Models_v7.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "--hxTFqqbbR_",
    "outputId": "6db3915f-54c5-4e6b-bcfd-241833e8c483"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
     ]
    }
   ],
   "source": [
    "from google.colab import drive\n",
    "drive.mount('/content/drive')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "aaHMICMyYe9h"
   },
   "source": [
    "### Install Libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "j0E-zFEfbbWZ",
    "outputId": "e468cf2d-98bd-47a8-ca86-fe54e377f779"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: catboost in /usr/local/lib/python3.7/dist-packages (0.24.4)\n",
      "Requirement already satisfied: pandas>=0.24.0 in /usr/local/lib/python3.7/dist-packages (from catboost) (1.1.5)\n",
      "Requirement already satisfied: graphviz in /usr/local/lib/python3.7/dist-packages (from catboost) (0.10.1)\n",
      "Requirement already satisfied: plotly in /usr/local/lib/python3.7/dist-packages (from catboost) (4.4.1)\n",
      "Requirement already satisfied: six in /usr/local/lib/python3.7/dist-packages (from catboost) (1.15.0)\n",
      "Requirement already satisfied: scipy in /usr/local/lib/python3.7/dist-packages (from catboost) (1.4.1)\n",
      "Requirement already satisfied: numpy>=1.16.0 in /usr/local/lib/python3.7/dist-packages (from catboost) (1.19.5)\n",
      "Requirement already satisfied: matplotlib in /usr/local/lib/python3.7/dist-packages (from catboost) (3.2.2)\n",
      "Requirement already satisfied: python-dateutil>=2.7.3 in /usr/local/lib/python3.7/dist-packages (from pandas>=0.24.0->catboost) (2.8.1)\n",
      "Requirement already satisfied: pytz>=2017.2 in /usr/local/lib/python3.7/dist-packages (from pandas>=0.24.0->catboost) (2018.9)\n",
      "Requirement already satisfied: retrying>=1.3.3 in /usr/local/lib/python3.7/dist-packages (from plotly->catboost) (1.3.3)\n",
      "Requirement already satisfied: kiwisolver>=1.0.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib->catboost) (1.3.1)\n",
      "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib->catboost) (2.4.7)\n",
      "Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.7/dist-packages (from matplotlib->catboost) (0.10.0)\n"
     ]
    }
   ],
   "source": [
    "!pip install catboost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "t7EsoB4euvsi",
    "outputId": "4216f8f5-1581-44d9-dd3f-76e85c23ec67"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: xgboost in /usr/local/lib/python3.7/dist-packages (0.90)\n",
      "Requirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from xgboost) (1.19.5)\n",
      "Requirement already satisfied: scipy in /usr/local/lib/python3.7/dist-packages (from xgboost) (1.4.1)\n"
     ]
    }
   ],
   "source": [
    "!pip install xgboost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "UotGg8J3bbXZ",
    "outputId": "e0b26194-d1e6-4a71-f69c-81068dc8d289"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/content/drive/MyDrive\n"
     ]
    }
   ],
   "source": [
    "%cd '/content/drive/MyDrive'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Cf5LRGVPYnfP"
   },
   "source": [
    "### Load Libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "K8QoaCA9bbZq"
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import seaborn as sns\n",
    "from sklearn.model_selection import train_test_split, GridSearchCV, RandomizedSearchCV\n",
    "from catboost import CatBoostClassifier, Pool\n",
    "import xgboost as xgb\n",
    "\n",
    "#F1-Score Expectation Maximization in O(n^2)\n",
    "#https://www.kaggle.com/mmueller/f1-score-expectation-maximization-in-o-n\n",
    "#https://github.com/KazukiOnodera/Instacart/blob/master/py_model/opt_fscore.py\n",
    "\n",
    "from f1optimization_faron import get_best_prediction\n",
    "\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.calibration import CalibratedClassifierCV\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.model_selection import GridSearchCV,RandomizedSearchCV\n",
    "from sklearn.metrics import confusion_matrix, accuracy_score, precision_score , log_loss\n",
    "\n",
    "from datetime import datetime\n",
    "from tqdm import tqdm\n",
    "import pickle\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "hI3mtUF0YJI9"
   },
   "source": [
    "### Load Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "5YMyqzVrbbbi"
   },
   "outputs": [],
   "source": [
    "#HDF5 format preserves data type after changing them\n",
    "data = pd.HDFStore(\"Data_v2.h5\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "U29Jnj4hbbeE"
   },
   "outputs": [],
   "source": [
    "#accessing dataset using keys\n",
    "train_data = data['train']\n",
    "test_data  = data['test']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "uDJ3PDaDYck-"
   },
   "source": [
    "#### Utility Functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "_6ybKg2xbbgE"
   },
   "outputs": [],
   "source": [
    "def display_target_distribution(y_train, y_val):\n",
    "\n",
    "  \"\"\"\n",
    "  Display target distribution in train and validation data\n",
    "  \"\"\"\n",
    "  _, train_label_counts = np.unique(y_train, return_counts=True)\n",
    "  _, val_label_counts = np.unique(y_val, return_counts=True)\n",
    "\n",
    "\n",
    "  width = 0.35  # the width of the bars\n",
    "  x = np.arange(2)\n",
    "  fig, ax = plt.subplots(figsize=(12,6))\n",
    "  rects1 = ax.bar(x - width/2, train_label_counts, width, label='Train')\n",
    "  rects2 = ax.bar(x + width/2, val_label_counts, width, label='validation')\n",
    "\n",
    "  # Add some text for labels, title and custom x-axis tick labels, etc.\n",
    "  ax.set_ylabel('Number of datapoints')\n",
    "  ax.set_title('Distribution of the Output Variable')\n",
    "  ax.set_xticks(x)\n",
    "  ax.set_xticklabels([0,1])\n",
    "  ax.legend()\n",
    "  plt.show()\n",
    "\n",
    "def random_split(test_size = 0.1):\n",
    "\n",
    "  \"\"\"\n",
    "  Split data randomly into train and validation data \n",
    "  \"\"\"\n",
    "  train_y = train_data['reordered'].values.tolist()\n",
    "  train_x = train_data.drop(['user_id', 'product_id', 'order_id', 'reordered'], axis = 1)\n",
    "\n",
    "  # since there are Millions of data ,we are taking 10% of data in test set , \n",
    "  X_train, X_val, y_train, y_val = train_test_split(train_x, train_y, stratify=train_y, test_size=0.1, random_state = 42)\n",
    "\n",
    "  #display distribution of data\n",
    "  display_target_distribution(y_train, y_val)\n",
    "  return (X_train, y_train),(X_val, y_val)\n",
    "\n",
    "def split_by_user(test_size = 0.1):\n",
    "  \n",
    "  #split on users\n",
    "  sampled_users = train_data['user_id'].sample(n=int(0.1 * train_data['user_id'].nunique()), random_state=42,replace=False).reset_index()\n",
    "  val = train_data[train_data['user_id'].isin(sampled_users['user_id'].tolist())]\n",
    "  train = train_data[~train_data['user_id'].isin(sampled_users['user_id'].tolist())]\n",
    "\n",
    "  y_train = train['reordered'].values.tolist()\n",
    "  X_train = train.drop(['user_id', 'product_id', 'order_id', 'reordered'], axis = 1)\n",
    "\n",
    "  y_val = val['reordered'].values.tolist()\n",
    "  X_val = val.drop(['user_id', 'product_id', 'order_id', 'reordered'], axis = 1)\n",
    "\n",
    "  display_target_distribution(y_train, y_val)\n",
    "  return (X_train, y_train),(X_val, y_val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "78nyy2QrxyFl"
   },
   "outputs": [],
   "source": [
    "def model_stats(predict_y,y_val):# Plot confusion Matrix\n",
    "\n",
    "  \"\"\"\n",
    "  Display model statistics such as accuracy , precision score and confusion matrix\n",
    "  \"\"\"\n",
    "\n",
    "  #predicted_y =np.array(predict_y>0.5,dtype=int).reshape(-1,).tolist()\n",
    "  cm = confusion_matrix(y_val, predict_y)\n",
    "  sns.heatmap(cm, cmap='Blues', annot = True, fmt=\"g\", annot_kws={\"size\": 12})\n",
    "  plt.show()\n",
    "  print(\"Accuracy :\",accuracy_score(y_val, predict_y))\n",
    "  print(\"Precision :\",precision_score(y_val, predict_y))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "CoQmM6oiwLfP"
   },
   "source": [
    "#### Function to create submission file based on 3 different global threshold"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "fI6KPmOprLBP"
   },
   "outputs": [],
   "source": [
    "def convert_to_string(row):\n",
    "    \"\"\"\n",
    "    convert series value to ist of strings\n",
    "    \"\"\"\n",
    "    list_string = list(map(str, row.tolist())) # convert series > list > list of strings\n",
    "    return \" \".join(list_string)\n",
    "\n",
    "def globl_threshold_products(model = None, scaler = None, model_type = 'xgb', val_file=\"Data_v2.h5\"):\n",
    "\n",
    "    \"\"\"\n",
    "    Get all products above threshold and join them\n",
    "    \"\"\"\n",
    "\n",
    "    #loading test data\n",
    "    test_data = pd.HDFStore(val_file)\n",
    "    test_data = test_data['test']\n",
    "    data = test_data.drop(['user_id', 'product_id', 'order_id'], axis = 1)\n",
    "\n",
    "    #scale test data\n",
    "    if scaler != None:\n",
    "        data = scaler.transform(data)\n",
    "\n",
    "    #if model is xgb use model.predict(), else use model.predict_proba()\n",
    "    if model_type == 'xgb':\n",
    "      dtest = xgb.DMatrix(data)\n",
    "      ypred = model.predict(dtest)\n",
    "    \n",
    "    elif model_type == 'mlp':\n",
    "      ypred = model.predict(data)\n",
    "      ypred = ypred.reshape(-1,).tolist()\n",
    "\n",
    "    else:\n",
    "      ypred = model.predict_proba(data)\n",
    "      #get probabilities of class 1\n",
    "      ypred = ypred[:,-1]\n",
    "\n",
    "    #add output probablities to a column\n",
    "    test_data['reordered'] =  ypred\n",
    "\n",
    "    #create a dataframe wiith unique order id\n",
    "    df = pd.DataFrame({'order_id':test_data['order_id'].unique()})\n",
    "\n",
    "    # 3 best threshold values based on multiple runs\n",
    "    for i, thresh in enumerate([0.18,0.19,0.20]):\n",
    "\n",
    "      #get all products above threshold and join them \n",
