4. Training Model.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "586e0e61",
   "metadata": {},
   "source": [
    "# Training Model"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6e146939",
   "metadata": {},
   "source": [
    "# Import Library"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "e86bcca6",
   "metadata": {},
   "outputs": [],
   "source": [
    "#import all realated libraries\n",
    "#import libraries for data analysis\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "# import library for visualization\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# import pickle and json file for columns and model file\n",
    "import pickle\n",
    "import json\n",
    "import joblib\n",
    "import yaml\n",
    "import scipy.stats as scs\n",
    "import src.util as util\n",
    "import yaml\n",
    "\n",
    "# import warnings for ignore the warnings\n",
    "import warnings \n",
    "warnings.filterwarnings(\"ignore\")\n",
    "\n",
    "# library for model selection and models\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "# evaluation metrics for classification model\n",
    "from sklearn.metrics import accuracy_score,classification_report,confusion_matrix\n",
    "from sklearn.metrics import classification_report, ConfusionMatrixDisplay\n",
    "from sklearn.metrics import accuracy_score, roc_auc_score\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "import json\n",
    "from datetime import datetime\n",
    "from sklearn.metrics import classification_report\n",
    "import uuid\n",
    "\n",
    "import lightgbm as lgb\n",
    "from sklearn.model_selection import cross_val_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "10ce5802-0e8a-4cfe-9eca-813f17b9faf7",
   "metadata": {},
   "outputs": [],
   "source": [
    "params_dir = \"config/config.yaml\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "98300325-6cca-467a-af8f-c9cffbcc6dec",
   "metadata": {},
   "outputs": [],
   "source": [
    "def load_params(param_dir):\n",
    "    with open(param_dir, 'r') as file:\n",
    "        params = yaml.safe_load(file)\n",
    "        \n",
    "    return params"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "098bc70c-038f-491d-baab-698f80640169",
   "metadata": {},
   "outputs": [],
   "source": [
    "params = load_params(params_dir)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "50c215ab-1591-4ab2-a5ed-c4e0e533f2d2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'raw_dataset_dir': 'dataset/1 - raw data/',\n",
       " 'train_set_path': ['dataset/2 - processed/X_train.pkl',\n",
       "  'dataset/2 - processed/y_train.pkl'],\n",
       " 'valid_set_path': ['dataset/2 - processed/X_valid.pkl',\n",
       "  'dataset/2 - processed/y_valid.pkl'],\n",
       " 'test_set_path': ['dataset/2 - processed/X_test.pkl',\n",
       "  'dataset/2 - processed/y_test.pkl'],\n",
       " 'train_feng_set_path': ['dataset/2 - processed/X_train_feng.pkl',\n",
       "  'dataset/2 - processed/y_train_feng.pkl'],\n",
       " 'valid_feng_set_path': ['dataset/2 - processed/X_valid_feng.pkl',\n",
       "  'dataset/2 - processed/y_valid_feng.pkl'],\n",
       " 'test_feng_set_path': ['dataset/2 - processed/X_test_feng.pkl',\n",
       "  'dataset/2 - processed/y_test_feng.pkl'],\n",
       " 'impute_data_train': ['dataset/2 - processed/X_train_impute.pkl',\n",
       "  'dataset/2 - processed/y_test_impute.pkl'],\n",
       " 'impute_data_test': ['dataset/2 - processed/X_test_impute.pkl',\n",
       "  'dataset/2 - processed/y_test_impute.pkl'],\n",
       " 'impute_data_valid': ['dataset/2 - processed/X_valid_impute.pkl',\n",
       "  'dataset/2 - processed/y_valid_impute.pkl'],\n",
       " 'train_ohe_balance': ['dataset/3 - final/1 - Encode - One Hot Encoding dan Balancing Data/X_train_ohebalance.pkl',\n",
       "  'dataset/3 - final/1 - Encode - One Hot Encoding dan Balancing Data/y_train_ohebalance.pkl'],\n",
       " 'valid_ohe_balance': ['dataset/3 - final/1 - Encode - One Hot Encoding dan Balancing Data/x_valid_feng.pkl',\n",
       "  'dataset/3 - final/1 - Encode - One Hot Encoding dan Balancing Data/y_valid_feng.pkl'],\n",
       " 'test_ohe_balance': ['dataset/3 - final/1 - Encode - One Hot Encoding dan Balancing Data/x_test_feng.pkl',\n",
       "  'dataset/3 - final/1 - Encode - One Hot Encoding dan Balancing Data/y_test_feng.pkl'],\n",
       " 'rus_ohe_balance': ['dataset/3 - final/1 - Encode - One Hot Encoding dan Balancing Data/x_rus.pkl',\n",
       "  'dataset/3 - final/1 - Encode - One Hot Encoding dan Balancing Data/y_rus.pkl'],\n",
       " 'sm_ohe_balance': ['dataset/3 - final/1 - Encode - One Hot Encoding dan Balancing Data/x_sm.pkl',\n",
       "  'dataset/3 - final/1 - Encode - One Hot Encoding dan Balancing Data/y_sm.pkl'],\n",
       " 'ros_ohe_balance': ['dataset/3 - final/1 - Encode - One Hot Encoding dan Balancing Data/x_ros.pkl',\n",
       "  'dataset/3 - final/1 - Encode - One Hot Encoding dan Balancing Data/y_ros.pkl'],\n",
       " 'standar_scaler_train': ['dataset/3 - final/2 - Scaling Data - Standart Scaler/X_train_clean.pkl',\n",
       "  'dataset/3 - final/2 - Scaling Data - Standart Scaler/y_train.pkl'],\n",
       " 'standar_scaler_valid': ['dataset/3 - final/2 - Scaling Data - Standart Scaler/X_valid_clean.pkl',\n",
       "  'dataset/3 - final/2 - Scaling Data - Standart Scaler/y_valid.pkl'],\n",
       " 'standar_scaler_test': ['dataset/3 - final/2 - Scaling Data - Standart Scaler/X_test_clean.pkl',\n",
       "  'dataset/3 - final/2 - Scaling Data - Standart Scaler/y_test.pkl'],\n",
       " 'standar_scaler_rus': ['dataset/3 - final/2 - Scaling Data - Standart Scaler/X_rus_clean.pkl',\n",
       "  'dataset/3 - final/2 - Scaling Data - Standart Scaler/y_rus.pkl'],\n",
