",
       "<polygon fill=\"GreenYellow\" stroke=\"black\" points=\"757.5,-297 635.5,-297 635.5,-261 757.5,-261 757.5,-297\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"696.5\" y=\"-275.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">leaf=0.0115977377</text>\r\n",
       "</g>\r\n",
       "<!-- 11&#45;&gt;18 -->\r\n",
       "<g id=\"edge22\" class=\"edge\"><title>11&#45;&gt;18</title>\r\n",
       "<path fill=\"none\" stroke=\"#ff0000\" d=\"M689.321,-347.799C690.553,-336.163 692.207,-320.548 693.616,-307.237\"/>\r\n",
       "<polygon fill=\"#ff0000\" stroke=\"#ff0000\" points=\"697.109,-307.488 694.681,-297.175 690.148,-306.751 697.109,-307.488\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"701\" y=\"-318.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">no</text>\r\n",
       "</g>\r\n",
       "<!-- 19 -->\r\n",
       "<g id=\"node26\" class=\"node\"><title>19</title>\r\n",
       "<polygon fill=\"GreenYellow\" stroke=\"black\" points=\"913,-297 784,-297 784,-261 913,-261 913,-297\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"848.5\" y=\"-275.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">leaf=0.00118298805</text>\r\n",
       "</g>\r\n",
       "<!-- 13&#45;&gt;19 -->\r\n",
       "<g id=\"edge25\" class=\"edge\"><title>13&#45;&gt;19</title>\r\n",
       "<path fill=\"none\" stroke=\"#0000ff\" d=\"M956.05,-347.899C945.105,-342.465 933.175,-336.239 922.5,-330 908.361,-321.736 893.292,-311.78 880.467,-302.934\"/>\r\n",
       "<polygon fill=\"#0000ff\" stroke=\"#0000ff\" points=\"882.191,-299.869 871.986,-297.022 878.188,-305.612 882.191,-299.869\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"957\" y=\"-318.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">yes, missing</text>\r\n",
       "</g>\r\n",
       "<!-- 20 -->\r\n",
       "<g id=\"node27\" class=\"node\"><title>20</title>\r\n",
       "<polygon fill=\"GreenYellow\" stroke=\"black\" points=\"1065.5,-297 931.5,-297 931.5,-261 1065.5,-261 1065.5,-297\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"998.5\" y=\"-275.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">leaf=&#45;0.00499183591</text>\r\n",
       "</g>\r\n",
       "<!-- 13&#45;&gt;20 -->\r\n",
       "<g id=\"edge26\" class=\"edge\"><title>13&#45;&gt;20</title>\r\n",
       "<path fill=\"none\" stroke=\"#ff0000\" d=\"M993.714,-347.799C994.536,-336.163 995.638,-320.548 996.577,-307.237\"/>\r\n",
       "<polygon fill=\"#ff0000\" stroke=\"#ff0000\" points=\"1000.07,-307.397 997.288,-297.175 993.092,-306.904 1000.07,-307.397\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"1004\" y=\"-318.8\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">no</text>\r\n",
       "</g>\r\n",
       "</g>\r\n",
       "</svg>\r\n"
      ],
      "text/plain": [
       "<graphviz.files.Source at 0x2ce422b6fc8>"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 트리모델 시각화\n",
    "xgb.to_graphviz(model, num_trees=4, size='5,10',\n",
    "               condition_node_params=node_params,\n",
    "               leaf_node_params=leaf_params)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 파라미터 튜닝"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> 파라미터 튜닝시 RandomizedSearchCV를 이용하였음"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 최적의 파라미터 검색을 위한 파라미터 정의\n",
    "\n",
    "xgb_clf = xgb.XGBClassifier(eval_metric = [\"merror\", \"map\", \"auc\"],silent=0 ,objective = \"multi:softmax\",random_state=16)\n",
    "parameters = {\"learning_rate\": [0.1, 0.01, 0.001],\n",
    "               \"gamma\" : [0.01, 1, 2],\n",
    "               \"max_depth\": range(1,50),\n",
    "               \"colsample_bytree\": [0.3, 0.8, 1.0],\n",
    "               \"colsample_bylevel\" : [0.8,0.85,0.9],\n",
    "               \"subsample\": [0.8,0.85,0.88,0.9],\n",
    "               'num_class': [1,4,8],\n",
    "               \"reg_lambda\": [0, 0.5, 1],\n",
    "               \"eta\" : [0.1,0.01],\n",
    "               \"min_child_weight\": [1, 3, 5, 7],\n",
    "               \"n_estimators\": range(500,2000)}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    ">xgb.XGBClassifier(eval_metric = [\"merror\", \"map\", \"auc\"]\n",
    ">\n",
    ">  merror: Multiclass classification error rate. It is calculated as #(wrong cases)/#(all cases) \n",
    ">\n",
    "> map: Mean Average Precision \n",
    ">\n",
    "> auc: Area under the curve"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 10 folds for each of 10 candidates, totalling 100 fits\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[02:49:11] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.\n",
      "[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    5.0s remaining:    0.0s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.850, total=   5.1s\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[02:49:16] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.809, total=   5.3s\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[02:49:21] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=1)]: Done   2 out of   2 | elapsed:   10.2s remaining:    0.0s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.807, total=   5.0s\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[02:49:26] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.925, total=   5.4s\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[02:49:32] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.883, total=   5.1s\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[02:49:37] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.852, total=   5.2s\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[02:49:42] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.882, total=   5.0s\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[02:49:47] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.971, total=   4.9s\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[02:49:52] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.807, total=   5.2s\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[02:49:57] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0, num_class=4, n_estimators=1663, min_child_weight=7, max_depth=30, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.926, total=   5.0s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9 \n",
      "[02:50:02] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9, score=0.882, total=   5.3s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9 \n",
      "[02:50:07] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9, score=0.826, total=   5.1s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9 \n",
      "[02:50:13] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9, score=0.811, total=   5.5s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9 \n",
      "[02:50:18] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9, score=0.897, total=   5.3s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9 \n",
      "[02:50:23] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9, score=0.927, total=   5.4s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9 \n",
      "[02:50:29] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9, score=0.866, total=   5.4s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9 \n",
      "[02:50:34] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9, score=0.881, total=   5.1s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9 \n",
      "[02:50:39] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9, score=0.912, total=   5.2s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9 \n",
      "[02:50:44] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9, score=0.837, total=   5.1s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9 \n",