    "      test_data_copy = test_data[test_data['reordered']>=thresh].groupby(['order_id'])['product_id'].apply(convert_to_string).reset_index(name = \"products\")\n",
    "      \n",
    "      #create a submission by merging the order_id with test_data_copy\n",
    "      submission = pd.merge(test_data_copy,df, on = 'order_id', how = 'right')\n",
    "      submission['order_id'] = submission['order_id'].apply(int)\n",
    "\n",
    "      # replace empty cells with None , where no product is above threshold probability\n",
    "      submission = submission.where(pd.notnull(submission), None)\n",
    "\n",
    "      #save submission file\n",
    "      submission.to_csv(\"submission_sample_v8_{}_{}.csv\".format(i+1,model_type), index= False)\n",
    "      print(\"Generated submission_sample_v8_{}_{}.csv successfully\".format(i+1,model_type))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "BTZw8pI-rLBP"
   },
   "source": [
    "##### Function to return product list which maximizes F1-score ( uses faron's kernel)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "GqBQjgNQrLBP"
   },
   "outputs": [],
   "source": [
    "def getscores_on_testdata(model = None ,scaler = None, model_type = 'xgb', val_file = \"Data_v2.h5\"):\n",
    "    \"\"\"\n",
    "    Returns product list for every order-ID which maximizes F1-score\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    model      : The default is None.\n",
    "                 Trained model is passed here  \n",
    "    model_type : The default is xgb\n",
    "                 Identifier to deal with both sgb and catboost models\n",
    "    val_file   : The default is Data_v2.h5\n",
    "                 Path of dataset is passed\n",
    "    Returns\n",
    "    -------\n",
    "    test_products : Dataframe object\n",
    "                  This dataframe contains orderID, productID's(string)\n",
    "\n",
    "    \"\"\"\n",
    "\n",
    "    #loading test data\n",
    "    test_data = pd.HDFStore(val_file)\n",
    "    test_data = test_data['test']\n",
    "    data = test_data.drop(['user_id', 'product_id', 'order_id'], axis = 1)\n",
    "\n",
    "    #scale test data\n",
    "    if scaler != None:\n",
    "        data = scaler.transform(data)\n",
    "\n",
    "    #if model is xgb use model.predict(), else use model.predict_proba()\n",
    "    if model_type == 'xgb':\n",
    "      dtest = xgb.DMatrix(data)\n",
    "      ypred = model.predict(dtest)\n",
    "    \n",
    "    elif model_type == 'mlp':\n",
    "      ypred = model.predict(data)\n",
    "      ypred = ypred.reshape(-1,).tolist()\n",
    "    \n",
    "    else:\n",
    "      ypred = model.predict_proba(data)\n",
    "      #get probabilities of class 1\n",
    "      ypred = ypred[:,-1]\n",
    "\n",
    "    #add output probablities to a column\n",
    "    test_data['reordered'] =  ypred\n",
    "    \n",
    "    #create a dataframe with orderID and list of all possible productID's\n",
    "    test_products = test_data.groupby(['order_id'])['product_id'].apply(list).reset_index()\n",
    "    \n",
    "    #collect all probs in a single list for each OrderID\n",
    "    probs = test_data.groupby(['order_id'])['reordered'].apply(list).reset_index()\n",
    "    test_products['probs'] = probs['reordered']\n",
    "\n",
    "    #run faron's optimization code to get most probable set of products which might be reordered\n",
    "    products = []\n",
    "    for i in tqdm(test_products.index):\n",
    "        prods = get_best_prediction(test_products.iloc[i]['product_id'], test_products.iloc[i]['probs'], None)\n",
    "        products.append(prods)\n",
    "\n",
    "    test_products['products']= products\n",
    "    \n",
    "    test_products['order_id'] = test_products['order_id'].apply(int)\n",
    "    test_products.drop(['product_id','probs'], axis = 1, inplace = True)\n",
    "   \n",
    "    return test_products\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "WmVZBj--hL7l"
   },
   "source": [
    "## Models"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "GoDzEiFRuDmd"
   },
   "source": [
    "#### Logistic Regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 390
    },
    "id": "2gdwOqcmrfaW",
    "outputId": "54772593-a3ac-4472-fe71-32a12b11fc03"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#data split randomly\n",
    "\n",
    "(X_train, y_train),(X_val, y_val) = random_split(test_size = 0.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "6MV8Re2GuDDG"
   },
   "outputs": [],
   "source": [
    "#feature scaling for logistic regression\n",
    "\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "# fit scaler on training data\n",
    "scaler = StandardScaler().fit(X_train)\n",
    "\n",
    "# transforming the data\n",
    "X_train_norm = scaler.transform(X_train)\n",
    "X_val_norm = scaler.transform(X_val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "JEzIGe2Uchsr",
    "outputId": "f5daa09a-423d-4393-8a31-e1b97da3155b"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.7/dist-packages/sklearn/externals/joblib/__init__.py:15: FutureWarning: sklearn.externals.joblib is deprecated in 0.21 and will be removed in 0.23. Please import this functionality directly from joblib, which can be installed with: pip install joblib. If this warning is raised when loading pickled models, you may need to re-serialize those models with scikit-learn 0.21+.\n",
      "  warnings.warn(msg, category=FutureWarning)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "['standrdscaler.pkl']"
      ]
     },
     "execution_count": 14,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#saving standard scalar\n",
    "from sklearn.externals import joblib \n",
    "joblib.dump(scaler, 'standrdscaler.pkl') "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 350
    },
    "id": "rvmbBYQcuDAy",
    "outputId": "84415896-f901-4d28-85ec-b1f664958a63"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "logloss on validation data : 0.2550918280106341\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy : 0.9059975196674325\n",
      "Precision : 0.6415127528583993\n",
      "CPU times: user 6.98 s, sys: 1.11 s, total: 8.09 s\n",
      "Wall time: 2min 19s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "log_reg = LogisticRegression(random_state=0, n_jobs = -1)\n",
    "log_reg.fit(X_train_norm,y_train)\n",
    "\n",
    "predict_y = log_reg.predict_proba(X_val_norm)\n",
    "print(\"logloss on validation data :\",log_loss(y_val, predict_y, labels=[0,1], eps=1e-15))\n",
    "model_stats(np.argmax(predict_y,1),y_val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "-aplLd5fuC73",
    "outputId": "f862b082-3ed1-456d-92c3-fd1e342ee8a9"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generated submission_sample_v8_1_log_reg.csv successfully\n",
      "Generated submission_sample_v8_2_log_reg.csv successfully\n",
      "Generated submission_sample_v8_3_log_reg.csv successfully\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 75000/75000 [41:42<00:00, 29.97it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 41min 58s, sys: 10.9 s, total: 42min 9s\n",
      "Wall time: 42min 2s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "#get submission files on global thresholds\n",
    "globl_threshold_products(model = log_reg, scaler = scaler, model_type = 'log_reg', val_file=\"Data_v2.h5\")\n",
    "\n",
    "#get products for order ID's and create a submission file based on local threshold\n",
    "test_products = getscores_on_testdata(model = log_reg, scaler = scaler, model_type = 'log_reg', val_file=\"Data_v2.h5\")\n",
    "test_products.to_csv(\"submission_log_reg_v8.csv\", index= False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "nZOsn_MobajM"
   },
   "source": [
    "### Decision Tree"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "UIcRgToPbaSL",
    "outputId": "bb9d0c1a-0a73-41cc-902f-33d3c165c918"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 10 candidates, totalling 50 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 4 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done  42 tasks      | elapsed: 39.9min\n",
      "[Parallel(n_jobs=-1)]: Done  50 out of  50 | elapsed: 48.7min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 18min 11s, sys: 1.44 s, total: 18min 13s\n",
      "Wall time: 51min 3s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "param_grid = {}\n",
    "param_grid['max_depth'] = [5,10,15,20]\n",
    "param_grid['min_samples_split'] = [2,3,4,5]\n",
    "\n",
    "dt_clf = DecisionTreeClassifier()\n",
    "r_search = RandomizedSearchCV(dt_clf, param_distributions=param_grid, cv = 5, verbose = True, n_jobs = -1)\n",
    "r_search.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "QIrXtOPbcvwC",
    "outputId": "68481ead-128c-4648-aebb-88b6299b6f83"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'max_depth': 10, 'min_samples_split': 5}"
      ]
     },
     "execution_count": 30,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "r_search.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 316
    },
    "id": "HQHapWXMbaPc",
    "outputId": "5ccb4432-c8fa-4f1a-b936-f1045ae7bd56"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "logloss on validation data : 0.2509911734828939\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy : 0.9077474403133101\n",
      "Precision : 0.6117317107952654\n"
     ]
    }
   ],
   "source": [
    "predict_y = r_search.predict_proba(X_val)\n",
    "print(\"logloss on validation data :\",log_loss(y_val, predict_y, labels=[0,1], eps=1e-15))\n",
    "model_stats(np.argmax(predict_y,1),y_val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "1tgojWAlpFGE"
   },
   "outputs": [],
   "source": [
    "%%time\n",
    "\n",
    "#get submission files on global thresholds\n",
    "globl_threshold_products(model = r_search, scaler = scaler, model_type = 'dt', val_file=\"Data_v2.h5\")\n",
    "\n",
    "#get products for order ID's and create a submission file based on local threshold\n",
    "test_products = getscores_on_testdata(model = r_search, scaler = scaler, model_type = 'dt', val_file=\"Data_v2.h5\")\n",