       " 'standar_scaler_ros': ['dataset/3 - final/2 - Scaling Data - Standart Scaler/X_ros_clean.pkl',\n",
       "  'dataset/3 - final/2 - Scaling Data - Standart Scaler/y_ros.pkl'],\n",
       " 'standar_scaler_sm': ['dataset/3 - final/2 - Scaling Data - Standart Scaler/X_sm_clean.pkl',\n",
       "  'dataset/3 - final/2 - Scaling Data - Standart Scaler/y_sm.pkl'],\n",
       " 'raw_dataset_path': 'dataset/processed/Churn_Modelling.csv',\n",
       " 'cleaned_raw_dataset_path': 'dataset/2 - processed/data_bank.pkl',\n",
       " 'production_model_path': 'model/5 - Model Final',\n",
       " 'ohe_stasiun_path': 'model/5 - Model Final/ohe_stasiun.pkl',\n",
       " 'le_encoder_path': 'model/5 - Model Final/le_encoder.pkl',\n",
       " 'model_scaler': 'model/5 - Model Final/scaler.pkl',\n",
       " 'scaler': 'model/5 - Model Final/',\n",
       " 'model_final': 'model/5 - Model Final/xgboost_cv.pkl',\n",
       " 'print_debug': True,\n",
       " 'int64_columns': ['CreditScore',\n",
       "  'Age',\n",
       "  'Tenure',\n",
       "  'NumOfProducts',\n",
       "  'HasCrCard',\n",
       "  'IsActiveMember'],\n",
       " 'float64_columns': ['Balance', 'EstimatedSalary'],\n",
       " 'object_columns': ['Geography', 'Gender'],\n",
       " 'label': 'Exited',\n",
       " 'predictors': ['CreditScore',\n",
       "  'Geography',\n",
       "  'Gender',\n",
       "  'Age',\n",
       "  'Tenure',\n",
       "  'Balance',\n",
       "  'NumOfProducts',\n",
       "  'HasCrCard',\n",
       "  'IsActiveMember',\n",
       "  'EstimatedSalary'],\n",
       " 'range_Geography': ['France', 'Spain', 'Germany'],\n",
       " 'range_Gender': ['Female', 'Male'],\n",
       " 'range_CreditScore': [350, 850],\n",
       " 'range_Age': [18, 92],\n",
       " 'range_Tenure': [0, 10],\n",
       " 'range_Balance': [0, 250898.09],\n",
       " 'range_NumOfProducts': [1, 4],\n",
       " 'range_IsActiveMember': [0, 1],\n",
       " 'range_EstimatedSalary': [11.58, 199992.48],\n",
       " 'range_HasCrCard': [0, 1],\n",
       " 'Exited_categories': [0, 1]}"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "params"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "4863e3c1-63e0-4d0d-955a-75fb91090e7c",
   "metadata": {},
   "outputs": [],
   "source": [
    "config_data = util.load_config()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "254cb49d",
   "metadata": {},
   "source": [
    "## Read Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "1af710e2",
   "metadata": {},
   "outputs": [],
   "source": [
    "#Read data X_train dan y_sm hasil dari data preparation\n",
    "X_ros_clean = util.pickle_load(config_data[\"standar_scaler_ros\"][0])\n",
    "y_ros = util.pickle_load(config_data[\"standar_scaler_ros\"][1])\n",
    "\n",
    "#Read data X_valid dan y_valid hasil dari data preparation\n",
    "X_valid_clean = util.pickle_load(config_data[\"standar_scaler_valid\"][0])\n",
    "y_valid = util.pickle_load(config_data[\"standar_scaler_valid\"][1])\n",
    "\n",
    "#Read data X_test dan y_test hasil dari data preparation\n",
    "X_test_clean = util.pickle_load(config_data[\"standar_scaler_test\"][0])\n",
    "y_test = util.pickle_load(config_data[\"standar_scaler_test\"][1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "e1ca0819",
   "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>Age</th>\n",
       "      <th>Balance</th>\n",
       "      <th>CreditScore</th>\n",
       "      <th>EstimatedSalary</th>\n",
       "      <th>Gender_Female</th>\n",
       "      <th>Gender_Male</th>\n",
       "      <th>Geography_France</th>\n",
       "      <th>Geography_Germany</th>\n",
       "      <th>Geography_Spain</th>\n",
       "      <th>HasCrCard</th>\n",
       "      <th>IsActiveMember</th>\n",
       "      <th>NumOfProducts</th>\n",
       "      <th>Tenure</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.306988</td>\n",
       "      <td>-0.011981</td>\n",
       "      <td>0.003435</td>\n",
       "      <td>1.409191</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1.493860</td>\n",
       "      <td>-1.336548</td>\n",
       "      <td>0.463506</td>\n",
       "      <td>0.610455</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-0.748604</td>\n",
       "      <td>-1.336548</td>\n",
       "      <td>-0.630442</td>\n",
       "      <td>1.178883</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-0.935476</td>\n",
       "      <td>-0.060350</td>\n",
       "      <td>-2.143566</td>\n",
       "      <td>-0.961045</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.185756</td>\n",
       "      <td>0.767652</td>\n",
       "      <td>1.230292</td>\n",
       "      <td>-1.408500</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11131</th>\n",
       "      <td>0.933244</td>\n",
       "      <td>0.254658</td>\n",
       "      <td>-0.773575</td>\n",
       "      <td>1.608396</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11132</th>\n",
       "      <td>0.466064</td>\n",
       "      <td>0.860213</td>\n",
       "      <td>-2.378714</td>\n",
       "      <td>1.358060</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11133</th>\n",
       "      <td>0.372628</td>\n",
       "      <td>0.561275</td>\n",
       "      <td>1.608573</td>\n",
       "      <td>1.546180</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11134</th>\n",
       "      <td>0.466064</td>\n",
       "      <td>0.581646</td>\n",
       "      <td>0.831564</td>\n",
       "      <td>0.064106</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11135</th>\n",
       "      <td>0.466064</td>\n",
       "      <td>0.584652</td>\n",
       "      <td>1.608573</td>\n",