      "[02:50:50] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.88, reg_lambda=0, num_class=8, n_estimators=597, min_child_weight=3, max_depth=28, learning_rate=0.1, gamma=2, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.9, score=0.896, total=   5.3s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[02:50:55] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.866, total=   2.6s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[02:50:57] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.824, total=   2.6s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[02:51:00] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.868, total=   2.3s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[02:51:02] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.925, total=   2.5s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[02:51:05] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.883, total=   2.8s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[02:51:08] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.881, total=   2.7s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[02:51:10] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.882, total=   2.4s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[02:51:13] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.971, total=   2.4s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[02:51:15] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.865, total=   2.4s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[02:51:18] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=994, min_child_weight=5, max_depth=20, learning_rate=0.1, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.911, total=   2.6s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:51:20] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.866, total=   5.6s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:51:26] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.827, total=   5.8s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:51:32] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.842, total=   5.9s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:51:38] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.911, total=   5.9s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:51:44] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.927, total=   5.8s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:51:49] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.898, total=   5.7s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:51:55] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.867, total=   5.9s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:52:01] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.911, total=   5.8s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:52:07] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.837, total=   6.3s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:52:13] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=728, min_child_weight=3, max_depth=30, learning_rate=0.01, gamma=1, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.911, total=   5.8s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:52:19] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.866, total=  16.5s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:52:35] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.813, total=  16.0s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:52:51] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.870, total=  15.9s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:53:07] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.911, total=  16.8s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:53:24] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.926, total=  16.2s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:53:40] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.912, total=  15.8s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:53:56] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.897, total=  16.2s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:54:12] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.941, total=  16.1s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:54:28] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.851, total=  15.8s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:54:44] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1419, min_child_weight=1, max_depth=29, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.897, total=  16.0s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[02:55:00] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.850, total=   7.4s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[02:55:08] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.810, total=   7.7s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[02:55:15] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.824, total=   7.3s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[02:55:23] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.863, total=   7.1s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[02:55:30] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.868, total=   7.2s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[02:55:37] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.851, total=   7.1s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[02:55:44] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.854, total=   7.3s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[02:55:51] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.867, total=   7.3s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[02:55:58] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.844, total=   7.1s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[02:56:06] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=0, num_class=8, n_estimators=916, min_child_weight=5, max_depth=12, learning_rate=0.001, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.867, total=   7.2s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:56:13] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.866, total=   2.1s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:56:15] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.809, total=   1.7s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:56:17] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.853, total=   1.9s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:56:18] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.941, total=   1.6s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:56:20] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.868, total=   2.0s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:56:22] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.867, total=   1.6s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:56:24] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.853, total=   1.8s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:56:26] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.941, total=   1.8s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:56:27] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.837, total=   1.9s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[02:56:29] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=4, n_estimators=524, min_child_weight=7, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.896, total=   1.9s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[02:56:31] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.799, total=   