    "test_products.to_csv(\"submission_dt_v8.csv\", index= False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "cS0TjjNnbkSY"
   },
   "source": [
    "### Random Forest"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "TmM8t0gHcT3h"
   },
   "source": [
    "##### When to calibrate your model ?\n",
    "ref: https://machinelearningmastery.com/calibrated-classification-model-in-scikit-learn/\n",
    "\n",
    "![image.png](data:image/png;base64,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)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 350
    },
    "id": "Tk-TR2I6_u_m",
    "outputId": "84da7898-aa5b-4c32-ed11-ca0fe50828c9"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "logloss on validation data : 0.25187675305313206\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy : 0.9103976910015376\n",
      "Precision : 0.6540423078628764\n",
      "CPU times: user 2h 9min 57s, sys: 42.7 s, total: 2h 10min 40s\n",
      "Wall time: 34min 57s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "#coderef: microsoft malware assignment\n",
    "\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "clf=RandomForestClassifier(n_estimators=25,random_state=42,n_jobs=-1)\n",
    "clf.fit(X_train, y_train)\n",
    "sig_clf = CalibratedClassifierCV(clf, method=\"sigmoid\")\n",
    "sig_clf.fit(X_train, y_train)\n",
    "\n",
    "predict_y = sig_clf.predict_proba(X_val)\n",
    "print(\"logloss on validation data :\",log_loss(y_val, predict_y, labels=[0,1], eps=1e-15))\n",
    "model_stats(np.argmax(predict_y,1),y_val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "BKL4IKdKZMOJ",
    "outputId": "0ccc5ad3-c675-44f7-bb77-2062fd87df93"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generated submission_sample_v8_1_rf.csv successfully\n",
      "Generated submission_sample_v8_2_rf.csv successfully\n",
      "Generated submission_sample_v8_3_rf.csv successfully\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 75000/75000 [41:58<00:00, 29.78it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 54min 22s, sys: 10.1 s, total: 54min 33s\n",
      "Wall time: 45min 36s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "#get submission files on global thresholds\n",
    "globl_threshold_products(model = sig_clf, scaler = None, model_type = 'rf', val_file=\"Data_v2.h5\")\n",
    "\n",
    "#get products for order ID's and create a submission file based on local threshold\n",
    "test_products = getscores_on_testdata(model = sig_clf, scaler = None, model_type = 'rf', val_file=\"Data_v2.h5\")\n",
    "test_products.to_csv(\"submission_rf_v8.csv\", index= False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "xCst2kKVZMGF"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "mjZE1yA9FbkU"
   },
   "source": [
    "#### Multi Layer Perceptron Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "nlxuFRoj6c3J"
   },
   "outputs": [],
   "source": [
    "## import tensorflow packages\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras.models import Model\n",
    "from tensorflow.keras.optimizers import Adam\n",
    "from tensorflow.keras.layers import Dense,Input,Activation,Dropout\n",
    "from tensorflow.keras.callbacks import LearningRateScheduler, EarlyStopping, ModelCheckpoint\n",
    "\n",
    "#SEED\n",
    "SEED_VALUE = 0\n",
    "np.random.seed(SEED_VALUE)\n",
    "tf.random.set_seed(SEED_VALUE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "bnL1mHKy6c0d"
   },
   "outputs": [],
   "source": [
    "def mlp_model():\n",
    "    \"\"\"\n",
    "    Create mlp model with 4 hidden layers .\n",
    "    \"\"\"\n",
    "    inp = Input(shape =  (28,)) #no of features\n",
    "    mlp = Dense(256, activation = \"relu\", kernel_initializer=tf.keras.initializers.he_normal(seed=SEED_VALUE), name = 'fc_1')(inp)\n",
    "    mlp = Dropout(0.3)(mlp)\n",
    "    mlp = Dense(128, activation = 'relu', kernel_initializer=tf.keras.initializers.he_normal(seed=SEED_VALUE), name = 'fc_2')(mlp)\n",
    "    mlp = Dropout(0.3)(mlp)\n",
    "    mlp = Dense(64, activation = 'relu', kernel_initializer=tf.keras.initializers.he_normal(seed=SEED_VALUE), name = 'fc_3')(mlp)\n",
    "    mlp = Dropout(0.3)(mlp)\n",
    "    mlp = Dense(32, activation = 'relu', kernel_initializer=tf.keras.initializers.he_normal(seed=SEED_VALUE), name = 'fc_4')(mlp)\n",
    "    mlp = Dropout(0.3)(mlp)\n",
    "    mlp = Dense(1, activation = 'sigmoid', kernel_initializer=tf.keras.initializers.he_normal(seed=SEED_VALUE), name = 'fc_5')(mlp)\n",
    "\n",
    "    model = Model(inputs = inp, outputs = mlp)\n",
    "    \n",
    "    opt = Adam(lr= 0.001)\n",
    "    model.compile(loss=\"binary_crossentropy\", optimizer=opt, metrics=[\"accuracy\"])\n",
    "    \n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "xb22a9Eb6cxb"
   },
   "outputs": [],
   "source": [
    "import os\n",
    "\n",
    "#save model checkpoint\n",
    "filepath=\"saved_model/mlp/checkpoint.hdf5\"\n",
    "\n",
    "if not os.path.exists(\"saved_model/mlp/\"):\n",
    "    os.makedirs(\"saved_model/mlp/\")    \n",
    "\n",
    "checkpoint = ModelCheckpoint(filepath=filepath, monitor='val_accuracy',  verbose=1, save_best_only=True, mode='auto')\n",
    "earlystop = EarlyStopping(monitor='val_accuracy', min_delta=0.001, patience=2, verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "dzrreidU6cvD",
    "outputId": "5409b283-46e1-4c8a-d3cb-b5fb77ce4f46"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         [(None, 28)]              0         \n",
      "_________________________________________________________________\n",
      "fc_1 (Dense)                 (None, 256)               7424      \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "fc_2 (Dense)                 (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "fc_3 (Dense)                 (None, 64)                8256      \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "fc_4 (Dense)                 (None, 32)                2080      \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 32)                0         \n",
      "_________________________________________________________________\n",
      "fc_5 (Dense)                 (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 50,689\n",
      "Trainable params: 50,689\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "tf.keras.backend.clear_session()\n",
    "\n",
    "model = mlp_model()\n",
    "print(model.summary())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "C-yVdgFo9IJE",
    "outputId": "07400c9f-730a-428f-f194-3cd42c86c95d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "238350/238350 [==============================] - 515s 2ms/step - loss: 0.2631 - accuracy: 0.9042 - val_loss: 0.2548 - val_accuracy: 0.9044\n",
      "\n",
      "Epoch 00001: val_accuracy improved from -inf to 0.90443, saving model to saved_model/mlp/checkpoint.hdf5\n",
      "Epoch 2/10\n",
      "238350/238350 [==============================] - 504s 2ms/step - loss: 0.2582 - accuracy: 0.9053 - val_loss: 0.2558 - val_accuracy: 0.9072\n",
      "\n",
      "Epoch 00002: val_accuracy improved from 0.90443 to 0.90719, saving model to saved_model/mlp/checkpoint.hdf5\n",
      "Epoch 3/10\n",
      "238350/238350 [==============================] - 505s 2ms/step - loss: 0.2597 - accuracy: 0.9049 - val_loss: 0.2513 - val_accuracy: 0.9074\n",
      "\n",
      "Epoch 00003: val_accuracy improved from 0.90719 to 0.90741, saving model to saved_model/mlp/checkpoint.hdf5\n",
      "Epoch 4/10\n",
      "238350/238350 [==============================] - 502s 2ms/step - loss: 0.2605 - accuracy: 0.9046 - val_loss: 0.2543 - val_accuracy: 0.9046\n",
      "\n",
      "Epoch 00004: val_accuracy did not improve from 0.90741\n",
      "Epoch 00004: early stopping\n"
     ]
    }
   ],
   "source": [
    "#train model\n",
    "batch_size = 128\n",
    "history = model.fit(X_train_norm,np.array(y_train).reshape(-1,1), epochs = 10, validation_data = (X_val_norm,np.array(y_val).reshape(-1,1)), callbacks=[checkpoint, earlystop])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "tKSmkB_Z9IGN",
    "outputId": "0b92ea78-d9eb-4dea-8600-035b60acf210"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training Accuracy :90.75 %\n",
      "Validation Accuracy :90.74 %\n"
     ]
    }
   ],
   "source": [
    "#load the saved model and print accuracy\n",
    "\n",
    "saved_model = tf.keras.models.load_model('saved_model/mlp/checkpoint.hdf5')\n",
    "print(\"Training Accuracy :{:.2f} %\".format(saved_model.evaluate(X_train_norm,np.array(y_train).reshape(-1,1), verbose = 0)[1]*100))\n",
    "print(\"Validation Accuracy :{:.2f} %\".format(saved_model.evaluate(X_val_norm,np.array(y_val).reshape(-1,1), verbose = 0)[1]*100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 316
    },
    "id": "eXlznw57CE6L",
    "outputId": "ef4dad8c-be62-4f1e-e823-ca20b9f49dca"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "logloss on validation data : 0.2513314122715033\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy : 0.9074087840588483\n",
      "Precision : 0.6539266285395453\n"
     ]
    }
   ],
   "source": [
    "#print model statistics\n",
    "\n",
    "predict_y = saved_model.predict(X_val_norm)\n",
    "print(\"logloss on validation data :\",log_loss(y_val, predict_y.reshape(-1,).tolist(), labels=[0,1], eps=1e-15))\n",
    "model_stats(predict_y,y_val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "VNBltAW79IDV",
    "outputId": "41fcfe37-88d7-439f-e550-da28bbc2909d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generated submission_sample_v8_1_mlp.csv successfully\n",
      "Generated submission_sample_v8_2_mlp.csv successfully\n",
      "Generated submission_sample_v8_3_mlp.csv successfully\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 75000/75000 [42:05<00:00, 29.70it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 46min 50s, sys: 42.1 s, total: 47min 32s\n",
      "Wall time: 45min 18s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "#get submission files on global thresholds\n",
    "globl_threshold_products(model = saved_model, scaler = scaler, model_type = 'mlp', val_file=\"Data_v2.h5\")\n",
    "\n",