       "      <td>1.081395</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>11136 rows × 13 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            Age   Balance  CreditScore  EstimatedSalary  Gender_Female  \\\n",
       "0      1.306988 -0.011981     0.003435         1.409191              0   \n",
       "1      1.493860 -1.336548     0.463506         0.610455              1   \n",
       "2     -0.748604 -1.336548    -0.630442         1.178883              1   \n",
       "3     -0.935476 -0.060350    -2.143566        -0.961045              0   \n",
       "4      0.185756  0.767652     1.230292        -1.408500              1   \n",
       "...         ...       ...          ...              ...            ...   \n",
       "11131  0.933244  0.254658    -0.773575         1.608396              0   \n",
       "11132  0.466064  0.860213    -2.378714         1.358060              0   \n",
       "11133  0.372628  0.561275     1.608573         1.546180              1   \n",
       "11134  0.466064  0.581646     0.831564         0.064106              0   \n",
       "11135  0.466064  0.584652     1.608573         1.081395              0   \n",
       "\n",
       "       Gender_Male  Geography_France  Geography_Germany  Geography_Spain  \\\n",
       "0                1                 0                  0                1   \n",
       "1                0                 0                  0                1   \n",
       "2                0                 0                  0                1   \n",
       "3                1                 0                  1                0   \n",
       "4                0                 0                  1                0   \n",
       "...            ...               ...                ...              ...   \n",
       "11131            1                 0                  1                0   \n",
       "11132            1                 1                  0                0   \n",
       "11133            0                 0                  1                0   \n",
       "11134            1                 0                  1                0   \n",
       "11135            1                 1                  0                0   \n",
       "\n",
       "       HasCrCard  IsActiveMember  NumOfProducts  Tenure  \n",
       "0              0               1              1       1  \n",
       "1              1               1              2       9  \n",
       "2              1               1              2       7  \n",
       "3              1               0              1       8  \n",
       "4              0               0              2       2  \n",
       "...          ...             ...            ...     ...  \n",
       "11131          1               1              3       8  \n",
       "11132          0               0              3       9  \n",
       "11133          1               0              1       9  \n",
       "11134          1               0              1       5  \n",
       "11135          1               0              3       6  \n",
       "\n",
       "[11136 rows x 13 columns]"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_ros_clean"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "ee1f4945",
   "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>Age</th>\n",
       "      <th>Balance</th>\n",
       "      <th>CreditScore</th>\n",
       "      <th>EstimatedSalary</th>\n",
       "      <th>Gender_Female</th>\n",
       "      <th>Gender_Male</th>\n",
       "      <th>Geography_France</th>\n",
       "      <th>Geography_Germany</th>\n",
       "      <th>Geography_Spain</th>\n",
       "      <th>HasCrCard</th>\n",
       "      <th>IsActiveMember</th>\n",
       "      <th>NumOfProducts</th>\n",
       "      <th>Tenure</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>9970</th>\n",
       "      <td>0.092320</td>\n",
       "      <td>1.121917</td>\n",
       "      <td>-1.325661</td>\n",
       "      <td>0.330646</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>-0.187988</td>\n",
       "      <td>-1.336548</td>\n",
       "      <td>-0.579323</td>\n",
       "      <td>0.696635</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7394</th>\n",
       "      <td>-1.776400</td>\n",
       "      <td>-1.336548</td>\n",
       "      <td>-1.315437</td>\n",
       "      <td>1.166592</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3471</th>\n",
       "      <td>-1.309220</td>\n",
       "      <td>0.411349</td>\n",
       "      <td>-0.160146</td>\n",
       "      <td>-0.652763</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7844</th>\n",
       "      <td>-0.281424</td>\n",
       "      <td>-1.336548</td>\n",
       "      <td>-0.364623</td>\n",
       "      <td>-1.580995</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4085</th>\n",
       "      <td>-0.655168</td>\n",
       "      <td>0.844874</td>\n",
       "      <td>-1.305214</td>\n",
       "      <td>1.611777</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>681</th>\n",
       "      <td>1.587296</td>\n",
       "      <td>0.754705</td>\n",
       "      <td>-0.579323</td>\n",
       "      <td>0.640851</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3983</th>\n",
       "      <td>0.279192</td>\n",
       "      <td>-1.336548</td>\n",
       "      <td>-0.221489</td>\n",
       "      <td>1.263142</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6219</th>\n",
       "      <td>-0.374860</td>\n",
       "      <td>0.537692</td>\n",
       "      <td>-0.119251</td>\n",
       "      <td>-0.830090</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6072</th>\n",
       "      <td>0.372628</td>\n",