2.7s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[02:56:34] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.824, total=   2.9s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[02:56:37] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.838, total=   2.7s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[02:56:39] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.863, total=   2.9s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[02:56:42] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.868, total=   2.8s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[02:56:45] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.835, total=   2.8s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[02:56:48] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.823, total=   2.7s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[02:56:51] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.867, total=   2.6s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[02:56:53] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.848, total=   2.8s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[02:56:56] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.85, reg_lambda=1, num_class=8, n_estimators=559, min_child_weight=7, max_depth=24, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.808, total=   2.7s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9 \n",
      "[02:56:59] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9, score=0.846, total=   6.3s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9 \n",
      "[02:57:05] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9, score=0.812, total=   6.5s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9 \n",
      "[02:57:12] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9, score=0.799, total=   6.0s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9 \n",
      "[02:57:18] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9, score=0.926, total=   6.1s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9 \n",
      "[02:57:24] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9, score=0.897, total=   6.1s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9 \n",
      "[02:57:30] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9, score=0.798, total=   5.8s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9 \n",
      "[02:57:36] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9, score=0.869, total=   6.1s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9 \n",
      "[02:57:42] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9, score=0.912, total=   6.1s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9 \n",
      "[02:57:48] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9, score=0.740, total=   5.8s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9 \n",
      "[02:57:54] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=626, min_child_weight=7, max_depth=19, learning_rate=0.001, gamma=0.01, eta=0.1, colsample_bytree=1.0, colsample_bylevel=0.9, score=0.825, total=   5.9s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[02:58:00] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.850, total=   6.2s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[02:58:06] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.841, total=   6.2s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[02:58:12] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.823, total=   6.1s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[02:58:18] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.927, total=   6.4s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[02:58:25] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.897, total=   6.1s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[02:58:31] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.898, total=   5.9s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[02:58:37] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.882, total=   6.2s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[02:58:43] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.941, total=   6.3s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[02:58:49] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.823, total=   6.9s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[02:58:56] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=8, n_estimators=1021, min_child_weight=5, max_depth=48, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.882, total=   6.9s\n",
      "[02:59:03] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=1)]: Done 100 out of 100 | elapsed:  9.9min finished\n"
     ]
    }
   ],
   "source": [
    "# RandomizedSearchCV를  사용하여 최적의 파라미터 검색 \n",
    "# stratifed kfold 이용\n",
    "xgb_rscv = RandomizedSearchCV(xgb_clf, param_distributions = parameters, scoring = 'f1_weighted',\n",
    "                             cv = kfold, verbose = 3, random_state = 16)\n",
    "model_xgboost = xgb_rscv.fit(dat1, target)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> RandomizedSearchCV를 통해서 모델에 최적화된 파라미터를 찾음"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'subsample': 0.85,\n",
       " 'reg_lambda': 0.5,\n",
       " 'num_class': 1,\n",
       " 'n_estimators': 1419,\n",
       " 'min_child_weight': 1,\n",
       " 'max_depth': 29,\n",
       " 'learning_rate': 0.001,\n",
       " 'gamma': 0.01,\n",
       " 'eta': 0.1,\n",
       " 'colsample_bytree': 0.8,\n",
       " 'colsample_bylevel': 0.8}"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 최적의 파라미터 목록\n",
    "\n",
    "best_param = model_xgboost.best_params_\n",
    "best_param"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> xgboost 최적화된 파라미터"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "XGBClassifier(base_score=0.5, booster='gbtree', colsample_bylevel=0.8,\n",
       "              colsample_bynode=1, colsample_bytree=0.8, eta=0.1, gamma=0.01,\n",
       "              gpu_id=-1, importance_type='gain', interaction_constraints='',\n",
       "              learning_rate=0.001, max_delta_step=0, max_depth=29,\n",
       "              min_child_weight=1, missing=nan, monotone_constraints='()',\n",
       "              n_estimators=1419, n_jobs=0, num_class=1, num_parallel_tree=1,\n",
       "              objective='multi:softprob', random_state=0, reg_alpha=0,\n",
       "              reg_lambda=0.5, scale_pos_weight=None, subsample=0.85,\n",
       "              tree_method='exact', validate_parameters=1, verbosity=None)"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_xgb = xgb.XGBClassifier(**best_param)\n",
    "model_xgb.fit(dat1, target)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "Predicted_xgb= model_xgb.predict(test_x1)\n",
    "test_xgb_f1 = cross_val_score(model_xgb, dat1, target, cv=kfold, scoring='f1_weighted').mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "F1 Cross_validate 0.8870841092994164\n",
      "F1 Macro: 0.8863759657543425\n",
      "F1 Micro: 0.8903508771929824\n",
      "F1 Weighted: 0.8917989593415837\n",
      "\n",
      "Matrix of confusion\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([[66,  8,  0,  0],\n",
       "       [ 0, 44,  5,  0],\n",
       "       [ 0,  2, 57,  5],\n",
       "       [ 0,  0,  5, 36]], dtype=int64)"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "four_f1(test_xgb_f1,Predicted_xgb)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> 첫 파라미터 조절 후 f1_score 89.1% "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "xgb_clf = xgb.XGBClassifier(eval_metric = [\"merror\", \"map\", \"auc\"],silent=0 ,objective = \"multi:softmax\",random_state=16)\n",
    "parameters = {\"learning_rate\": [0.1, 0.01, 0.001],\n",
    "               \"gamma\" : [0.01, 1, 2],\n",