    "#get products for order ID's and create a submission file based on local threshold\n",
    "test_products = getscores_on_testdata(model = saved_model, scaler = scaler, model_type = 'mlp', val_file=\"Data_v2.h5\")\n",
    "test_products.to_csv(\"submission_mlp_v8.csv\", index= False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "r-um4SlYb_TV"
   },
   "source": [
    "##### Ensemble: XGBoost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "cT6bmKcjvbQc"
   },
   "outputs": [],
   "source": [
    "def train_xgb(X_train, X_test, y_train, y_test, plot_importance = True, save = True, file_name = None, params = None):\n",
    "    \n",
    "    \"\"\"\n",
    "    Returns trained XGB model and output probabilities of validation set\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    X_train              : X_train data is passed \n",
    "    X_test               : X_test data is passed\n",
    "    y_train              : y_train data is passed\n",
    "    y_test               : y_test data is passed\n",
    "    plot_importance      : The default is True\n",
    "                          Boolean variable , Set to True if feature Importance\n",
    "                          needs to be plotted\n",
    "    save                 : The default is True\n",
    "                          Boolean variable , Set to True if trained model \n",
    "                          needs to be saved\n",
    "    file_name            : The default is None\n",
    "                          Filename to be used while saving model\n",
    "    params               : The default is None\n",
    "                          dict of parameter set used to train XGBoost model\n",
    "    Returns\n",
    "    -------\n",
    "    xgb_model            :  Trained XGBoost Model\n",
    "    predict_y            : output probabilities for validation data\n",
    "    \"\"\"\n",
    "\n",
    "    #defining set of parameters, these are optimal parameters obatined after \n",
    "    # rigorous hyperparameter Tuning\n",
    "    if params is None:\n",
    "        params = {}\n",
    "        params['objective'] = 'binary:logistic'\n",
    "        params['eval_metric'] = ['logloss']\n",
    "        params['eta'] = 0.02\n",
    "        params['max_depth'] = 15\n",
    "        params['nthread']=-1\n",
    "        params['colsample_bytree'] = 0.4\n",
    "        params['tree_method'] = 'gpu_hist'\n",
    "        \n",
    "    #Create DataMatrix for XGBoost\n",
    "    d_train = xgb.DMatrix(X_train, label=y_train)\n",
    "    d_test = xgb.DMatrix(X_val, label=y_val)\n",
    "\n",
    "    watchlist = [(d_train, 'train'), (d_test, 'valid')]\n",
    "    \n",
    "    #output Training Time\n",
    "    start_time = datetime.now()\n",
    "    print(\"Training Started :\")\n",
    "    xgb_model = xgb.train(params, d_train, 300, watchlist, early_stopping_rounds=20, verbose_eval=10)\n",
    "    print(\"Training Completed \")\n",
    "    end_time = datetime.now()\n",
    "    difference = end_time - start_time\n",
    "    time = divmod(difference.total_seconds() , 360)\n",
    "    print(\"Total Time : {} minutes {} seconds\".format(time[0], time[1]))\n",
    "    \n",
    "    #get probabilities for validation set\n",
    "    predict_y = xgb_model.predict(d_test)\n",
    "    print(\"The test log loss is:\",log_loss(y_val, predict_y, labels=[0,1], eps=1e-15))\n",
    "    \n",
    "    #save trained model\n",
    "    if save:\n",
    "        #file_name = \"xgb_v1.pkl\"\n",
    "        \n",
    "        # save\n",
    "        pickle.dump(bst, open(file_name, \"wb\"))\n",
    "    \n",
    "    #plot Feature Importance\n",
    "    if plot_importance:\n",
    "        print(\"Feature Importance\")\n",
    "        fig, ax = plt.subplots(1,1,figsize=(10,10))\n",
    "        xgb.plot_importance(booster = bst, ax = ax)\n",
    "\n",
    "    return xgb_model, predict_y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "id": "kUYWA4HpvbN9",
    "outputId": "5d7a1164-ee14-4b70-d555-5e19dd3c32fc"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training Started :\n",
      "[0]\ttrain-logloss:0.67958\tvalid-logloss:0.679697\n",
      "Multiple eval metrics have been passed: 'valid-logloss' will be used for early stopping.\n",
      "\n",
      "Will train until valid-logloss hasn't improved in 20 rounds.\n",
      "[10]\ttrain-logloss:0.565498\tvalid-logloss:0.566859\n",
      "[20]\ttrain-logloss:0.485983\tvalid-logloss:0.488555\n",
      "[30]\ttrain-logloss:0.427436\tvalid-logloss:0.431356\n",
      "[40]\ttrain-logloss:0.381913\tvalid-logloss:0.387167\n",
      "[50]\ttrain-logloss:0.348537\tvalid-logloss:0.355125\n",
      "[60]\ttrain-logloss:0.322154\tvalid-logloss:0.330022\n",
      "[70]\ttrain-logloss:0.302482\tvalid-logloss:0.311674\n",
      "[80]\ttrain-logloss:0.28651\tvalid-logloss:0.297039\n",
      "[90]\ttrain-logloss:0.274486\tvalid-logloss:0.28634\n",
      "[100]\ttrain-logloss:0.263946\tvalid-logloss:0.277057\n",
      "[110]\ttrain-logloss:0.255393\tvalid-logloss:0.269897\n",
      "[120]\ttrain-logloss:0.248648\tvalid-logloss:0.264373\n",
      "[130]\ttrain-logloss:0.243342\tvalid-logloss:0.260319\n",
      "[140]\ttrain-logloss:0.238704\tvalid-logloss:0.257066\n",
      "[150]\ttrain-logloss:0.234887\tvalid-logloss:0.254566\n",
      "[160]\ttrain-logloss:0.231591\tvalid-logloss:0.252578\n",
      "[170]\ttrain-logloss:0.22869\tvalid-logloss:0.251122\n",
      "[180]\ttrain-logloss:0.226143\tvalid-logloss:0.249827\n",
      "[190]\ttrain-logloss:0.223813\tvalid-logloss:0.248738\n",
      "[200]\ttrain-logloss:0.221736\tvalid-logloss:0.247826\n",
      "[210]\ttrain-logloss:0.219797\tvalid-logloss:0.247171\n",
      "[220]\ttrain-logloss:0.218108\tvalid-logloss:0.246626\n",
      "[230]\ttrain-logloss:0.216371\tvalid-logloss:0.245915\n",
      "[240]\ttrain-logloss:0.214782\tvalid-logloss:0.245433\n",
      "[250]\ttrain-logloss:0.213254\tvalid-logloss:0.244975\n",
      "[260]\ttrain-logloss:0.211805\tvalid-logloss:0.244492\n",
      "[270]\ttrain-logloss:0.210502\tvalid-logloss:0.244188\n",
      "[280]\ttrain-logloss:0.209101\tvalid-logloss:0.243867\n",
      "[290]\ttrain-logloss:0.207939\tvalid-logloss:0.243613\n",
      "[299]\ttrain-logloss:0.20696\tvalid-logloss:0.243453\n",
      "Training Completed \n",
      "Total Time : 0.0 hours 306.009438 seconds\n",
      "The test log loss is: 0.24345293402046597\n",
      "Feature Importance\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "xgb_model, predict_y = train_xgb(X_train, X_val, y_train, y_val, True, True, \"xgb_v3.pkl\", None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 299
    },
    "id": "J0siJzDtDKKp",
    "outputId": "d35d7f9d-285e-4e2e-b757-2679698e8ca6"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy : 0.9104625902837514\n",
      "Precision : 0.6673358187553771\n"
     ]
    }
   ],
   "source": [
    "# Plot confusion Matrix\n",
    "predicted_y =np.array(predict_y>0.5,dtype=int)\n",
    "cm = confusion_matrix(y_val, predicted_y)\n",
    "sns.heatmap(cm, cmap='Blues', annot = True, fmt=\"g\", annot_kws={\"size\": 16})\n",
    "plt.show()\n",
    "print(\"Accuracy :\",accuracy_score(y_val, predicted_y))\n",
    "print(\"Precision :\",precision_score(y_val, predicted_y))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "_xnb636MeoIi"
   },
   "source": [
    "#### Create a Submission File"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "jRJwOmM2ozyg",
    "outputId": "06a2514e-65a1-4a4a-ed6d-1bf68d92eadb"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generated submission_sample_v6.1_1_xgb.csv successfully\n",
      "Generated submission_sample_v6.1_2_xgb.csv successfully\n",
      "Generated submission_sample_v6.1_3_xgb.csv successfully\n"
     ]
    }
   ],
   "source": [
    "#generate submission files based on global threshold\n",
    "globl_threshold_products(model = pickle.load(open(\"xgb_v3.pkl\", \"rb\")), model_type = 'xgb')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "7WV-CuIBvbLq",
    "outputId": "e7d7f9c2-7709-4d26-c054-0492a57f6448"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 75000/75000 [42:23<00:00, 29.48it/s]\n"
     ]
    }
   ],
   "source": [
    "#get products for order ID's and create a submission file based on local threshold\n",
    "test_products = getscores_on_testdata(model = pickle.load(open(\"xgb_v3.pkl\", \"rb\")), model_type = 'xgb')\n",
    "test_products.to_csv(\"submission_xgb_v6.csv\", index= False) # 0.39155 - 0.39303"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "dPd9Ratre4Uu"
   },
   "source": [
    "### Ensemble: Catboost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "-emRO3I1ymlB"
   },
   "outputs": [],
   "source": [
    "def train_catboost(X_train, X_test, y_train, y_test, plot_importance = True, save = True, file_name = None):\n",
    "    \n",
    "    \"\"\"\n",
    "    Returns trained catboost model and output probabilities of validation set\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    X_train              : X_train data is passed \n",
    "    X_test               : X_test data is passed\n",
    "    y_train              : y_train data is passed\n",
    "    y_test               : y_test data is passed\n",
    "    plot_importance      : The default is True\n",
    "                          Boolean variable , Set to True if feature Importance\n",
    "                          needs to be plotted\n",
    "    save                 : The default is True\n",
    "                          Boolean variable , Set to True if trained model \n",
    "                          needs to be saved\n",
    "    file_name            : The default is None\n",
    "                          Filename to be used while saving model\n",
    "    Returns\n",
    "    -------\n",
    "    xgb_model            :  Trained XGBoost Model\n",
    "    predict_y            : output probabilities for validation data\n",
    "    \"\"\"\n",
    "\n",
    "\n",
    "    start_time = datetime.now()\n",
    "    print(\"Training Started :\")\n",
    "\n",
    "    #defining set of parameters, these are optimal parameters obatined after \n",
    "    # rigorous hyperparameter Tuning\n",