       "      <td>-1.336548</td>\n",
       "      <td>-0.201042</td>\n",
       "      <td>-0.060032</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1800 rows × 13 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           Age   Balance  CreditScore  EstimatedSalary  Gender_Female  \\\n",
       "9970  0.092320  1.121917    -1.325661         0.330646              0   \n",
       "30   -0.187988 -1.336548    -0.579323         0.696635              1   \n",
       "7394 -1.776400 -1.336548    -1.315437         1.166592              1   \n",
       "3471 -1.309220  0.411349    -0.160146        -0.652763              1   \n",
       "7844 -0.281424 -1.336548    -0.364623        -1.580995              1   \n",
       "...        ...       ...          ...              ...            ...   \n",
       "4085 -0.655168  0.844874    -1.305214         1.611777              0   \n",
       "681   1.587296  0.754705    -0.579323         0.640851              0   \n",
       "3983  0.279192 -1.336548    -0.221489         1.263142              1   \n",
       "6219 -0.374860  0.537692    -0.119251        -0.830090              0   \n",
       "6072  0.372628 -1.336548    -0.201042        -0.060032              0   \n",
       "\n",
       "      Gender_Male  Geography_France  Geography_Germany  Geography_Spain  \\\n",
       "9970            1                 1                  0                0   \n",
       "30              0                 0                  0                1   \n",
       "7394            0                 1                  0                0   \n",
       "3471            0                 1                  0                0   \n",
       "7844            0                 1                  0                0   \n",
       "...           ...               ...                ...              ...   \n",
       "4085            1                 1                  0                0   \n",
       "681             1                 0                  0                1   \n",
       "3983            0                 1                  0                0   \n",
       "6219            1                 0                  0                1   \n",
       "6072            1                 1                  0                0   \n",
       "\n",
       "      HasCrCard  IsActiveMember  NumOfProducts  Tenure  \n",
       "9970          1               0              2       7  \n",
       "30            1               0              3       3  \n",
       "7394          0               1              1       8  \n",
       "3471          1               1              1       3  \n",
       "7844          1               1              2       1  \n",
       "...         ...             ...            ...     ...  \n",
       "4085          1               1              1       4  \n",
       "681           0               1              1       5  \n",
       "3983          0               1              1       2  \n",
       "6219          1               0              1       1  \n",
       "6072          1               1              2       9  \n",
       "\n",
       "[1800 rows x 13 columns]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_valid_clean"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "348f2342-6cd0-40ce-bfce-934f4a4411a2",
   "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>Age</th>\n",
       "      <th>Balance</th>\n",
       "      <th>CreditScore</th>\n",
       "      <th>EstimatedSalary</th>\n",
       "      <th>Gender_Female</th>\n",
       "      <th>Gender_Male</th>\n",
       "      <th>Geography_France</th>\n",
       "      <th>Geography_Germany</th>\n",
       "      <th>Geography_Spain</th>\n",
       "      <th>HasCrCard</th>\n",
       "      <th>IsActiveMember</th>\n",
       "      <th>NumOfProducts</th>\n",
       "      <th>Tenure</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>9970</th>\n",
       "      <td>0.092320</td>\n",
       "      <td>1.121917</td>\n",
       "      <td>-1.325661</td>\n",
       "      <td>0.330646</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>-0.187988</td>\n",
       "      <td>-1.336548</td>\n",
       "      <td>-0.579323</td>\n",
       "      <td>0.696635</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7394</th>\n",
       "      <td>-1.776400</td>\n",
       "      <td>-1.336548</td>\n",
       "      <td>-1.315437</td>\n",
       "      <td>1.166592</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3471</th>\n",
       "      <td>-1.309220</td>\n",
       "      <td>0.411349</td>\n",
       "      <td>-0.160146</td>\n",
       "      <td>-0.652763</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7844</th>\n",
       "      <td>-0.281424</td>\n",
       "      <td>-1.336548</td>\n",
       "      <td>-0.364623</td>\n",
       "      <td>-1.580995</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4085</th>\n",
       "      <td>-0.655168</td>\n",
       "      <td>0.844874</td>\n",
       "      <td>-1.305214</td>\n",
       "      <td>1.611777</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>681</th>\n",
       "      <td>1.587296</td>\n",
       "      <td>0.754705</td>\n",
       "      <td>-0.579323</td>\n",
       "      <td>0.640851</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3983</th>\n",
       "      <td>0.279192</td>\n",
       "      <td>-1.336548</td>\n",
       "      <td>-0.221489</td>\n",
       "      <td>1.263142</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6219</th>\n",
       "      <td>-0.374860</td>\n",
       "      <td>0.537692</td>\n",
       "      <td>-0.119251</td>\n",
       "      <td>-0.830090</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6072</th>\n",