    "               \"max_depth\": range(20,40),\n",
    "               \"colsample_bytree\": [0.3, 0.8, 1.0],\n",
    "               \"colsample_bylevel\" : [0.8,0.85,0.9],\n",
    "               \"subsample\": [0.8,0.85,0.88,0.9],\n",
    "               'num_class': range(1,5),\n",
    "               \"reg_lambda\": [0, 0.5, 1],\n",
    "               \"eta\" : [0.1,0.01],\n",
    "               \"min_child_weight\": range(1,10),\n",
    "               \"n_estimators\": range(500,2000)}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> 두번째 파라미터 조정 같은 경우에는 스코어에 영향이 클거같은 n_estimators, max_depth, num_class에 대해 하였음"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 10 folds for each of 10 candidates, totalling 100 fits\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[03:02:50] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.865, total=   9.2s\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[03:02:59] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    9.1s remaining:    0.0s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.826, total=   9.0s\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[03:03:08] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=1)]: Done   2 out of   2 | elapsed:   18.1s remaining:    0.0s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.823, total=   9.4s\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[03:03:17] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.941, total=   9.4s\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[03:03:27] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.868, total=   9.4s\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[03:03:36] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.867, total=   9.0s\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[03:03:45] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.882, total=   9.0s\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[03:03:54] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.941, total=   9.2s\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[03:04:04] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.823, total=   9.2s\n",
      "[CV] subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85 \n",
      "[03:04:13] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0, num_class=3, n_estimators=1747, min_child_weight=9, max_depth=39, learning_rate=0.01, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.85, score=0.896, total=   9.3s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:04:22] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.882, total=   4.6s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:04:27] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.793, total=   4.6s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:04:31] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.839, total=   4.6s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:04:36] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.941, total=   4.6s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:04:40] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.883, total=   4.7s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:04:45] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.896, total=   4.7s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:04:50] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.868, total=   4.6s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:04:54] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.971, total=   4.7s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:04:59] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.865, total=   4.7s\n",
      "[CV] subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:05:04] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.88, reg_lambda=0, num_class=2, n_estimators=1698, min_child_weight=1, max_depth=25, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.940, total=   4.7s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:05:09] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.850, total=   8.0s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:05:17] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.810, total=   8.0s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:05:25] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.855, total=   8.3s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:05:33] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.895, total=   8.1s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:05:41] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.882, total=   7.9s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:05:49] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.881, total=   7.2s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:05:56] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.882, total=   7.3s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:06:03] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.897, total=   7.1s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:06:10] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.803, total=   7.3s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:06:18] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=1, n_estimators=1621, min_child_weight=5, max_depth=33, learning_rate=0.01, gamma=2, eta=0.1, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.882, total=   7.4s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[03:06:25] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.882, total=   5.5s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[03:06:31] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.869, total=   5.6s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[03:06:36] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.838, total=   5.5s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[03:06:42] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.927, total=   5.4s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[03:06:47] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.868, total=   5.5s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[03:06:53] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.898, total=   5.4s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[03:06:58] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.882, total=   5.4s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[03:07:04] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.956, total=   5.8s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[03:07:09] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.837, total=   5.7s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8 \n",
      "[03:07:15] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=1, n_estimators=1796, min_child_weight=1, max_depth=29, learning_rate=0.1, gamma=0.01, eta=0.1, colsample_bytree=0.8, colsample_bylevel=0.8, score=0.955, total=   5.2s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:07:20] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.850, total=   8.2s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:07:28] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.810, total=   8.2s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:07:37] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.869, total=   8.0s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:07:45] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.881, total=   7.9s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:07:53] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.897, total=   8.3s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:08:01] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.881, total=   7.8s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:08:09] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.882, total=   7.9s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:08:17] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.897, total=   8.2s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:08:25] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.835, total=   7.7s\n",