    "    c_model = CatBoostClassifier(task_type = \"GPU\",verbose=True,depth = 13, iterations= 2000,learning_rate= 0.02,scale_pos_weight= 1.0)\n",
    "    c_model.fit(X_train,y_train)\n",
    "    print(\"Training Completed \")\n",
    "    end_time = datetime.now()\n",
    "    difference = end_time - start_time\n",
    "    time = divmod(difference.total_seconds() , 3600)\n",
    "    print(\"Total Time : {} hours {} seconds\".format(time[0], time[1]))\n",
    "    \n",
    "    #get output probabilities\n",
    "    predict_y = c_model.predict_proba(X_test)\n",
    "    print(\"The Test log loss is:\",log_loss(y_test, predict_y, labels=[0,1], eps=1e-15))\n",
    "    \n",
    "    #get probs for class 1\n",
    "    predict_y = predict_y[:,-1]\n",
    "    \n",
    "    #save the model\n",
    "    if save:\n",
    "        # save\n",
    "        pickle.dump(c_model, open(file_name, \"wb\"))\n",
    "    \n",
    "    #plot feature importance\n",
    "    if plot_importance:\n",
    "        #ref: https://stackoverflow.com/a/65842279/11533069\n",
    "        f_imp=pd.DataFrame({'features':train_x.columns.to_numpy(),'feature_importance': c_model.get_feature_importance()})\n",
    "        f_imp.sort_values(by = 'feature_importance', ascending = False, inplace = True)\n",
    "\n",
    "        print(\"Feature Importance\")\n",
    "        fig, ax = plt.subplots(1,1,figsize=(10,10))\n",
    "        sns.barplot(x=f_imp['feature_importance'], y=f_imp['features'])\n",
    "        plt.title('Feature Importance')\n",
    "        plt.xlabel('Importance')\n",
    "        plt.ylabel('Features')\n",
    "\n",
    "    return c_model, predict_y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "id": "3LSn_6_P3qWy",
    "outputId": "85c7b677-92bb-4ea8-f4c4-1e56aa0074fe"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training Started :\n",
      "0:\tlearn: 0.6682243\ttotal: 213ms\tremaining: 7m 4s\n",
      "1:\tlearn: 0.6447123\ttotal: 413ms\tremaining: 6m 52s\n",
      "2:\tlearn: 0.6226377\ttotal: 616ms\tremaining: 6m 50s\n",
      "3:\tlearn: 0.6019817\ttotal: 818ms\tremaining: 6m 48s\n",
      "4:\tlearn: 0.5821403\ttotal: 1.02s\tremaining: 6m 46s\n",
      "5:\tlearn: 0.5640026\ttotal: 1.23s\tremaining: 6m 48s\n",
      "6:\tlearn: 0.5468472\ttotal: 1.44s\tremaining: 6m 49s\n",
      "7:\tlearn: 0.5307860\ttotal: 1.64s\tremaining: 6m 49s\n",
      "8:\tlearn: 0.5154758\ttotal: 1.84s\tremaining: 6m 48s\n",
      "9:\tlearn: 0.5011645\ttotal: 2.05s\tremaining: 6m 48s\n",
      "10:\tlearn: 0.4874874\ttotal: 2.26s\tremaining: 6m 48s\n",
      "11:\tlearn: 0.4746784\ttotal: 2.47s\tremaining: 6m 49s\n",
      "12:\tlearn: 0.4626853\ttotal: 2.68s\tremaining: 6m 49s\n",
      "13:\tlearn: 0.4509007\ttotal: 2.88s\tremaining: 6m 49s\n",
      "14:\tlearn: 0.4401764\ttotal: 3.09s\tremaining: 6m 48s\n",
      "15:\tlearn: 0.4299006\ttotal: 3.29s\tremaining: 6m 48s\n",
      "16:\tlearn: 0.4204155\ttotal: 3.49s\tremaining: 6m 47s\n",
      "17:\tlearn: 0.4112568\ttotal: 3.7s\tremaining: 6m 47s\n",
      "18:\tlearn: 0.4027624\ttotal: 3.9s\tremaining: 6m 47s\n",
      "19:\tlearn: 0.3948050\ttotal: 4.1s\tremaining: 6m 46s\n",
      "20:\tlearn: 0.3871880\ttotal: 4.31s\tremaining: 6m 46s\n",
      "21:\tlearn: 0.3802699\ttotal: 4.51s\tremaining: 6m 45s\n",
      "22:\tlearn: 0.3733568\ttotal: 4.72s\tremaining: 6m 45s\n",
      "23:\tlearn: 0.3668651\ttotal: 4.93s\tremaining: 6m 45s\n",
      "24:\tlearn: 0.3609307\ttotal: 5.13s\tremaining: 6m 45s\n",
      "25:\tlearn: 0.3553763\ttotal: 5.34s\tremaining: 6m 45s\n",
      "26:\tlearn: 0.3501678\ttotal: 5.54s\tremaining: 6m 44s\n",
      "27:\tlearn: 0.3452065\ttotal: 5.75s\tremaining: 6m 44s\n",
      "28:\tlearn: 0.3404242\ttotal: 5.96s\tremaining: 6m 44s\n",
      "29:\tlearn: 0.3359604\ttotal: 6.16s\tremaining: 6m 44s\n",
      "30:\tlearn: 0.3317464\ttotal: 6.37s\tremaining: 6m 44s\n",
      "31:\tlearn: 0.3277068\ttotal: 6.57s\tremaining: 6m 44s\n",
      "32:\tlearn: 0.3239622\ttotal: 6.78s\tremaining: 6m 43s\n",
      "33:\tlearn: 0.3203739\ttotal: 6.98s\tremaining: 6m 43s\n",
      "34:\tlearn: 0.3169466\ttotal: 7.19s\tremaining: 6m 43s\n",
      "35:\tlearn: 0.3137291\ttotal: 7.4s\tremaining: 6m 43s\n",
      "36:\tlearn: 0.3105916\ttotal: 7.6s\tremaining: 6m 43s\n",
      "37:\tlearn: 0.3077527\ttotal: 7.82s\tremaining: 6m 43s\n",
      "38:\tlearn: 0.3049923\ttotal: 8.03s\tremaining: 6m 43s\n",
      "39:\tlearn: 0.3024168\ttotal: 8.23s\tremaining: 6m 43s\n",
      "40:\tlearn: 0.2999867\ttotal: 8.44s\tremaining: 6m 43s\n",
      "41:\tlearn: 0.2976672\ttotal: 8.65s\tremaining: 6m 43s\n",
      "42:\tlearn: 0.2954416\ttotal: 8.86s\tremaining: 6m 43s\n",
      "43:\tlearn: 0.2933098\ttotal: 9.06s\tremaining: 6m 42s\n",
      "44:\tlearn: 0.2912776\ttotal: 9.27s\tremaining: 6m 42s\n",
      "45:\tlearn: 0.2893922\ttotal: 9.48s\tremaining: 6m 42s\n",
      "46:\tlearn: 0.2875704\ttotal: 9.68s\tremaining: 6m 42s\n",
      "47:\tlearn: 0.2858041\ttotal: 9.89s\tremaining: 6m 42s\n",
      "48:\tlearn: 0.2841284\ttotal: 10.1s\tremaining: 6m 42s\n",
      "49:\tlearn: 0.2826105\ttotal: 10.3s\tremaining: 6m 41s\n",
      "50:\tlearn: 0.2811426\ttotal: 10.5s\tremaining: 6m 42s\n",
      "51:\tlearn: 0.2797211\ttotal: 10.7s\tremaining: 6m 41s\n",
      "52:\tlearn: 0.2783655\ttotal: 10.9s\tremaining: 6m 41s\n",
      "53:\tlearn: 0.2770374\ttotal: 11.1s\tremaining: 6m 41s\n",
      "54:\tlearn: 0.2757932\ttotal: 11.4s\tremaining: 6m 41s\n",
      "55:\tlearn: 0.2746249\ttotal: 11.6s\tremaining: 6m 41s\n",
      "56:\tlearn: 0.2734662\ttotal: 11.8s\tremaining: 6m 41s\n",
      "57:\tlearn: 0.2723479\ttotal: 12s\tremaining: 6m 40s\n",
      "58:\tlearn: 0.2713512\ttotal: 12.2s\tremaining: 6m 40s\n",
      "59:\tlearn: 0.2704143\ttotal: 12.4s\tremaining: 6m 40s\n",
      "60:\tlearn: 0.2695112\ttotal: 12.6s\tremaining: 6m 40s\n",
      "61:\tlearn: 0.2686362\ttotal: 12.8s\tremaining: 6m 39s\n",
      "62:\tlearn: 0.2677411\ttotal: 13s\tremaining: 6m 39s\n",
      "63:\tlearn: 0.2669638\ttotal: 13.2s\tremaining: 6m 39s\n",
      "64:\tlearn: 0.2662086\ttotal: 13.4s\tremaining: 6m 39s\n",
      "65:\tlearn: 0.2654710\ttotal: 13.6s\tremaining: 6m 39s\n",
      "66:\tlearn: 0.2647336\ttotal: 13.8s\tremaining: 6m 39s\n",
      "67:\tlearn: 0.2640018\ttotal: 14s\tremaining: 6m 38s\n",
      "68:\tlearn: 0.2633695\ttotal: 14.2s\tremaining: 6m 38s\n",
      "69:\tlearn: 0.2627835\ttotal: 14.4s\tremaining: 6m 38s\n",
      "70:\tlearn: 0.2621980\ttotal: 14.6s\tremaining: 6m 37s\n",
      "71:\tlearn: 0.2616417\ttotal: 14.8s\tremaining: 6m 37s\n",
      "72:\tlearn: 0.2611125\ttotal: 15.1s\tremaining: 6m 37s\n",
      "73:\tlearn: 0.2606183\ttotal: 15.3s\tremaining: 6m 37s\n",
      "74:\tlearn: 0.2601488\ttotal: 15.5s\tremaining: 6m 37s\n",
      "75:\tlearn: 0.2596869\ttotal: 15.7s\tremaining: 6m 36s\n",
      "76:\tlearn: 0.2592371\ttotal: 15.9s\tremaining: 6m 36s\n",
      "77:\tlearn: 0.2587636\ttotal: 16.1s\tremaining: 6m 36s\n",
      "78:\tlearn: 0.2583267\ttotal: 16.3s\tremaining: 6m 36s\n",
      "79:\tlearn: 0.2579011\ttotal: 16.5s\tremaining: 6m 35s\n",
      "80:\tlearn: 0.2575210\ttotal: 16.7s\tremaining: 6m 35s\n",
      "81:\tlearn: 0.2571692\ttotal: 16.9s\tremaining: 6m 35s\n",
      "82:\tlearn: 0.2568284\ttotal: 17.1s\tremaining: 6m 35s\n",
      "83:\tlearn: 0.2564872\ttotal: 17.3s\tremaining: 6m 35s\n",
      "84:\tlearn: 0.2561523\ttotal: 17.5s\tremaining: 6m 35s\n",
      "85:\tlearn: 0.2558117\ttotal: 17.7s\tremaining: 6m 34s\n",
      "86:\tlearn: 0.2555080\ttotal: 17.9s\tremaining: 6m 34s\n",
      "87:\tlearn: 0.2552380\ttotal: 18.1s\tremaining: 6m 34s\n",
      "88:\tlearn: 0.2549723\ttotal: 18.4s\tremaining: 6m 34s\n",
      "89:\tlearn: 0.2547211\ttotal: 18.6s\tremaining: 6m 34s\n",
      "90:\tlearn: 0.2544365\ttotal: 18.8s\tremaining: 6m 33s\n",
      "91:\tlearn: 0.2542034\ttotal: 19s\tremaining: 6m 33s\n",
      "92:\tlearn: 0.2539791\ttotal: 19.2s\tremaining: 6m 33s\n",
      "93:\tlearn: 0.2537298\ttotal: 19.4s\tremaining: 6m 32s\n",
      "94:\tlearn: 0.2534924\ttotal: 19.6s\tremaining: 6m 32s\n",
      "95:\tlearn: 0.2532913\ttotal: 19.8s\tremaining: 6m 32s\n",
      "96:\tlearn: 0.2530829\ttotal: 20s\tremaining: 6m 32s\n",
      "97:\tlearn: 0.2528487\ttotal: 20.2s\tremaining: 6m 32s\n",
      "98:\tlearn: 0.2526426\ttotal: 20.4s\tremaining: 6m 32s\n",
      "99:\tlearn: 0.2524426\ttotal: 20.6s\tremaining: 6m 31s\n",
      "100:\tlearn: 0.2522680\ttotal: 20.8s\tremaining: 6m 31s\n",
      "101:\tlearn: 0.2520716\ttotal: 21s\tremaining: 6m 31s\n",
      "102:\tlearn: 0.2519122\ttotal: 21.2s\tremaining: 6m 31s\n",
      "103:\tlearn: 0.2517454\ttotal: 21.4s\tremaining: 6m 30s\n",
      "104:\tlearn: 0.2516038\ttotal: 21.6s\tremaining: 6m 30s\n",
      "105:\tlearn: 0.2514583\ttotal: 21.9s\tremaining: 6m 30s\n",
      "106:\tlearn: 0.2513334\ttotal: 22.1s\tremaining: 6m 30s\n",
      "107:\tlearn: 0.2511795\ttotal: 22.3s\tremaining: 6m 30s\n",
      "108:\tlearn: 0.2510185\ttotal: 22.5s\tremaining: 6m 29s\n",
      "109:\tlearn: 0.2508938\ttotal: 22.7s\tremaining: 6m 29s\n",
      "110:\tlearn: 0.2507455\ttotal: 22.9s\tremaining: 6m 29s\n",
      "111:\tlearn: 0.2506060\ttotal: 23.1s\tremaining: 6m 29s\n",
      "112:\tlearn: 0.2504977\ttotal: 23.3s\tremaining: 6m 29s\n",
      "113:\tlearn: 0.2503822\ttotal: 23.5s\tremaining: 6m 28s\n",
      "114:\tlearn: 0.2502850\ttotal: 23.7s\tremaining: 6m 28s\n",
      "115:\tlearn: 0.2501600\ttotal: 23.9s\tremaining: 6m 28s\n",
      "116:\tlearn: 0.2500549\ttotal: 24.1s\tremaining: 6m 27s\n",
      "117:\tlearn: 0.2499646\ttotal: 24.3s\tremaining: 6m 27s\n",
      "118:\tlearn: 0.2498409\ttotal: 24.5s\tremaining: 6m 27s\n",
      "119:\tlearn: 0.2497434\ttotal: 24.7s\tremaining: 6m 27s\n",
      "120:\tlearn: 0.2496518\ttotal: 24.9s\tremaining: 6m 27s\n",
      "121:\tlearn: 0.2495569\ttotal: 25.1s\tremaining: 6m 26s\n",
      "122:\tlearn: 0.2494645\ttotal: 25.3s\tremaining: 6m 26s\n",
      "123:\tlearn: 0.2493822\ttotal: 25.6s\tremaining: 6m 26s\n",
      "124:\tlearn: 0.2493123\ttotal: 25.8s\tremaining: 6m 26s\n",
      "125:\tlearn: 0.2492199\ttotal: 26s\tremaining: 6m 26s\n",
      "126:\tlearn: 0.2491406\ttotal: 26.2s\tremaining: 6m 26s\n",