       "      <td>0.372628</td>\n",
       "      <td>-1.336548</td>\n",
       "      <td>-0.201042</td>\n",
       "      <td>-0.060032</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1800 rows × 13 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           Age   Balance  CreditScore  EstimatedSalary  Gender_Female  \\\n",
       "9970  0.092320  1.121917    -1.325661         0.330646              0   \n",
       "30   -0.187988 -1.336548    -0.579323         0.696635              1   \n",
       "7394 -1.776400 -1.336548    -1.315437         1.166592              1   \n",
       "3471 -1.309220  0.411349    -0.160146        -0.652763              1   \n",
       "7844 -0.281424 -1.336548    -0.364623        -1.580995              1   \n",
       "...        ...       ...          ...              ...            ...   \n",
       "4085 -0.655168  0.844874    -1.305214         1.611777              0   \n",
       "681   1.587296  0.754705    -0.579323         0.640851              0   \n",
       "3983  0.279192 -1.336548    -0.221489         1.263142              1   \n",
       "6219 -0.374860  0.537692    -0.119251        -0.830090              0   \n",
       "6072  0.372628 -1.336548    -0.201042        -0.060032              0   \n",
       "\n",
       "      Gender_Male  Geography_France  Geography_Germany  Geography_Spain  \\\n",
       "9970            1                 1                  0                0   \n",
       "30              0                 0                  0                1   \n",
       "7394            0                 1                  0                0   \n",
       "3471            0                 1                  0                0   \n",
       "7844            0                 1                  0                0   \n",
       "...           ...               ...                ...              ...   \n",
       "4085            1                 1                  0                0   \n",
       "681             1                 0                  0                1   \n",
       "3983            0                 1                  0                0   \n",
       "6219            1                 0                  0                1   \n",
       "6072            1                 1                  0                0   \n",
       "\n",
       "      HasCrCard  IsActiveMember  NumOfProducts  Tenure  \n",
       "9970          1               0              2       7  \n",
       "30            1               0              3       3  \n",
       "7394          0               1              1       8  \n",
       "3471          1               1              1       3  \n",
       "7844          1               1              2       1  \n",
       "...         ...             ...            ...     ...  \n",
       "4085          1               1              1       4  \n",
       "681           0               1              1       5  \n",
       "3983          0               1              1       2  \n",
       "6219          1               0              1       1  \n",
       "6072          1               1              2       9  \n",
       "\n",
       "[1800 rows x 13 columns]"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_valid_clean"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "4807ec07-e73b-4d01-b219-12b53c148645",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "9970    0\n",
       "30      1\n",
       "7394    0\n",
       "3471    0\n",
       "7844    0\n",
       "       ..\n",
       "4085    0\n",
       "681     0\n",
       "3983    1\n",
       "6219    0\n",
       "6072    0\n",
       "Name: Exited, Length: 1800, dtype: int64"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_valid"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "1223d2b9-74a7-41fc-9f07-b5f08ea04878",
   "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>Age</th>\n",
       "      <th>Balance</th>\n",
       "      <th>CreditScore</th>\n",
       "      <th>EstimatedSalary</th>\n",
       "      <th>Gender_Female</th>\n",
       "      <th>Gender_Male</th>\n",
       "      <th>Geography_France</th>\n",
       "      <th>Geography_Germany</th>\n",
       "      <th>Geography_Spain</th>\n",
       "      <th>HasCrCard</th>\n",
       "      <th>IsActiveMember</th>\n",
       "      <th>NumOfProducts</th>\n",
       "      <th>Tenure</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2755</th>\n",
       "      <td>-0.935476</td>\n",
       "      <td>-1.336548</td>\n",
       "      <td>0.535073</td>\n",
       "      <td>0.777888</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4997</th>\n",
       "      <td>0.185756</td>\n",
       "      <td>-1.336548</td>\n",
       "      <td>0.759997</td>\n",
       "      <td>-0.967584</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6555</th>\n",
       "      <td>-0.561732</td>\n",
       "      <td>-1.336548</td>\n",
       "      <td>-0.068132</td>\n",
       "      <td>-0.124471</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5552</th>\n",
       "      <td>-0.001116</td>\n",
       "      <td>-1.336548</td>\n",
       "      <td>-0.211266</td>\n",
       "      <td>0.562164</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>796</th>\n",
       "      <td>0.466064</td>\n",
       "      <td>0.265957</td>\n",
       "      <td>0.555521</td>\n",
       "      <td>0.160352</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7089</th>\n",
       "      <td>0.652936</td>\n",
       "      <td>-1.336548</td>\n",
       "      <td>-0.129475</td>\n",
       "      <td>0.628648</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1956</th>\n",