      "[CV] subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:08:33] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=0.5, num_class=2, n_estimators=1189, min_child_weight=1, max_depth=36, learning_rate=0.01, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.882, total=   7.9s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[03:08:41] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.878, total=   9.0s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[03:08:50] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.842, total=   9.0s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[03:08:59] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.826, total=   9.1s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[03:09:08] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.911, total=   9.3s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[03:09:17] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.912, total=   8.9s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[03:09:26] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.869, total=   8.9s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[03:09:35] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.897, total=   9.0s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[03:09:44] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.941, total=   9.1s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[03:09:53] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.821, total=   8.6s\n",
      "[CV] subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85 \n",
      "[03:10:02] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=0, num_class=3, n_estimators=1562, min_child_weight=5, max_depth=25, learning_rate=0.1, gamma=1, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.85, score=0.883, total=   9.0s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:10:11] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.850, total=   4.1s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:10:15] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.826, total=   4.4s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:10:19] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.883, total=   4.3s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:10:23] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.911, total=   4.1s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:10:27] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.897, total=   4.3s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:10:32] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.867, total=   4.4s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:10:36] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.897, total=   4.1s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:10:40] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.911, total=   4.3s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:10:44] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.803, total=   4.1s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8 \n",
      "[03:10:49] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=893, min_child_weight=3, max_depth=22, learning_rate=0.1, gamma=2, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.8, score=0.882, total=   4.3s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8 \n",
      "[03:10:53] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8, score=0.878, total=   6.1s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8 \n",
      "[03:10:59] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8, score=0.797, total=   6.1s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8 \n",
      "[03:11:05] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8, score=0.793, total=   6.5s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8 \n",
      "[03:11:12] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8, score=0.911, total=   6.6s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8 \n",
      "[03:11:18] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8, score=0.912, total=   6.2s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8 \n",
      "[03:11:24] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8, score=0.881, total=   6.3s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8 \n",
      "[03:11:31] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8, score=0.867, total=   6.3s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8 \n",
      "[03:11:37] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8, score=0.956, total=   6.2s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8 \n",
      "[03:11:43] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8, score=0.823, total=   6.0s\n",
      "[CV] subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8 \n",
      "[03:11:49] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.85, reg_lambda=1, num_class=4, n_estimators=919, min_child_weight=6, max_depth=22, learning_rate=0.01, gamma=0.01, eta=0.01, colsample_bytree=1.0, colsample_bylevel=0.8, score=0.883, total=   6.0s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:11:55] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.813, total=   4.8s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:12:00] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.824, total=   4.6s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:12:05] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.824, total=   5.0s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:12:10] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.880, total=   4.6s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:12:14] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.868, total=   4.6s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:12:19] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.835, total=   4.5s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:12:23] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.823, total=   4.7s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:12:28] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.867, total=   4.8s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:12:33] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.848, total=   4.7s\n",
      "[CV] subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9 \n",
      "[03:12:38] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.9, reg_lambda=1, num_class=4, n_estimators=969, min_child_weight=9, max_depth=30, learning_rate=0.001, gamma=0.01, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.9, score=0.823, total=   4.9s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[03:12:43] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.850, total=   8.1s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[03:12:51] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.810, total=   8.6s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[03:12:59] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.854, total=   8.7s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[03:13:08] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.895, total=   9.1s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[03:13:17] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.883, total=  10.0s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[03:13:27] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.881, total=   8.6s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[03:13:36] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.912, total=   8.5s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[03:13:44] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.941, total=   8.6s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[03:13:53] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.835, total=   8.7s\n",