      "127:\tlearn: 0.2490629\ttotal: 26.4s\tremaining: 6m 25s\n",
      "128:\tlearn: 0.2489790\ttotal: 26.6s\tremaining: 6m 25s\n",
      "129:\tlearn: 0.2489136\ttotal: 26.8s\tremaining: 6m 25s\n",
      "130:\tlearn: 0.2488313\ttotal: 27s\tremaining: 6m 25s\n",
      "131:\tlearn: 0.2487735\ttotal: 27.2s\tremaining: 6m 25s\n",
      "132:\tlearn: 0.2487007\ttotal: 27.4s\tremaining: 6m 25s\n",
      "133:\tlearn: 0.2486291\ttotal: 27.7s\tremaining: 6m 25s\n",
      "134:\tlearn: 0.2485794\ttotal: 27.9s\tremaining: 6m 24s\n",
      "135:\tlearn: 0.2485312\ttotal: 28.1s\tremaining: 6m 24s\n",
      "136:\tlearn: 0.2484646\ttotal: 28.3s\tremaining: 6m 24s\n",
      "137:\tlearn: 0.2483991\ttotal: 28.5s\tremaining: 6m 24s\n",
      "138:\tlearn: 0.2483438\ttotal: 28.7s\tremaining: 6m 24s\n",
      "139:\tlearn: 0.2482893\ttotal: 28.9s\tremaining: 6m 24s\n",
      "140:\tlearn: 0.2482356\ttotal: 29.1s\tremaining: 6m 23s\n",
      "141:\tlearn: 0.2481758\ttotal: 29.3s\tremaining: 6m 23s\n",
      "142:\tlearn: 0.2481256\ttotal: 29.5s\tremaining: 6m 23s\n",
      "143:\tlearn: 0.2480814\ttotal: 29.8s\tremaining: 6m 23s\n",
      "144:\tlearn: 0.2480160\ttotal: 30s\tremaining: 6m 23s\n",
      "145:\tlearn: 0.2479793\ttotal: 30.2s\tremaining: 6m 23s\n",
      "146:\tlearn: 0.2479277\ttotal: 30.4s\tremaining: 6m 23s\n",
      "147:\tlearn: 0.2478782\ttotal: 30.6s\tremaining: 6m 22s\n",
      "148:\tlearn: 0.2478315\ttotal: 30.8s\tremaining: 6m 22s\n",
      "149:\tlearn: 0.2477890\ttotal: 31s\tremaining: 6m 22s\n",
      "150:\tlearn: 0.2477387\ttotal: 31.2s\tremaining: 6m 22s\n",
      "151:\tlearn: 0.2477000\ttotal: 31.4s\tremaining: 6m 22s\n",
      "152:\tlearn: 0.2476644\ttotal: 31.6s\tremaining: 6m 22s\n",
      "153:\tlearn: 0.2476274\ttotal: 31.9s\tremaining: 6m 21s\n",
      "154:\tlearn: 0.2475853\ttotal: 32.1s\tremaining: 6m 21s\n",
      "155:\tlearn: 0.2475313\ttotal: 32.3s\tremaining: 6m 21s\n",
      "156:\tlearn: 0.2474801\ttotal: 32.5s\tremaining: 6m 21s\n",
      "157:\tlearn: 0.2474376\ttotal: 32.7s\tremaining: 6m 20s\n",
      "158:\tlearn: 0.2473884\ttotal: 32.9s\tremaining: 6m 20s\n",
      "159:\tlearn: 0.2473429\ttotal: 33.1s\tremaining: 6m 20s\n",
      "160:\tlearn: 0.2473095\ttotal: 33.3s\tremaining: 6m 20s\n",
      "161:\tlearn: 0.2472654\ttotal: 33.5s\tremaining: 6m 20s\n",
      "162:\tlearn: 0.2472232\ttotal: 33.7s\tremaining: 6m 19s\n",
      "163:\tlearn: 0.2471837\ttotal: 33.9s\tremaining: 6m 19s\n",
      "164:\tlearn: 0.2471562\ttotal: 34.1s\tremaining: 6m 19s\n",
      "165:\tlearn: 0.2471159\ttotal: 34.3s\tremaining: 6m 19s\n",
      "166:\tlearn: 0.2470800\ttotal: 34.5s\tremaining: 6m 19s\n",
      "167:\tlearn: 0.2470355\ttotal: 34.7s\tremaining: 6m 18s\n",
      "168:\tlearn: 0.2469970\ttotal: 34.9s\tremaining: 6m 18s\n",
      "169:\tlearn: 0.2469608\ttotal: 35.2s\tremaining: 6m 18s\n",
      "170:\tlearn: 0.2469297\ttotal: 35.4s\tremaining: 6m 18s\n",
      "171:\tlearn: 0.2469053\ttotal: 35.6s\tremaining: 6m 18s\n",
      "172:\tlearn: 0.2468704\ttotal: 35.8s\tremaining: 6m 17s\n",
      "173:\tlearn: 0.2468375\ttotal: 36s\tremaining: 6m 17s\n",
      "174:\tlearn: 0.2468139\ttotal: 36.2s\tremaining: 6m 17s\n",
      "175:\tlearn: 0.2467880\ttotal: 36.4s\tremaining: 6m 17s\n",
      "176:\tlearn: 0.2467522\ttotal: 36.6s\tremaining: 6m 17s\n",
      "177:\tlearn: 0.2467247\ttotal: 36.8s\tremaining: 6m 16s\n",
      "178:\tlearn: 0.2466984\ttotal: 37s\tremaining: 6m 16s\n",
      "179:\tlearn: 0.2466723\ttotal: 37.2s\tremaining: 6m 16s\n",
      "180:\tlearn: 0.2466393\ttotal: 37.5s\tremaining: 6m 16s\n",
      "181:\tlearn: 0.2466079\ttotal: 37.7s\tremaining: 6m 16s\n",
      "182:\tlearn: 0.2465789\ttotal: 37.9s\tremaining: 6m 15s\n",
      "183:\tlearn: 0.2465536\ttotal: 38.1s\tremaining: 6m 15s\n",
      "184:\tlearn: 0.2465303\ttotal: 38.3s\tremaining: 6m 15s\n",
      "185:\tlearn: 0.2465053\ttotal: 38.5s\tremaining: 6m 15s\n",
      "186:\tlearn: 0.2464823\ttotal: 38.7s\tremaining: 6m 15s\n",
      "187:\tlearn: 0.2464609\ttotal: 38.9s\tremaining: 6m 14s\n",
      "188:\tlearn: 0.2464386\ttotal: 39.1s\tremaining: 6m 14s\n",
      "189:\tlearn: 0.2464136\ttotal: 39.3s\tremaining: 6m 14s\n",
      "190:\tlearn: 0.2463865\ttotal: 39.5s\tremaining: 6m 14s\n",
      "191:\tlearn: 0.2463625\ttotal: 39.7s\tremaining: 6m 13s\n",
      "192:\tlearn: 0.2463361\ttotal: 39.9s\tremaining: 6m 13s\n",
      "193:\tlearn: 0.2463077\ttotal: 40.1s\tremaining: 6m 13s\n",
      "194:\tlearn: 0.2462844\ttotal: 40.3s\tremaining: 6m 13s\n",
      "195:\tlearn: 0.2462546\ttotal: 40.5s\tremaining: 6m 13s\n",
      "196:\tlearn: 0.2462276\ttotal: 40.7s\tremaining: 6m 12s\n",
      "197:\tlearn: 0.2461963\ttotal: 41s\tremaining: 6m 12s\n",
      "198:\tlearn: 0.2461693\ttotal: 41.2s\tremaining: 6m 12s\n",
      "199:\tlearn: 0.2461477\ttotal: 41.4s\tremaining: 6m 12s\n",
      "200:\tlearn: 0.2461295\ttotal: 41.6s\tremaining: 6m 12s\n",
      "201:\tlearn: 0.2461126\ttotal: 41.8s\tremaining: 6m 11s\n",
      "202:\tlearn: 0.2460894\ttotal: 42s\tremaining: 6m 11s\n",
      "203:\tlearn: 0.2460692\ttotal: 42.2s\tremaining: 6m 11s\n",
      "204:\tlearn: 0.2460455\ttotal: 42.4s\tremaining: 6m 11s\n",
      "205:\tlearn: 0.2460204\ttotal: 42.6s\tremaining: 6m 11s\n",
      "206:\tlearn: 0.2460041\ttotal: 42.8s\tremaining: 6m 11s\n",
      "207:\tlearn: 0.2459850\ttotal: 43s\tremaining: 6m 10s\n",
      "208:\tlearn: 0.2459606\ttotal: 43.3s\tremaining: 6m 10s\n",
      "209:\tlearn: 0.2459396\ttotal: 43.5s\tremaining: 6m 10s\n",
      "210:\tlearn: 0.2459227\ttotal: 43.7s\tremaining: 6m 10s\n",
      "211:\tlearn: 0.2459022\ttotal: 43.9s\tremaining: 6m 10s\n",
      "212:\tlearn: 0.2458811\ttotal: 44.1s\tremaining: 6m 9s\n",
      "213:\tlearn: 0.2458546\ttotal: 44.3s\tremaining: 6m 9s\n",
      "214:\tlearn: 0.2458365\ttotal: 44.5s\tremaining: 6m 9s\n",
      "215:\tlearn: 0.2458168\ttotal: 44.7s\tremaining: 6m 9s\n",
      "216:\tlearn: 0.2457982\ttotal: 44.9s\tremaining: 6m 9s\n",
      "217:\tlearn: 0.2457746\ttotal: 45.1s\tremaining: 6m 8s\n",
      "218:\tlearn: 0.2457546\ttotal: 45.3s\tremaining: 6m 8s\n",
      "219:\tlearn: 0.2457398\ttotal: 45.5s\tremaining: 6m 8s\n",
      "220:\tlearn: 0.2457142\ttotal: 45.7s\tremaining: 6m 8s\n",
      "221:\tlearn: 0.2456921\ttotal: 45.9s\tremaining: 6m 7s\n",
      "222:\tlearn: 0.2456760\ttotal: 46.2s\tremaining: 6m 7s\n",
      "223:\tlearn: 0.2456577\ttotal: 46.4s\tremaining: 6m 7s\n",
      "224:\tlearn: 0.2456310\ttotal: 46.6s\tremaining: 6m 7s\n",
      "225:\tlearn: 0.2456124\ttotal: 46.8s\tremaining: 6m 7s\n",
      "226:\tlearn: 0.2455939\ttotal: 47s\tremaining: 6m 7s\n",
      "227:\tlearn: 0.2455735\ttotal: 47.2s\tremaining: 6m 6s\n",
      "228:\tlearn: 0.2455542\ttotal: 47.4s\tremaining: 6m 6s\n",
      "229:\tlearn: 0.2455375\ttotal: 47.6s\tremaining: 6m 6s\n",
      "230:\tlearn: 0.2455197\ttotal: 47.9s\tremaining: 6m 6s\n",
      "231:\tlearn: 0.2454992\ttotal: 48.1s\tremaining: 6m 6s\n",
      "232:\tlearn: 0.2454778\ttotal: 48.3s\tremaining: 6m 6s\n",
      "233:\tlearn: 0.2454609\ttotal: 48.5s\tremaining: 6m 5s\n",
      "234:\tlearn: 0.2454417\ttotal: 48.7s\tremaining: 6m 5s\n",
      "235:\tlearn: 0.2454136\ttotal: 48.9s\tremaining: 6m 5s\n",
      "236:\tlearn: 0.2453912\ttotal: 49.1s\tremaining: 6m 5s\n",
      "237:\tlearn: 0.2453689\ttotal: 49.3s\tremaining: 6m 5s\n",
      "238:\tlearn: 0.2453503\ttotal: 49.5s\tremaining: 6m 5s\n",
      "239:\tlearn: 0.2453378\ttotal: 49.8s\tremaining: 6m 4s\n",
      "240:\tlearn: 0.2453157\ttotal: 50s\tremaining: 6m 4s\n",
      "241:\tlearn: 0.2452893\ttotal: 50.2s\tremaining: 6m 4s\n",
      "242:\tlearn: 0.2452770\ttotal: 50.4s\tremaining: 6m 4s\n",
      "243:\tlearn: 0.2452578\ttotal: 50.6s\tremaining: 6m 4s\n",
      "244:\tlearn: 0.2452407\ttotal: 50.8s\tremaining: 6m 4s\n",
      "245:\tlearn: 0.2452212\ttotal: 51s\tremaining: 6m 3s\n",
      "246:\tlearn: 0.2452036\ttotal: 51.2s\tremaining: 6m 3s\n",
      "247:\tlearn: 0.2451899\ttotal: 51.4s\tremaining: 6m 3s\n",
      "248:\tlearn: 0.2451765\ttotal: 51.6s\tremaining: 6m 3s\n",
      "249:\tlearn: 0.2451573\ttotal: 51.9s\tremaining: 6m 2s\n",
      "250:\tlearn: 0.2451439\ttotal: 52.1s\tremaining: 6m 2s\n",
      "251:\tlearn: 0.2451226\ttotal: 52.3s\tremaining: 6m 2s\n",
      "252:\tlearn: 0.2451115\ttotal: 52.5s\tremaining: 6m 2s\n",
      "253:\tlearn: 0.2450899\ttotal: 52.7s\tremaining: 6m 2s\n",
      "254:\tlearn: 0.2450687\ttotal: 52.9s\tremaining: 6m 1s\n",
      "255:\tlearn: 0.2450553\ttotal: 53.1s\tremaining: 6m 1s\n",
      "256:\tlearn: 0.2450390\ttotal: 53.3s\tremaining: 6m 1s\n",
      "257:\tlearn: 0.2450272\ttotal: 53.5s\tremaining: 6m 1s\n",
      "258:\tlearn: 0.2450065\ttotal: 53.7s\tremaining: 6m 1s\n",
      "259:\tlearn: 0.2449876\ttotal: 53.9s\tremaining: 6m\n",
      "260:\tlearn: 0.2449760\ttotal: 54.1s\tremaining: 6m\n",
      "261:\tlearn: 0.2449614\ttotal: 54.4s\tremaining: 6m\n",
      "262:\tlearn: 0.2449440\ttotal: 54.6s\tremaining: 6m\n",
      "263:\tlearn: 0.2449348\ttotal: 54.8s\tremaining: 6m\n",
      "264:\tlearn: 0.2449195\ttotal: 55s\tremaining: 6m\n",
      "265:\tlearn: 0.2449055\ttotal: 55.2s\tremaining: 5m 59s\n",
      "266:\tlearn: 0.2448935\ttotal: 55.4s\tremaining: 5m 59s\n",
      "267:\tlearn: 0.2448815\ttotal: 55.6s\tremaining: 5m 59s\n",
      "268:\tlearn: 0.2448616\ttotal: 55.8s\tremaining: 5m 59s\n",
      "269:\tlearn: 0.2448491\ttotal: 56s\tremaining: 5m 58s\n",
      "270:\tlearn: 0.2448314\ttotal: 56.2s\tremaining: 5m 58s\n",
      "271:\tlearn: 0.2448192\ttotal: 56.5s\tremaining: 5m 58s\n",
      "272:\tlearn: 0.2447968\ttotal: 56.7s\tremaining: 5m 58s\n",
      "273:\tlearn: 0.2447875\ttotal: 56.9s\tremaining: 5m 58s\n",
      "274:\tlearn: 0.2447655\ttotal: 57.1s\tremaining: 5m 58s\n",
      "275:\tlearn: 0.2447477\ttotal: 57.3s\tremaining: 5m 57s\n",
      "276:\tlearn: 0.2447310\ttotal: 57.5s\tremaining: 5m 57s\n",
      "277:\tlearn: 0.2447153\ttotal: 57.7s\tremaining: 5m 57s\n",
      "278:\tlearn: 0.2447001\ttotal: 57.9s\tremaining: 5m 57s\n",
      "279:\tlearn: 0.2446830\ttotal: 58.2s\tremaining: 5m 57s\n",
      "280:\tlearn: 0.2446684\ttotal: 58.4s\tremaining: 5m 57s\n",