       "      <td>1.306988</td>\n",
       "      <td>0.597370</td>\n",
       "      <td>-0.845142</td>\n",
       "      <td>0.482430</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7791</th>\n",
       "      <td>-0.935476</td>\n",
       "      <td>1.147886</td>\n",
       "      <td>-0.650890</td>\n",
       "      <td>-0.017758</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8854</th>\n",
       "      <td>-0.187988</td>\n",
       "      <td>-1.336548</td>\n",
       "      <td>0.269254</td>\n",
       "      <td>-1.574651</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>918</th>\n",
       "      <td>-0.561732</td>\n",
       "      <td>-1.336548</td>\n",
       "      <td>1.393873</td>\n",
       "      <td>-0.326887</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1200 rows × 13 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           Age   Balance  CreditScore  EstimatedSalary  Gender_Female  \\\n",
       "2755 -0.935476 -1.336548     0.535073         0.777888              0   \n",
       "4997  0.185756 -1.336548     0.759997        -0.967584              0   \n",
       "6555 -0.561732 -1.336548    -0.068132        -0.124471              0   \n",
       "5552 -0.001116 -1.336548    -0.211266         0.562164              0   \n",
       "796   0.466064  0.265957     0.555521         0.160352              1   \n",
       "...        ...       ...          ...              ...            ...   \n",
       "7089  0.652936 -1.336548    -0.129475         0.628648              1   \n",
       "1956  1.306988  0.597370    -0.845142         0.482430              0   \n",
       "7791 -0.935476  1.147886    -0.650890        -0.017758              1   \n",
       "8854 -0.187988 -1.336548     0.269254        -1.574651              1   \n",
       "918  -0.561732 -1.336548     1.393873        -0.326887              0   \n",
       "\n",
       "      Gender_Male  Geography_France  Geography_Germany  Geography_Spain  \\\n",
       "2755            1                 0                  0                1   \n",
       "4997            1                 0                  0                1   \n",
       "6555            1                 1                  0                0   \n",
       "5552            1                 1                  0                0   \n",
       "796             0                 1                  0                0   \n",
       "...           ...               ...                ...              ...   \n",
       "7089            0                 0                  0                1   \n",
       "1956            1                 1                  0                0   \n",
       "7791            0                 0                  1                0   \n",
       "8854            0                 0                  0                1   \n",
       "918             1                 0                  0                1   \n",
       "\n",
       "      HasCrCard  IsActiveMember  NumOfProducts  Tenure  \n",
       "2755          0               1              2       7  \n",
       "4997          1               0              1       1  \n",
       "6555          1               0              2       5  \n",
       "5552          1               0              2       3  \n",
       "796           0               1              1       9  \n",
       "...         ...             ...            ...     ...  \n",
       "7089          1               1              2       2  \n",
       "1956          1               1              2       4  \n",
       "7791          1               0              1       6  \n",
       "8854          1               1              2       6  \n",
       "918           0               0              2       3  \n",
       "\n",
       "[1200 rows x 13 columns]"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test_clean"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "fe9cf523-de60-407e-8fbf-2b47d8f578c4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2755    0\n",
       "4997    1\n",
       "6555    0\n",
       "5552    0\n",
       "796     0\n",
       "       ..\n",
       "7089    0\n",
       "1956    1\n",
       "7791    0\n",
       "8854    0\n",
       "918     0\n",
       "Name: Exited, Length: 1200, dtype: int64"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_test"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5bb686f7",
   "metadata": {},
   "source": [
    "## LightGBM Classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "b75e2996-b544-4531-a716-c6640d939237",
   "metadata": {},
   "outputs": [],
   "source": [
    "def binary_classification_lgbm_tuned(x_train, y_train, x_valid, y_valid, x_test, y_test):\n",
    "    # set hyperparameters for tuning\n",
    "    # instantiate the classifier\n",
    "    best_lgbm_clf = lgb.LGBMClassifier(random_state=123)\n",
    "    \n",
    "    best_lgbm_clf.fit(X_ros_clean, y_ros)\n",
    "    \n",
    "    # evaluate on validation set\n",
    "    valid_pred = best_lgbm_clf.predict(x_valid)\n",
    "    report = classification_report(y_valid, valid_pred, output_dict=True)\n",
    "    valid_recall = report['weighted avg']['recall']\n",
    "    print('Validation recall:', valid_recall)\n",
    "    \n",
    "    # evaluate on test set\n",
    "    test_pred = best_lgbm_clf.predict(x_test)\n",
    "    report = classification_report(y_test, test_pred, output_dict=True)\n",
    "    test_recall = report['weighted avg']['recall']\n",
    "    print('Test recall:', test_recall)\n",
    "    \n",
    "    return best_lgbm_clf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "d1e51561-cdd1-4257-a7e7-7ab51b6b31f6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Validation recall: 0.8327777777777777\n",