      "[CV] subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85 \n",
      "[03:14:02] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n",
      "[CV]  subsample=0.8, reg_lambda=1, num_class=2, n_estimators=1641, min_child_weight=2, max_depth=35, learning_rate=0.01, gamma=1, eta=0.01, colsample_bytree=0.3, colsample_bylevel=0.85, score=0.896, total=   8.6s\n",
      "[03:14:10] WARNING: C:\\Users\\Administrator\\workspace\\xgboost-win64_release_1.2.0\\src\\learner.cc:516: \n",
      "Parameters: { silent } might not be used.\n",
      "\n",
      "  This may not be accurate due to some parameters are only used in language bindings but\n",
      "  passed down to XGBoost core.  Or some parameters are not used but slip through this\n",
      "  verification. Please open an issue if you find above cases.\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=1)]: Done 100 out of 100 | elapsed: 11.3min finished\n"
     ]
    }
   ],
   "source": [
    "xgb_rscv1 = RandomizedSearchCV(xgb_clf, param_distributions = parameters, scoring = 'f1_weighted',\n",
    "                             cv = kfold, verbose = 3, random_state = 16)\n",
    "model_xgboost1 = xgb_rscv1.fit(dat1, target)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'subsample': 0.85,\n",
       " 'reg_lambda': 0.5,\n",
       " 'num_class': 1,\n",
       " 'n_estimators': 1796,\n",
       " 'min_child_weight': 1,\n",
       " 'max_depth': 29,\n",
       " 'learning_rate': 0.1,\n",
       " 'gamma': 0.01,\n",
       " 'eta': 0.1,\n",
       " 'colsample_bytree': 0.8,\n",
       " 'colsample_bylevel': 0.8}"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#RandomizedSearchCV를 통한 내 데이터셋에 최적화된 파라미터\n",
    "best_param1 = model_xgboost1.best_params_\n",
    "best_param1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "XGBClassifier(base_score=0.5, booster='gbtree', colsample_bylevel=0.8,\n",
       "              colsample_bynode=1, colsample_bytree=0.8, eta=0.1, gamma=0.01,\n",
       "              gpu_id=-1, importance_type='gain', interaction_constraints='',\n",
       "              learning_rate=0.1, max_delta_step=0, max_depth=29,\n",
       "              min_child_weight=1, missing=nan, monotone_constraints='()',\n",
       "              n_estimators=1796, n_jobs=0, num_class=1, num_parallel_tree=1,\n",
       "              objective='multi:softprob', random_state=0, reg_alpha=0,\n",
       "              reg_lambda=0.5, scale_pos_weight=None, subsample=0.85,\n",
       "              tree_method='exact', validate_parameters=1, verbosity=None)"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_xgb1 = xgb.XGBClassifier(**best_param1)\n",
    "model_xgb1.fit(dat1, target)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "Predicted_xgb1= model_xgb1.predict(test_x1)\n",
    "test_xgb_f1_1 = cross_val_score(model_xgb1, dat1, target, cv=kfold, scoring='f1_weighted').mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "F1 Cross_validate 0.8861209383882039\n",
      "F1 Macro: 0.8987678193651479\n",
      "F1 Micro: 0.9035087719298246\n",
      "F1 Weighted: 0.9045246073567706\n",
      "\n",
      "Matrix of confusion\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([[68,  6,  0,  0],\n",
       "       [ 0, 45,  4,  0],\n",
       "       [ 0,  2, 57,  5],\n",
       "       [ 0,  0,  5, 36]], dtype=int64)"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "four_f1(test_xgb_f1_1,Predicted_xgb1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='Predicted Label', ylabel='Actual Label'>"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cnf_matrix_model(Predicted_xgb1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> 3개의 파라미터를 조정했더니 f1_score가 90.4%까지 상승 "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### XGB voting"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    ">class_weightdict_'balanced' y 값을 사용하여 입력 데이터의 클래스 빈도에 반비례하여 가중치를 자동으로 조정합니다. 다중 출력의 경우 y의 각 열의 가중치가 곱해집니다.\n",
    ">sample_weight -> 클래스 가중치는 클래스 불균형을 수정하기 위해 가중치를 변경합니다\n",
    ">\n",
    ">sample_weight 사용법은 학습 데이터 세트의 목표 클래스를 균형 잡는 것이 목적이라는 것을 기억하고 sample_weight 간단합니다. 즉, X 가 관측 값이고 y 가 클래스 (레이블) 인 경우 len(X) == len(y) == len(sample_wight)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "#target 불균형에 대해 class_weigt 라이브러리 사용\n",
    "# 클래스 불균형 문제를 해소하기 위해 compute_sample_weight 사용\n",
    "\n",
    "y_train_weights = class_weight.compute_sample_weight('balanced', target, indices=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "#파라미터 설정\n",
    "fit_params={'verbose': False}\n",
    "fit_params['verbose'] = 50"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 여러가지 모델을 사용해보았지만 XGBoost 제외하고는 성능이 굉장히 떨어지는 모습을 보였음\n",
    "# 다중 분류기들을 Voting하는 방법을 사용하기 위해 XGBoost 모델의 random_state(난수값)을 변경해가면서 10개의 XGB 분류기를 만들어냄\n",
    "# 파라미터는 위에서 정의해준 최적 파라미터 사용 \n",
    "\n",
    "clfs = []\n",
    "for i in range(10):\n",
    "    clf = xgb.XGBClassifier(random_state=217+i, n_jobs=4, **best_param1)\n",
    "    \n",
    "    clfs.append(('xgb{}'.format(i), clf))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "# VotingClassifier 정의 \n",
    "# 위에서 정의해준 10개의 XGB 분류기를 사용\n",
    "# 앙상블의 소프트 보팅\n",
    "vc = VotingClassifier(clfs, voting='soft')\n",
    "del(clfs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    ">voting 서로 다른 ml알고리즘으로 여러 개의 분류기를 생성하고,투표를 통해 최종 예측 결과를 결정하는 방식\n",
    ">\n",
    ">하드 보팅 - 분류기 간 닫수결로 최종 분류 값 예측\n",
    ">\n",
    ">소프트 보팅 - 분류기들의 분류 확률들을 평균 내어 최종 분류값 예측"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 보팅에 불균형을 해결하기 위해 이용했떤 라이브러리를 sample_weight 파라미터를 통해 이용\n",
    "# 10개의 XGB 분류기로 이루어진 VotingClassifier로 학습 \n",
    "\n",
    "_ = vc.fit(dat1, target, sample_weight=y_train_weights)\n",
    "\n",
    "clf_final = vc.estimators_[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Validation score of a single XGB Classifier: 0.9046\n",
      "Validation score of a VotingClassifier on 3 XGB with hard voting strategy: 0.9045\n",
      "Validation score of a VotingClassifier on 3 XGB with soft voting strategy: 0.9045\n"
     ]
    }
   ],
   "source": [
    "# 단일 모델, soft voting, hard voting 3가지 방법에 대한 점수\n",
    "# Hard Voting 방법이 가장 높은 점수를 보임 \n",
    "\n",
    "global_score = f1_score(test_target, clf_final.predict(test_x1), average='weighted')\n",
    "vc.voting = 'hard'\n",
    "global_score_hard = f1_score(test_target, vc.predict(test_x1), average='weighted')\n",
    "vc.voting = 'soft'\n",
    "global_score_soft = f1_score(test_target, vc.predict(test_x1),average='weighted')\n",
    "print('Validation score of a single XGB Classifier: {:.4f}'.format(global_score))\n",
    "print('Validation score of a VotingClassifier on 3 XGB with hard voting strategy: {:.4f}'.format(global_score_hard))\n",
    "print('Validation score of a VotingClassifier on 3 XGB with soft voting strategy: {:.4f}'.format(global_score_soft))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    ">average 파라미터 설명 -> weigted 선택\n",
    ">\n",
    ">(f1_score(test_target, vc.predict(test_x1), average='weighted')\n",
    ">\n",
    ">macro 평균은 클래스별 f1 score에 가중치를 주지 않음 클래스 크기에 상관없이 모든 클래스를 같은 비중으로 다룸\n",
    ">\n",