      "281:\tlearn: 0.2446475\ttotal: 58.6s\tremaining: 5m 56s\n",
      "282:\tlearn: 0.2446262\ttotal: 58.8s\tremaining: 5m 56s\n",
      "283:\tlearn: 0.2446115\ttotal: 59s\tremaining: 5m 56s\n",
      "284:\tlearn: 0.2445909\ttotal: 59.2s\tremaining: 5m 56s\n",
      "285:\tlearn: 0.2445719\ttotal: 59.4s\tremaining: 5m 56s\n",
      "286:\tlearn: 0.2445619\ttotal: 59.7s\tremaining: 5m 56s\n",
      "287:\tlearn: 0.2445431\ttotal: 59.9s\tremaining: 5m 55s\n",
      "288:\tlearn: 0.2445275\ttotal: 1m\tremaining: 5m 55s\n",
      "289:\tlearn: 0.2445158\ttotal: 1m\tremaining: 5m 55s\n",
      "290:\tlearn: 0.2445003\ttotal: 1m\tremaining: 5m 55s\n",
      "291:\tlearn: 0.2444833\ttotal: 1m\tremaining: 5m 55s\n",
      "292:\tlearn: 0.2444688\ttotal: 1m\tremaining: 5m 54s\n",
      "293:\tlearn: 0.2444504\ttotal: 1m 1s\tremaining: 5m 54s\n",
      "294:\tlearn: 0.2444397\ttotal: 1m 1s\tremaining: 5m 54s\n",
      "295:\tlearn: 0.2444188\ttotal: 1m 1s\tremaining: 5m 54s\n",
      "296:\tlearn: 0.2443979\ttotal: 1m 1s\tremaining: 5m 54s\n",
      "297:\tlearn: 0.2443825\ttotal: 1m 1s\tremaining: 5m 53s\n",
      "298:\tlearn: 0.2443662\ttotal: 1m 2s\tremaining: 5m 53s\n",
      "299:\tlearn: 0.2443497\ttotal: 1m 2s\tremaining: 5m 53s\n",
      "300:\tlearn: 0.2443363\ttotal: 1m 2s\tremaining: 5m 53s\n",
      "301:\tlearn: 0.2443276\ttotal: 1m 2s\tremaining: 5m 53s\n",
      "302:\tlearn: 0.2443090\ttotal: 1m 3s\tremaining: 5m 52s\n",
      "303:\tlearn: 0.2442913\ttotal: 1m 3s\tremaining: 5m 52s\n",
      "304:\tlearn: 0.2442814\ttotal: 1m 3s\tremaining: 5m 52s\n",
      "305:\tlearn: 0.2442659\ttotal: 1m 3s\tremaining: 5m 52s\n",
      "306:\tlearn: 0.2442434\ttotal: 1m 3s\tremaining: 5m 52s\n",
      "307:\tlearn: 0.2442276\ttotal: 1m 4s\tremaining: 5m 52s\n",
      "308:\tlearn: 0.2441985\ttotal: 1m 4s\tremaining: 5m 51s\n",
      "309:\tlearn: 0.2441868\ttotal: 1m 4s\tremaining: 5m 51s\n",
      "310:\tlearn: 0.2441626\ttotal: 1m 4s\tremaining: 5m 51s\n",
      "311:\tlearn: 0.2441419\ttotal: 1m 4s\tremaining: 5m 51s\n",
      "312:\tlearn: 0.2441278\ttotal: 1m 5s\tremaining: 5m 51s\n",
      "313:\tlearn: 0.2441160\ttotal: 1m 5s\tremaining: 5m 50s\n",
      "314:\tlearn: 0.2441028\ttotal: 1m 5s\tremaining: 5m 50s\n",
      "315:\tlearn: 0.2440860\ttotal: 1m 5s\tremaining: 5m 50s\n",
      "316:\tlearn: 0.2440633\ttotal: 1m 5s\tremaining: 5m 50s\n",
      "317:\tlearn: 0.2440498\ttotal: 1m 6s\tremaining: 5m 50s\n",
      "318:\tlearn: 0.2440336\ttotal: 1m 6s\tremaining: 5m 50s\n",
      "319:\tlearn: 0.2440200\ttotal: 1m 6s\tremaining: 5m 49s\n",
      "320:\tlearn: 0.2440058\ttotal: 1m 6s\tremaining: 5m 49s\n",
      "321:\tlearn: 0.2439924\ttotal: 1m 7s\tremaining: 5m 49s\n",
      "322:\tlearn: 0.2439766\ttotal: 1m 7s\tremaining: 5m 49s\n",
      "323:\tlearn: 0.2439565\ttotal: 1m 7s\tremaining: 5m 48s\n",
      "324:\tlearn: 0.2439406\ttotal: 1m 7s\tremaining: 5m 48s\n",
      "325:\tlearn: 0.2439273\ttotal: 1m 7s\tremaining: 5m 48s\n",
      "326:\tlearn: 0.2439131\ttotal: 1m 8s\tremaining: 5m 48s\n",
      "327:\tlearn: 0.2439025\ttotal: 1m 8s\tremaining: 5m 48s\n",
      "328:\tlearn: 0.2438910\ttotal: 1m 8s\tremaining: 5m 48s\n",
      "329:\tlearn: 0.2438791\ttotal: 1m 8s\tremaining: 5m 47s\n",
      "330:\tlearn: 0.2438609\ttotal: 1m 8s\tremaining: 5m 47s\n",
      "331:\tlearn: 0.2438386\ttotal: 1m 9s\tremaining: 5m 47s\n",
      "332:\tlearn: 0.2438188\ttotal: 1m 9s\tremaining: 5m 47s\n",
      "333:\tlearn: 0.2438061\ttotal: 1m 9s\tremaining: 5m 47s\n",
      "334:\tlearn: 0.2437940\ttotal: 1m 9s\tremaining: 5m 47s\n",
      "335:\tlearn: 0.2437849\ttotal: 1m 10s\tremaining: 5m 46s\n",
      "336:\tlearn: 0.2437745\ttotal: 1m 10s\tremaining: 5m 46s\n",
      "337:\tlearn: 0.2437556\ttotal: 1m 10s\tremaining: 5m 46s\n",
      "338:\tlearn: 0.2437360\ttotal: 1m 10s\tremaining: 5m 46s\n",
      "339:\tlearn: 0.2437218\ttotal: 1m 10s\tremaining: 5m 46s\n",
      "340:\tlearn: 0.2437104\ttotal: 1m 11s\tremaining: 5m 45s\n",
      "341:\tlearn: 0.2436964\ttotal: 1m 11s\tremaining: 5m 45s\n",
      "342:\tlearn: 0.2436747\ttotal: 1m 11s\tremaining: 5m 45s\n",
      "343:\tlearn: 0.2436593\ttotal: 1m 11s\tremaining: 5m 45s\n",
      "344:\tlearn: 0.2436498\ttotal: 1m 11s\tremaining: 5m 44s\n",
      "345:\tlearn: 0.2436374\ttotal: 1m 12s\tremaining: 5m 44s\n",
      "346:\tlearn: 0.2436264\ttotal: 1m 12s\tremaining: 5m 44s\n",
      "347:\tlearn: 0.2436138\ttotal: 1m 12s\tremaining: 5m 44s\n",
      "348:\tlearn: 0.2436024\ttotal: 1m 12s\tremaining: 5m 44s\n",
      "349:\tlearn: 0.2435801\ttotal: 1m 12s\tremaining: 5m 43s\n",
      "350:\tlearn: 0.2435674\ttotal: 1m 13s\tremaining: 5m 43s\n",
      "351:\tlearn: 0.2435590\ttotal: 1m 13s\tremaining: 5m 43s\n",
      "352:\tlearn: 0.2435451\ttotal: 1m 13s\tremaining: 5m 43s\n",
      "353:\tlearn: 0.2435365\ttotal: 1m 13s\tremaining: 5m 43s\n",
      "354:\tlearn: 0.2435224\ttotal: 1m 14s\tremaining: 5m 43s\n",
      "355:\tlearn: 0.2435070\ttotal: 1m 14s\tremaining: 5m 42s\n",
      "356:\tlearn: 0.2434823\ttotal: 1m 14s\tremaining: 5m 42s\n",
      "357:\tlearn: 0.2434624\ttotal: 1m 14s\tremaining: 5m 42s\n",
      "358:\tlearn: 0.2434470\ttotal: 1m 14s\tremaining: 5m 42s\n",
      "359:\tlearn: 0.2434265\ttotal: 1m 15s\tremaining: 5m 42s\n",
      "360:\tlearn: 0.2434124\ttotal: 1m 15s\tremaining: 5m 41s\n",
      "361:\tlearn: 0.2434001\ttotal: 1m 15s\tremaining: 5m 41s\n",
      "362:\tlearn: 0.2433811\ttotal: 1m 15s\tremaining: 5m 41s\n",
      "363:\tlearn: 0.2433673\ttotal: 1m 15s\tremaining: 5m 41s\n",
      "364:\tlearn: 0.2433522\ttotal: 1m 16s\tremaining: 5m 41s\n",
      "365:\tlearn: 0.2433362\ttotal: 1m 16s\tremaining: 5m 40s\n",
      "366:\tlearn: 0.2433189\ttotal: 1m 16s\tremaining: 5m 40s\n",
      "367:\tlearn: 0.2433077\ttotal: 1m 16s\tremaining: 5m 40s\n",
      "368:\tlearn: 0.2432930\ttotal: 1m 17s\tremaining: 5m 40s\n",
      "369:\tlearn: 0.2432790\ttotal: 1m 17s\tremaining: 5m 40s\n",
      "370:\tlearn: 0.2432582\ttotal: 1m 17s\tremaining: 5m 40s\n",
      "371:\tlearn: 0.2432465\ttotal: 1m 17s\tremaining: 5m 39s\n",
      "372:\tlearn: 0.2432291\ttotal: 1m 17s\tremaining: 5m 39s\n",
      "373:\tlearn: 0.2432085\ttotal: 1m 18s\tremaining: 5m 39s\n",
      "374:\tlearn: 0.2431916\ttotal: 1m 18s\tremaining: 5m 39s\n",
      "375:\tlearn: 0.2431778\ttotal: 1m 18s\tremaining: 5m 39s\n",
      "376:\tlearn: 0.2431644\ttotal: 1m 18s\tremaining: 5m 38s\n",
      "377:\tlearn: 0.2431557\ttotal: 1m 18s\tremaining: 5m 38s\n",
      "378:\tlearn: 0.2431365\ttotal: 1m 19s\tremaining: 5m 38s\n",
      "379:\tlearn: 0.2431201\ttotal: 1m 19s\tremaining: 5m 38s\n",
      "380:\tlearn: 0.2431011\ttotal: 1m 19s\tremaining: 5m 38s\n",
      "381:\tlearn: 0.2430885\ttotal: 1m 19s\tremaining: 5m 37s\n",
      "382:\tlearn: 0.2430728\ttotal: 1m 19s\tremaining: 5m 37s\n",
      "383:\tlearn: 0.2430618\ttotal: 1m 20s\tremaining: 5m 37s\n",
      "384:\tlearn: 0.2430471\ttotal: 1m 20s\tremaining: 5m 37s\n",
      "385:\tlearn: 0.2430353\ttotal: 1m 20s\tremaining: 5m 37s\n",
      "386:\tlearn: 0.2430240\ttotal: 1m 20s\tremaining: 5m 36s\n",
      "387:\tlearn: 0.2430092\ttotal: 1m 21s\tremaining: 5m 36s\n",
      "388:\tlearn: 0.2429979\ttotal: 1m 21s\tremaining: 5m 36s\n",
      "389:\tlearn: 0.2429794\ttotal: 1m 21s\tremaining: 5m 36s\n",
      "390:\tlearn: 0.2429688\ttotal: 1m 21s\tremaining: 5m 36s\n",
      "391:\tlearn: 0.2429566\ttotal: 1m 21s\tremaining: 5m 35s\n",
      "392:\tlearn: 0.2429433\ttotal: 1m 22s\tremaining: 5m 35s\n",
      "393:\tlearn: 0.2429307\ttotal: 1m 22s\tremaining: 5m 35s\n",
      "394:\tlearn: 0.2429163\ttotal: 1m 22s\tremaining: 5m 35s\n",
      "395:\tlearn: 0.2428986\ttotal: 1m 22s\tremaining: 5m 35s\n",
      "396:\tlearn: 0.2428833\ttotal: 1m 22s\tremaining: 5m 34s\n",
      "397:\tlearn: 0.2428732\ttotal: 1m 23s\tremaining: 5m 34s\n",
      "398:\tlearn: 0.2428607\ttotal: 1m 23s\tremaining: 5m 34s\n",
      "399:\tlearn: 0.2428484\ttotal: 1m 23s\tremaining: 5m 34s\n",
      "400:\tlearn: 0.2428282\ttotal: 1m 23s\tremaining: 5m 34s\n",
      "401:\tlearn: 0.2428130\ttotal: 1m 23s\tremaining: 5m 33s\n",
      "402:\tlearn: 0.2428039\ttotal: 1m 24s\tremaining: 5m 33s\n",
      "403:\tlearn: 0.2427913\ttotal: 1m 24s\tremaining: 5m 33s\n",
      "404:\tlearn: 0.2427801\ttotal: 1m 24s\tremaining: 5m 33s\n",
      "405:\tlearn: 0.2427686\ttotal: 1m 24s\tremaining: 5m 33s\n",
      "406:\tlearn: 0.2427603\ttotal: 1m 25s\tremaining: 5m 32s\n",
      "407:\tlearn: 0.2427476\ttotal: 1m 25s\tremaining: 5m 32s\n",
      "408:\tlearn: 0.2427343\ttotal: 1m 25s\tremaining: 5m 32s\n",
      "409:\tlearn: 0.2427212\ttotal: 1m 25s\tremaining: 5m 32s\n",
      "410:\tlearn: 0.2427066\ttotal: 1m 25s\tremaining: 5m 32s\n",
      "411:\tlearn: 0.2426937\ttotal: 1m 26s\tremaining: 5m 31s\n",
      "412:\tlearn: 0.2426857\ttotal: 1m 26s\tremaining: 5m 31s\n",
      "413:\tlearn: 0.2426756\ttotal: 1m 26s\tremaining: 5m 31s\n",
      "414:\tlearn: 0.2426602\ttotal: 1m 26s\tremaining: 5m 31s\n",
      "415:\tlearn: 0.2426499\ttotal: 1m 26s\tremaining: 5m 31s\n",
      "416:\tlearn: 0.2426381\ttotal: 1m 27s\tremaining: 5m 30s\n",
      "417:\tlearn: 0.2426248\ttotal: 1m 27s\tremaining: 5m 30s\n",
      "418:\tlearn: 0.2426098\ttotal: 1m 27s\tremaining: 5m 30s\n",
      "419:\tlearn: 0.2425948\ttotal: 1m 27s\tremaining: 5m 30s\n",
      "420:\tlearn: 0.2425802\ttotal: 1m 27s\tremaining: 5m 30s\n",
      "421:\tlearn: 0.2425686\ttotal: 1m 28s\tremaining: 5m 29s\n",
      "422:\tlearn: 0.2425504\ttotal: 1m 28s\tremaining: 5m 29s\n",
      "423:\tlearn: 0.2425418\ttotal: 1m 28s\tremaining: 5m 29s\n",
      "424:\tlearn: 0.2425212\ttotal: 1m 28s\tremaining: 5m 29s\n",
      "425:\tlearn: 0.2425011\ttotal: 1m 29s\tremaining: 5m 29s\n",
      "426:\tlearn: 0.2424894\ttotal: 1m 29s\tremaining: 5m 28s\n",
      "427:\tlearn: 0.2424743\ttotal: 1m 29s\tremaining: 5m 28s\n",
      "428:\tlearn: 0.2424617\ttotal: 1m 29s\tremaining: 5m 28s\n",
      "429:\tlearn: 0.2424411\ttotal: 1m 29s\tremaining: 5m 28s\n",
      "430:\tlearn: 0.2424285\ttotal: 1m 30s\tremaining: 5m 28s\n",
      "431:\tlearn: 0.2424132\ttotal: 1m 30s\tremaining: 5m 27s\n",