      "Test recall: 0.8391666666666666\n"
     ]
    }
   ],
   "source": [
    "lgbm_best = binary_classification_lgbm_tuned(x_train=X_ros_clean, y_train=y_ros,\n",
    "                                              x_valid=X_valid_clean, y_valid=y_valid,\n",
    "                                              x_test=X_test_clean, y_test=y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "6170ee05-bd70-427d-b694-40f1dc8e0663",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cross-validation Scores: [0.88150808 0.90296496 0.90925427 0.91382406 0.90843806]\n",
      "Average CV Score: 0.9031978849600204\n"
     ]
    }
   ],
   "source": [
    "# Perform cross-validation\n",
    "cross_val_scores = cross_val_score(lgbm_best, X_ros_clean, y_ros, cv=5, scoring='recall')\n",
    "\n",
    "# Print the cross-validation scores\n",
    "print(\"Cross-validation Scores:\", cross_val_scores)\n",
    "print(\"Average CV Score:\", cross_val_scores.mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "f10ca301-87f4-425e-8a15-2d6fea28e158",
   "metadata": {},
   "outputs": [],
   "source": [
    "valid_pred = lgbm_best.predict(X_valid_clean)\n",
    "valid_acc = accuracy_score(y_valid, valid_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "3b5d1859-5d51-419e-9159-869c53f7e673",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8327777777777777"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "valid_acc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "f25ade06-4b4e-4d90-9b31-74ef6beb2662",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.92      0.86      0.89      1437\n",
      "           1       0.57      0.72      0.64       363\n",
      "\n",
      "    accuracy                           0.83      1800\n",
      "   macro avg       0.75      0.79      0.76      1800\n",
      "weighted avg       0.85      0.83      0.84      1800\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_valid, valid_pred, labels=[0,1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "07eeb120-1810-4fc3-9dff-69f1b14938a0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ConfusionMatrixDisplay.from_predictions(y_valid, valid_pred, display_labels = [False, True])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "8bc7880d-75e0-4864-a421-9134320b70c5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC: 0.8730299771294266\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import roc_curve, auc\n",
    "\n",
    "# predict probabilities for the positive class\n",
    "probs = lgbm_best.predict_proba(X_valid_clean)[:, 1]\n",
    "\n",
    "# calculate fpr and tpr for various thresholds\n",
    "fpr, tpr, thresholds = roc_curve(y_valid, probs)\n",
    "\n",
    "# calculate AUC\n",
    "roc_auc = auc(fpr, tpr)\n",
    "print('AUC:', roc_auc)\n",
    "\n",
    "# plot ROC curve\n",
    "plt.plot(fpr, tpr, label='ROC curve (area = %0.2f)' % roc_auc)\n",
    "plt.plot([0, 1], [0, 1], 'k--')\n",
    "plt.xlim([0.0, 1.0])\n",
    "plt.ylim([0.0, 1.05])\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.title('Receiver Operating Characteristic (ROC) Curve')\n",
    "plt.legend(loc=\"lower right\")\n",
    "#plt.savefig('roc_auc.png')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "f77c27ae-8eb3-468d-96d0-a291296d92d9",
   "metadata": {},
   "outputs": [],
   "source": [
    "test_pred = lgbm_best.predict(X_test_clean)\n",
    "test_acc = accuracy_score(y_test, test_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "a9f79367-063d-49ec-832f-8d787f9cde20",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8391666666666666"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_acc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "fd36c7a8-f605-44bb-b6d0-eb93a69ab9e5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.93      0.86      0.90       958\n",
      "           1       0.58      0.75      0.65       242\n",
      "\n",
      "    accuracy                           0.84      1200\n",
      "   macro avg       0.75      0.81      0.77      1200\n",
      "weighted avg       0.86      0.84      0.85      1200\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test, test_pred, labels=[0,1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "8c11f3b7-086d-48f9-9055-af2e87ac0138",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ConfusionMatrixDisplay.from_predictions(y_test, test_pred, display_labels = [False, True])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "57d638d3-b439-40fc-9570-70fe9a6bad47",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC: 0.8744284753015061\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import roc_curve, auc\n",
    "\n",
    "# predict probabilities for the positive class\n",
    "probs = lgbm_best.predict_proba(X_test_clean)[:, 1]\n",
    "\n",
    "# calculate fpr and tpr for various thresholds\n",
    "fpr, tpr, thresholds = roc_curve(y_test, probs)\n",
    "\n",
    "# calculate AUC\n",
    "roc_auc = auc(fpr, tpr)\n",
    "print('AUC:', roc_auc)\n",
    "\n",
    "# plot ROC curve\n",
    "plt.plot(fpr, tpr, label='ROC curve (area = %0.2f)' % roc_auc)\n",
    "plt.plot([0, 1], [0, 1], 'k--')\n",
    "plt.xlim([0.0, 1.0])\n",
    "plt.ylim([0.0, 1.05])\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.title('Receiver Operating Characteristic (ROC) Curve')\n",
    "plt.legend(loc=\"lower right\")\n",
    "#plt.savefig('roc_auc.png')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "175f8e39-ef6e-4f1d-a903-c3f7beddb9ed",
   "metadata": {},
   "source": [
    "## Save Log Best Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "dc340c90-ac9c-4b6b-85ad-7cb211445994",
   "metadata": {},