    ">micro 평균은 모든 클래스의 FP,FN,TP의 총 수를 헤아린 다음 precision, recall, f1 score를 수리초 계산한다\n",
    ">\n",
    ">weighted 평균은 클래스별 샘플 수로 가중치를 두어 f1 score의 평균을 계산한다. 이 값이 분류 report에 나타나는 값이다(각각의 라벨에 동일한 가중치(target label을 갖는 인스턴스의 수)를 부여)\n",
    ">\n",
    ">samples 평균은 각 인스턴스에 대해 f1 score를 계산하는 함수고 다중 레이블 분류에 사용된다(여기선 이용x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "f1  0.9045246073567706\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       1.00      0.92      0.96        74\n",
      "           1       0.85      0.92      0.88        49\n",
      "           2       0.86      0.89      0.88        64\n",
      "           3       0.88      0.88      0.88        41\n",
      "\n",
      "    accuracy                           0.90       228\n",
      "   macro avg       0.90      0.90      0.90       228\n",
      "weighted avg       0.91      0.90      0.90       228\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 최종 모델 점수와 Confusion Matrix \n",
    "\n",
    "cnf_matrix = confusion_matrix(test_target, vc.predict(test_x1))\n",
    "g = sns.heatmap(pd.DataFrame(cnf_matrix), annot=True, cmap=\"BuGn\", fmt='g')\n",
    "buttom , top = g.get_ylim()\n",
    "g.set_ylim(buttom+0.5, top-0.5)\n",
    "plt.ylabel('Actual Label')\n",
    "plt.xlabel('Predicted Label')\n",
    "\n",
    "print('f1 ',f1_score(y_true=test_target, y_pred=vc.predict(test_x1), average='weighted'))\n",
    "print(classification_report(test_target, vc.predict(test_x1)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> f1_score 90.4%\n",
    ">\n",
    "> a를 a라 잘 예측 68개(소멸위험이 큰 지역 68개)\n",
    ">\n",
    "> d를 d라고 잘 예측 36개(소멸위험이 작은 지역 36개)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### e.앙상블 <a id=\"four-e\"></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    ">모델 결합 : 마지막으로, 여러 수준의 모델을 앙상블합니다. 최상의 결과를 얻으려면 모델끼리의 상관 관계 확인\n",
    "\n",
    ">앙상블 : 여러가지 예측모형들을 만든 후 조합하여 하나의 최종 예측모형을 만드는 방법 \n",
    ">\n",
    ">장점 : 성능이 많이 좋아짐. 오버피팅도 방지 됨. \n",
    ">\n",
    ">단점 : 해석력이 많이 떨어짐. 모델이 무거워짐. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_XGB = pd.Series(vc.predict(test_x1), name=\"XGB\")\n",
    "test_GBC = pd.Series(vc2.predict(test_x1), name=\"GBC\")\n",
    "test_RFC = pd.Series(vc3.predict(test_x1), name=\"RFC\")\n",
    "\n",
    "ensemble_results = pd.concat([test_XGB,test_GBC,test_RFC],axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "g= sns.heatmap(ensemble_results.corr(),annot=True,cmap='BuGn')\n",
    "buttom , top = g.get_ylim()\n",
    "g.set_ylim(buttom+0.5, top-0.5)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    ">모델끼리의 상관관계 확인결과 xgb와 gbc를 통해서 앙상블을 하기로 결정함"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "#xgb와 gbc에 각각의 가중치를 주는 함수 만듬\n",
    "def combine_voters(data, weights=[0.5, 0.5]):\n",
    "    # do soft voting with both classifiers\n",
    "    vc.voting=\"soft\"\n",
    "    vc1_probs = vc.predict_proba(data)\n",
    "    vc2.voting=\"soft\"\n",
    "    vc2_probs = vc2.predict_proba(data) \n",
    "    \n",
    "    final_vote = (vc1_probs * weights[0]) + (vc2_probs * weights[1])\n",
    "    predictions = np.argmax(final_vote, axis=1)\n",
    "    \n",
    "    return predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.913219839582179"
      ]
     },
     "execution_count": 75,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#xgb에는 0.8의 가중치를 gbc에는 0.2의 가중치를 줘서 앙상블\n",
    "combo_preds = combine_voters(test_x1, weights=[0.8, 0.2])\n",
    "global_combo_score_soft = f1_score(test_target, combo_preds, average='weighted')\n",
    "global_combo_score_soft"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> xgb와 rf, xgb와 gbc와 rf, gbc와 rf 모두 앙상블을 했을때보다 xgb와 gbc의 스코어가 높게 나왔다"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> 최종 앙상블 모델의 성능은 91.3%"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 결과 <a id=\"five\"></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1, 1, 1, 1, 1, 1, 0, 0, 2, 1, 0, 1, 2, 0, 0, 0, 1, 1, 1, 2, 2, 3,\n",
       "       3, 2, 3, 3, 2, 1, 3, 3, 3, 3, 3, 2, 3, 2, 1, 1, 1, 3, 3, 2, 2, 2,\n",
       "       2, 3, 1, 3, 3, 3, 0, 0, 3, 0, 1, 1, 0, 3, 0, 2, 0, 2, 1, 0, 1, 0,\n",
       "       0, 2, 1, 0, 3, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0,\n",
       "       2, 2, 3, 2, 1, 2, 2, 1, 2, 3, 2, 2, 1, 2, 1, 2, 2, 3, 3, 1, 3, 2,\n",
       "       2, 1, 1, 1, 2, 2, 2, 2, 1, 1, 2, 1, 1, 2, 3, 2, 2, 3, 3, 3, 2, 2,\n",
       "       2, 2, 2, 3, 3, 3, 3, 3, 2, 3, 2, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 2,\n",
       "       2, 0, 2, 2, 3, 1, 2, 3, 3, 0, 2, 0, 0, 0, 2, 0, 1, 0, 2, 1, 0, 2,\n",
       "       0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 1,\n",
       "       2, 0, 0, 2, 2, 1, 0, 1, 1, 1, 0, 1, 0, 3, 0, 3, 0, 0, 1, 0, 0, 0,\n",
       "       0, 0, 1, 1, 2, 2, 3, 1], dtype=int64)"
      ]
     },
     "execution_count": 76,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 등급 매핑 \n",
    "combo_preds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['b', 'b', 'b', 'b', 'b', 'b', 'a', 'a', 'c', 'b', 'a', 'b', 'c',\n",
       "       'a', 'a', 'a', 'b', 'b', 'b', 'c', 'c', 'd', 'd', 'c', 'd', 'd',\n",
       "       'c', 'b', 'd', 'd', 'd', 'd', 'd', 'c', 'd', 'c', 'b', 'b', 'b',\n",
       "       'd', 'd', 'c', 'c', 'c', 'c', 'd', 'b', 'd', 'd', 'd', 'a', 'a',\n",
       "       'd', 'a', 'b', 'b', 'a', 'd', 'a', 'c', 'a', 'c', 'b', 'a', 'b',\n",
       "       'a', 'a', 'c', 'b', 'a', 'd', 'a', 'b', 'a', 'a', 'a', 'a', 'b',\n",
       "       'a', 'a', 'a', 'b', 'b', 'a', 'a', 'a', 'a', 'a', 'c', 'c', 'd',\n",
       "       'c', 'b', 'c', 'c', 'b', 'c', 'd', 'c', 'c', 'b', 'c', 'b', 'c',\n",
       "       'c', 'd', 'd', 'b', 'd', 'c', 'c', 'b', 'b', 'b', 'c', 'c', 'c',\n",
       "       'c', 'b', 'b', 'c', 'b', 'b', 'c', 'd', 'c', 'c', 'd', 'd', 'd',\n",
       "       'c', 'c', 'c', 'c', 'c', 'd', 'd', 'd', 'd', 'd', 'c', 'd', 'c',\n",
       "       'c', 'c', 'c', 'c', 'c', 'c', 'd', 'd', 'd', 'd', 'c', 'c', 'a',\n",
       "       'c', 'c', 'd', 'b', 'c', 'd', 'd', 'a', 'c', 'a', 'a', 'a', 'c',\n",
       "       'a', 'b', 'a', 'c', 'b', 'a', 'c', 'a', 'b', 'a', 'a', 'a', 'a',\n",
       "       'a', 'a', 'a', 'a', 'a', 'a', 'b', 'a', 'a', 'a', 'a', 'a', 'b',\n",
       "       'b', 'a', 'b', 'c', 'a', 'a', 'c', 'c', 'b', 'a', 'b', 'b', 'b',\n",
       "       'a', 'b', 'a', 'd', 'a', 'd', 'a', 'a', 'b', 'a', 'a', 'a', 'a',\n",
       "       'a', 'b', 'b', 'c', 'c', 'd', 'b'], dtype=object)"
      ]
     },
     "execution_count": 77,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#예측치의 숫자를 다시 문자로 매핑\n",
    "pred = LB.inverse_transform(combo_preds)\n",
    "pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>sigun</th>\n",
       "      <th>소멸위험등급</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>강원도강릉시</td>\n",
       "      <td>b</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>강원도고성군</td>\n",
       "      <td>b</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>강원도동해시</td>\n",
       "      <td>b</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>강원도삼척시</td>\n",
       "      <td>b</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>강원도속초시</td>\n",
       "      <td>b</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    sigun 소멸위험등급\n",
       "0  강원도강릉시      b\n",
       "1  강원도고성군      b\n",
       "2  강원도동해시      b\n",
       "3  강원도삼척시      b\n",
       "4  강원도속초시      b"
      ]
     },
     "execution_count": 78,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 데이터프레임화\n",
    "result = pd.DataFrame({\n",
    "    \"sigun\" :test_x1.index,\n",
    "    \"소멸위험등급\":  pred\n",
    "})\n",
    "result.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>sigun</th>\n",
       "      <th>소멸위험등급</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>강원도양양군</td>\n",
       "      <td>a</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>강원도영월군</td>\n",
       "      <td>a</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>강원도정선군</td>\n",
       "      <td>a</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>강원도태백시</td>\n",
       "      <td>a</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>강원도평창군</td>\n",
       "      <td>a</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>217</th>\n",
       "      <td>충청북도괴산군</td>\n",
       "      <td>a</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>218</th>\n",
       "      <td>충청북도단양군</td>\n",
       "      <td>a</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>219</th>\n",
       "      <td>충청북도보은군</td>\n",
       "      <td>a</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>220</th>\n",
       "      <td>충청북도영동군</td>\n",
       "      <td>a</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>221</th>\n",
       "      <td>충청북도옥천군</td>\n",
       "      <td>a</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>67 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       sigun 소멸위험등급\n",
       "6     강원도양양군      a\n",
       "7     강원도영월군      a\n",
       "10    강원도정선군      a\n",
       "13    강원도태백시      a\n",
       "14    강원도평창군      a\n",
       "..       ...    ...\n",
       "217  충청북도괴산군      a\n",
       "218  충청북도단양군      a\n",
       "219  충청북도보은군      a\n",
       "220  충청북도영동군      a\n",
       "221  충청북도옥천군      a\n",
       "\n",
       "[67 rows x 2 columns]"
      ]
     },
     "execution_count": 79,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#소멸위험인 시군구\n",
    "result[result.소멸위험등급 == 'a']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [],
   "source": [
    "#정성적 분석을 할 때 이용하기 위해 csv형태로 저장\n",
    "result_a = result[result.소멸위험등급 == 'a']\n",
    "#result_a.to_csv('소멸위험시군구_최종.csv',encoding='cp949')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> 소멸위험지수가 큰 지방자치단체"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>sigun</th>\n",
       "      <th>소멸위험등급</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>경기도광명시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>경기도광주시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>경기도군포시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>경기도김포시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>경기도부천시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>경기도성남시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>경기도수원시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>경기도시흥시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>경기도안산시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>경기도안양시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>경기도오산시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>40</th>\n",
       "      <td>경기도용인시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45</th>\n",
       "      <td>경기도평택시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>47</th>\n",
       "      <td>경기도하남시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>48</th>\n",
       "      <td>경기도화성시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49</th>\n",
       "      <td>경상남도거제시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>52</th>\n",
       "      <td>경상남도김해시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>57</th>\n",
       "      <td>경상남도양산시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>70</th>\n",
       "      <td>경상북도구미시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>90</th>\n",
       "      <td>광주광역시광산구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>97</th>\n",
       "      <td>대구광역시달성군</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>105</th>\n",
       "      <td>대전광역시서구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>106</th>\n",
       "      <td>대전광역시유성구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>108</th>\n",
       "      <td>부산광역시강서구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>124</th>\n",
       "      <td>서울특별시강남구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>127</th>\n",
       "      <td>서울특별시강서구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>128</th>\n",
       "      <td>서울특별시관악구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>129</th>\n",
       "      <td>서울특별시광진구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>135</th>\n",
       "      <td>서울특별시동작구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>136</th>\n",
       "      <td>서울특별시마포구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>137</th>\n",
       "      <td>서울특별시서대문구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>138</th>\n",
       "      <td>서울특별시서초구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>139</th>\n",
       "      <td>서울특별시성동구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>141</th>\n",
       "      <td>서울특별시송파구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>149</th>\n",
       "      <td>세종특별자치시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>150</th>\n",
       "      <td>울산광역시남구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>151</th>\n",
       "      <td>울산광역시동구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>152</th>\n",
       "      <td>울산광역시북구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>158</th>\n",
       "      <td>인천광역시남동구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>161</th>\n",
       "      <td>인천광역시서구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>162</th>\n",
       "      <td>인천광역시연수구</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>211</th>\n",
       "      <td>충청남도아산시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>213</th>\n",
       "      <td>충청남도천안시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>226</th>\n",
       "      <td>충청북도청주시</td>\n",
       "      <td>d</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         sigun 소멸위험등급\n",
       "21      경기도광명시      d\n",
       "22      경기도광주시      d\n",
       "24      경기도군포시      d\n",
       "25      경기도김포시      d\n",
       "28      경기도부천시      d\n",
       "29      경기도성남시      d\n",
       "30      경기도수원시      d\n",
       "31      경기도시흥시      d\n",
       "32      경기도안산시      d\n",
       "34      경기도안양시      d\n",
       "39      경기도오산시      d\n",
       "40      경기도용인시      d\n",
       "45      경기도평택시      d\n",
       "47      경기도하남시      d\n",
       "48      경기도화성시      d\n",
       "49     경상남도거제시      d\n",
       "52     경상남도김해시      d\n",
       "57     경상남도양산시      d\n",
       "70     경상북도구미시      d\n",
       "90    광주광역시광산구      d\n",
       "97    대구광역시달성군      d\n",
       "105    대전광역시서구      d\n",
       "106   대전광역시유성구      d\n",
       "108   부산광역시강서구      d\n",
       "124   서울특별시강남구      d\n",
       "127   서울특별시강서구      d\n",
       "128   서울특별시관악구      d\n",
       "129   서울특별시광진구      d\n",
       "135   서울특별시동작구      d\n",
       "136   서울특별시마포구      d\n",
       "137  서울특별시서대문구      d\n",
       "138   서울특별시서초구      d\n",
       "139   서울특별시성동구      d\n",
       "141   서울특별시송파구      d\n",
       "149    세종특별자치시      d\n",
       "150    울산광역시남구      d\n",
       "151    울산광역시동구      d\n",
       "152    울산광역시북구      d\n",
       "158   인천광역시남동구      d\n",
       "161    인천광역시서구      d\n",
       "162   인천광역시연수구      d\n",
       "211    충청남도아산시      d\n",
       "213    충청남도천안시      d\n",
       "226    충청북도청주시      d"
      ]
     },
     "execution_count": 81,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result[result.소멸위험등급 == 'd']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> 소멸위험지수가 작은 지방자치단체"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}