   "outputs": [],
   "source": [
    "def save_model_log(model, model_name, X_test, y_test):\n",
    "    # generate unique id\n",
    "    model_uid = uuid.uuid4().hex\n",
    "    \n",
    "    # get current time and date\n",
    "    now = datetime.now()\n",
    "    training_time = now.strftime(\"%H:%M:%S\")\n",
    "    training_date = now.strftime(\"%Y-%m-%d\")\n",
    "    \n",
    "    # generate classification report\n",
    "    y_pred = model.predict(X_test)\n",
    "    report = classification_report(y_test, y_pred, output_dict=True)\n",
    "    \n",
    "    # create dictionary for log\n",
    "    log = {\"model_name\": model_name,\n",
    "           \"model_uid\": model_uid,\n",
    "           \"training_time\": training_time,\n",
    "           \"training_date\": training_date,\n",
    "           \"classification_report\": report}\n",
    "    \n",
    "    # menyimpan log sebagai file JSON\n",
    "    with open('training_log/training_log.json', 'w') as f:\n",
    "        json.dump(log, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "4b21fb19-bce5-47e6-8155-dfb70080cafe",
   "metadata": {},
   "outputs": [],
   "source": [
    "save_model_log(model = lgbm_best, model_name = \"LightGBM Default\", X_test = X_test_clean, y_test=y_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5a4bdf55-0188-4112-813d-c980f6f34adc",
   "metadata": {},
   "source": [
    "## Evaluation Metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "876de2ec",
   "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>precision</th>\n",
       "      <th>recall</th>\n",
       "      <th>f1_score</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>macro_avg_precision</th>\n",
       "      <th>macro_avg_recall</th>\n",
       "      <th>macro_avg_f1_score</th>\n",
       "      <th>weighted_avg_precision</th>\n",
       "      <th>weighted_avg_recall</th>\n",
       "      <th>weighted_avg_f1_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>LGBM</th>\n",
       "      <td>0.86005</td>\n",
       "      <td>0.839167</td>\n",
       "      <td>0.84636</td>\n",
       "      <td>0.839167</td>\n",
       "      <td>0.754752</td>\n",
       "      <td>0.805073</td>\n",
       "      <td>0.773823</td>\n",
       "      <td>0.86005</td>\n",
       "      <td>0.839167</td>\n",
       "      <td>0.84636</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      precision    recall  f1_score  accuracy  macro_avg_precision  \\\n",
       "LGBM    0.86005  0.839167   0.84636  0.839167             0.754752   \n",
       "\n",
       "      macro_avg_recall  macro_avg_f1_score  weighted_avg_precision  \\\n",
       "LGBM          0.805073            0.773823                 0.86005   \n",
       "\n",
       "      weighted_avg_recall  weighted_avg_f1_score  \n",
       "LGBM             0.839167                0.84636  "
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "from sklearn.metrics import classification_report\n",
    "\n",
    "# definisikan model-model yang ingin dibandingkan\n",
    "models = {\n",
    "          'LGBM':lgbm_best\n",
    "         }\n",
    "\n",
    "# loop untuk menghitung dan menyimpan metrik untuk setiap model\n",
    "results = {}\n",
    "for name, model in models.items():\n",
    "    # hitung prediksi pada data test\n",
    "    y_pred = model.predict(X_test_clean)\n",
    "    # hitung metrik\n",
    "    report = classification_report(y_test, y_pred, output_dict=True)\n",
    "    # simpan metrik ke dictionary\n",
    "    results[name] = {\n",
    "        'precision': report['weighted avg']['precision'],\n",
    "        'recall': report['weighted avg']['recall'],\n",
    "        'f1_score': report['weighted avg']['f1-score'],\n",
    "        'accuracy': report['accuracy'],\n",
    "        'macro_avg_precision': report['macro avg']['precision'],\n",
    "        'macro_avg_recall': report['macro avg']['recall'],\n",
    "        'macro_avg_f1_score': report['macro avg']['f1-score'],\n",
    "        'weighted_avg_precision': report['weighted avg']['precision'],\n",
    "        'weighted_avg_recall': report['weighted avg']['recall'],\n",
    "        'weighted_avg_f1_score': report['weighted avg']['f1-score']\n",
    "    }\n",
    "\n",
    "# buat dataframe dari dictionary\n",
    "df_results = pd.DataFrame.from_dict(results, orient='index')\n",
    "\n",
    "# tampilkan dataframe sebagai tabel\n",
    "df_results"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "66c320ef",
   "metadata": {},
   "source": [
    "# Save Model"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c9575423",
   "metadata": {},
   "source": [
    "## LightGBM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "5b82825a-0b4e-49d9-ae4a-5c6e74971316",
   "metadata": {},
   "outputs": [],
   "source": [
    "lightGBM_default = 'model/5 - Model Final/lightGBM.pkl'\n",
    "with open(lightGBM_default, 'wb') as file:\n",
    "    pickle.dump(lgbm_best, file)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2033f72a",
   "metadata": {},
   "source": [
    "## Cara Load Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "eba972ca",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\"\\nfilename = 'models/model.pkl'\\nwith open(filename, 'rb') as file:\\n    model = pickle.load(file)\\n\""
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Load model dari folder models\n",
    "\"\"\"\n",
    "filename = 'models/model.pkl'\n",
    "with open(filename, 'rb') as file:\n",
    "    model = pickle.load(file)\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d4357f0c-a1dc-4fc4-afd1-ba5a0367f2fb",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.10.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}