      "432:\tlearn: 0.2423958\ttotal: 1m 30s\tremaining: 5m 27s\n",
      "433:\tlearn: 0.2423749\ttotal: 1m 30s\tremaining: 5m 27s\n",
      "434:\tlearn: 0.2423582\ttotal: 1m 30s\tremaining: 5m 27s\n",
      "435:\tlearn: 0.2423483\ttotal: 1m 31s\tremaining: 5m 27s\n",
      "436:\tlearn: 0.2423343\ttotal: 1m 31s\tremaining: 5m 26s\n",
      "437:\tlearn: 0.2423232\ttotal: 1m 31s\tremaining: 5m 26s\n",
      "438:\tlearn: 0.2423142\ttotal: 1m 31s\tremaining: 5m 26s\n",
      "439:\tlearn: 0.2423051\ttotal: 1m 32s\tremaining: 5m 26s\n",
      "440:\tlearn: 0.2422987\ttotal: 1m 32s\tremaining: 5m 26s\n",
      "441:\tlearn: 0.2422842\ttotal: 1m 32s\tremaining: 5m 25s\n",
      "442:\tlearn: 0.2422742\ttotal: 1m 32s\tremaining: 5m 25s\n",
      "443:\tlearn: 0.2422636\ttotal: 1m 32s\tremaining: 5m 25s\n",
      "444:\tlearn: 0.2422487\ttotal: 1m 33s\tremaining: 5m 25s\n",
      "445:\tlearn: 0.2422376\ttotal: 1m 33s\tremaining: 5m 25s\n",
      "446:\tlearn: 0.2422279\ttotal: 1m 33s\tremaining: 5m 24s\n",
      "447:\tlearn: 0.2422098\ttotal: 1m 33s\tremaining: 5m 24s\n",
      "448:\tlearn: 0.2421977\ttotal: 1m 33s\tremaining: 5m 24s\n",
      "449:\tlearn: 0.2421862\ttotal: 1m 34s\tremaining: 5m 24s\n",
      "450:\tlearn: 0.2421749\ttotal: 1m 34s\tremaining: 5m 24s\n",
      "451:\tlearn: 0.2421604\ttotal: 1m 34s\tremaining: 5m 23s\n",
      "452:\tlearn: 0.2421503\ttotal: 1m 34s\tremaining: 5m 23s\n",
      "453:\tlearn: 0.2421397\ttotal: 1m 34s\tremaining: 5m 23s\n",
      "454:\tlearn: 0.2421277\ttotal: 1m 35s\tremaining: 5m 23s\n",
      "455:\tlearn: 0.2421082\ttotal: 1m 35s\tremaining: 5m 23s\n",
      "456:\tlearn: 0.2420929\ttotal: 1m 35s\tremaining: 5m 22s\n",
      "457:\tlearn: 0.2420797\ttotal: 1m 35s\tremaining: 5m 22s\n",
      "458:\tlearn: 0.2420664\ttotal: 1m 36s\tremaining: 5m 22s\n",
      "459:\tlearn: 0.2420546\ttotal: 1m 36s\tremaining: 5m 22s\n",
      "460:\tlearn: 0.2420464\ttotal: 1m 36s\tremaining: 5m 22s\n",
      "461:\tlearn: 0.2420343\ttotal: 1m 36s\tremaining: 5m 21s\n",
      "462:\tlearn: 0.2420173\ttotal: 1m 36s\tremaining: 5m 21s\n",
      "463:\tlearn: 0.2420034\ttotal: 1m 37s\tremaining: 5m 21s\n",
      "464:\tlearn: 0.2419976\ttotal: 1m 37s\tremaining: 5m 21s\n",
      "465:\tlearn: 0.2419889\ttotal: 1m 37s\tremaining: 5m 21s\n",
      "466:\tlearn: 0.2419705\ttotal: 1m 37s\tremaining: 5m 20s\n",
      "467:\tlearn: 0.2419572\ttotal: 1m 37s\tremaining: 5m 20s\n",
      "468:\tlearn: 0.2419417\ttotal: 1m 38s\tremaining: 5m 20s\n",
      "469:\tlearn: 0.2419288\ttotal: 1m 38s\tremaining: 5m 20s\n",
      "470:\tlearn: 0.2419134\ttotal: 1m 38s\tremaining: 5m 20s\n",
      "471:\tlearn: 0.2419062\ttotal: 1m 38s\tremaining: 5m 19s\n",
      "472:\tlearn: 0.2418938\ttotal: 1m 39s\tremaining: 5m 19s\n",
      "473:\tlearn: 0.2418803\ttotal: 1m 39s\tremaining: 5m 19s\n",
      "474:\tlearn: 0.2418676\ttotal: 1m 39s\tremaining: 5m 19s\n",
      "475:\tlearn: 0.2418565\ttotal: 1m 39s\tremaining: 5m 19s\n",
      "476:\tlearn: 0.2418379\ttotal: 1m 39s\tremaining: 5m 18s\n",
      "477:\tlearn: 0.2418278\ttotal: 1m 40s\tremaining: 5m 18s\n",
      "478:\tlearn: 0.2418177\ttotal: 1m 40s\tremaining: 5m 18s\n",
      "479:\tlearn: 0.2418018\ttotal: 1m 40s\tremaining: 5m 18s\n",
      "480:\tlearn: 0.2417915\ttotal: 1m 40s\tremaining: 5m 18s\n",
      "481:\tlearn: 0.2417774\ttotal: 1m 40s\tremaining: 5m 17s\n",
      "482:\tlearn: 0.2417668\ttotal: 1m 41s\tremaining: 5m 17s\n",
      "483:\tlearn: 0.2417486\ttotal: 1m 41s\tremaining: 5m 17s\n",
      "484:\tlearn: 0.2417372\ttotal: 1m 41s\tremaining: 5m 17s\n",
      "485:\tlearn: 0.2417171\ttotal: 1m 41s\tremaining: 5m 17s\n",
      "486:\tlearn: 0.2417037\ttotal: 1m 42s\tremaining: 5m 16s\n",
      "487:\tlearn: 0.2416914\ttotal: 1m 42s\tremaining: 5m 16s\n",
      "488:\tlearn: 0.2416751\ttotal: 1m 42s\tremaining: 5m 16s\n",
      "489:\tlearn: 0.2416666\ttotal: 1m 42s\tremaining: 5m 16s\n",
      "490:\tlearn: 0.2416480\ttotal: 1m 42s\tremaining: 5m 16s\n",
      "491:\tlearn: 0.2416370\ttotal: 1m 43s\tremaining: 5m 15s\n",
      "492:\tlearn: 0.2416238\ttotal: 1m 43s\tremaining: 5m 15s\n",
      "493:\tlearn: 0.2416106\ttotal: 1m 43s\tremaining: 5m 15s\n",
      "494:\tlearn: 0.2415997\ttotal: 1m 43s\tremaining: 5m 15s\n",
      "495:\tlearn: 0.2415913\ttotal: 1m 43s\tremaining: 5m 15s\n",
      "496:\tlearn: 0.2415733\ttotal: 1m 44s\tremaining: 5m 14s\n",
      "497:\tlearn: 0.2415555\ttotal: 1m 44s\tremaining: 5m 14s\n",
      "498:\tlearn: 0.2415426\ttotal: 1m 44s\tremaining: 5m 14s\n",
      "499:\tlearn: 0.2415328\ttotal: 1m 44s\tremaining: 5m 14s\n",
      "500:\tlearn: 0.2415186\ttotal: 1m 44s\tremaining: 5m 14s\n",
      "501:\tlearn: 0.2415047\ttotal: 1m 45s\tremaining: 5m 13s\n",
      "502:\tlearn: 0.2414963\ttotal: 1m 45s\tremaining: 5m 13s\n",
      "503:\tlearn: 0.2414845\ttotal: 1m 45s\tremaining: 5m 13s\n",
      "504:\tlearn: 0.2414728\ttotal: 1m 45s\tremaining: 5m 13s\n",
      "505:\tlearn: 0.2414605\ttotal: 1m 46s\tremaining: 5m 13s\n",
      "506:\tlearn: 0.2414483\ttotal: 1m 46s\tremaining: 5m 12s\n",
      "507:\tlearn: 0.2414326\ttotal: 1m 46s\tremaining: 5m 12s\n",
      "508:\tlearn: 0.2414226\ttotal: 1m 46s\tremaining: 5m 12s\n",
      "509:\tlearn: 0.2414103\ttotal: 1m 46s\tremaining: 5m 12s\n",
      "510:\tlearn: 0.2413969\ttotal: 1m 47s\tremaining: 5m 12s\n",
      "511:\tlearn: 0.2413811\ttotal: 1m 47s\tremaining: 5m 11s\n",
      "512:\tlearn: 0.2413663\ttotal: 1m 47s\tremaining: 5m 11s\n",
      "513:\tlearn: 0.2413510\ttotal: 1m 47s\tremaining: 5m 11s\n",
      "514:\tlearn: 0.2413362\ttotal: 1m 47s\tremaining: 5m 11s\n",
      "515:\tlearn: 0.2413263\ttotal: 1m 48s\tremaining: 5m 11s\n",
      "516:\tlearn: 0.2413063\ttotal: 1m 48s\tremaining: 5m 10s\n",
      "517:\tlearn: 0.2412881\ttotal: 1m 48s\tremaining: 5m 10s\n",
      "518:\tlearn: 0.2412738\ttotal: 1m 48s\tremaining: 5m 10s\n",
      "519:\tlearn: 0.2412617\ttotal: 1m 49s\tremaining: 5m 10s\n",
      "520:\tlearn: 0.2412412\ttotal: 1m 49s\tremaining: 5m 10s\n",
      "521:\tlearn: 0.2412249\ttotal: 1m 49s\tremaining: 5m 9s\n",
      "522:\tlearn: 0.2412068\ttotal: 1m 49s\tremaining: 5m 9s\n",
      "523:\tlearn: 0.2411874\ttotal: 1m 49s\tremaining: 5m 9s\n",
      "524:\tlearn: 0.2411690\ttotal: 1m 50s\tremaining: 5m 9s\n",
      "525:\tlearn: 0.2411580\ttotal: 1m 50s\tremaining: 5m 9s\n",
      "526:\tlearn: 0.2411426\ttotal: 1m 50s\tremaining: 5m 8s\n",
      "527:\tlearn: 0.2411236\ttotal: 1m 50s\tremaining: 5m 8s\n",
      "528:\tlearn: 0.2411137\ttotal: 1m 50s\tremaining: 5m 8s\n",
      "529:\tlearn: 0.2411018\ttotal: 1m 51s\tremaining: 5m 8s\n",
      "530:\tlearn: 0.2410858\ttotal: 1m 51s\tremaining: 5m 8s\n",
      "531:\tlearn: 0.2410728\ttotal: 1m 51s\tremaining: 5m 7s\n",
      "532:\tlearn: 0.2410587\ttotal: 1m 51s\tremaining: 5m 7s\n",
      "533:\tlearn: 0.2410507\ttotal: 1m 51s\tremaining: 5m 7s\n",
      "534:\tlearn: 0.2410384\ttotal: 1m 52s\tremaining: 5m 7s\n",
      "535:\tlearn: 0.2410231\ttotal: 1m 52s\tremaining: 5m 7s\n",
      "536:\tlearn: 0.2410037\ttotal: 1m 52s\tremaining: 5m 6s\n",
      "537:\tlearn: 0.2409892\ttotal: 1m 52s\tremaining: 5m 6s\n",
      "538:\tlearn: 0.2409808\ttotal: 1m 53s\tremaining: 5m 6s\n",
      "539:\tlearn: 0.2409633\ttotal: 1m 53s\tremaining: 5m 6s\n",
      "540:\tlearn: 0.2409459\ttotal: 1m 53s\tremaining: 5m 6s\n",
      "541:\tlearn: 0.2409298\ttotal: 1m 53s\tremaining: 5m 5s\n",
      "542:\tlearn: 0.2409177\ttotal: 1m 53s\tremaining: 5m 5s\n",
      "543:\tlearn: 0.2409048\ttotal: 1m 54s\tremaining: 5m 5s\n",
      "544:\tlearn: 0.2408964\ttotal: 1m 54s\tremaining: 5m 5s\n",
      "545:\tlearn: 0.2408807\ttotal: 1m 54s\tremaining: 5m 4s\n",
      "546:\tlearn: 0.2408694\ttotal: 1m 54s\tremaining: 5m 4s\n",
      "547:\tlearn: 0.2408530\ttotal: 1m 54s\tremaining: 5m 4s\n",
      "548:\tlearn: 0.2408372\ttotal: 1m 55s\tremaining: 5m 4s\n",
      "549:\tlearn: 0.2408282\ttotal: 1m 55s\tremaining: 5m 4s\n",
      "550:\tlearn: 0.2408171\ttotal: 1m 55s\tremaining: 5m 3s\n",
      "551:\tlearn: 0.2408024\ttotal: 1m 55s\tremaining: 5m 3s\n",
      "552:\tlearn: 0.2407857\ttotal: 1m 55s\tremaining: 5m 3s\n",
      "553:\tlearn: 0.2407710\ttotal: 1m 56s\tremaining: 5m 3s\n",
      "554:\tlearn: 0.2407539\ttotal: 1m 56s\tremaining: 5m 3s\n",
      "555:\tlearn: 0.2407437\ttotal: 1m 56s\tremaining: 5m 2s\n",
      "556:\tlearn: 0.2407291\ttotal: 1m 56s\tremaining: 5m 2s\n",
      "557:\tlearn: 0.2407185\ttotal: 1m 57s\tremaining: 5m 2s\n",
      "558:\tlearn: 0.2407087\ttotal: 1m 57s\tremaining: 5m 2s\n",
      "559:\tlearn: 0.2406976\ttotal: 1m 57s\tremaining: 5m 2s\n",
      "560:\tlearn: 0.2406850\ttotal: 1m 57s\tremaining: 5m 1s\n",
      "561:\tlearn: 0.2406778\ttotal: 1m 57s\tremaining: 5m 1s\n",
      "562:\tlearn: 0.2406643\ttotal: 1m 58s\tremaining: 5m 1s\n",
      "563:\tlearn: 0.2406453\ttotal: 1m 58s\tremaining: 5m 1s\n",
      "564:\tlearn: 0.2406300\ttotal: 1m 58s\tremaining: 5m 1s\n",
      "565:\tlearn: 0.2406173\ttotal: 1m 58s\tremaining: 5m\n",
      "566:\tlearn: 0.2406058\ttotal: 1m 58s\tremaining: 5m\n",
      "567:\tlearn: 0.2405944\ttotal: 1m 59s\tremaining: 5m\n",
      "568:\tlearn: 0.2405801\ttotal: 1m 59s\tremaining: 5m\n",
      "569:\tlearn: 0.2405705\ttotal: 1m 59s\tremaining: 4m 59s\n",
      "570:\tlearn: 0.2405555\ttotal: 1m 59s\tremaining: 4m 59s\n",
      "571:\tlearn: 0.2405381\ttotal: 1m 59s\tremaining: 4m 59s\n",
      "572:\tlearn: 0.2405210\ttotal: 2m\tremaining: 4m 59s\n",
      "573:\tlearn: 0.2405066\ttotal: 2m\tremaining: 4m 59s\n",
      "574:\tlearn: 0.2404976\ttotal: 2m\tremaining: 4m 58s\n",
      "575:\tlearn: 0.2404886\ttotal: 2m\tremaining: 4m 58s\n",
      "576:\tlearn: 0.2404756\ttotal: 2m 1s\tremaining: 4m 58s\n",
      "577:\tlearn: 0.2404578\ttotal: 2m 1s\tremaining: 4m 58s\n",
      "578:\tlearn: 0.2404387\ttotal: 2m 1s\tremaining: 4m 58s\n",
      "579:\tlearn: 0.2404301\ttotal: 2m 1s\tremaining: 4m 57s\n",
      "580:\tlearn: 0.2404191\ttotal: 2m 1s\tremaining: 4m 57s\n",
      "581:\tlearn: 0.2404117\ttotal: 2m 2s\tremaining: 4m 57s\n",
      "582:\tlearn: 0.2404034\ttotal: 2m 2s\tremaining: 4m 57s\n",
      "583:\tlearn: 0.2403888\ttotal: 2m 2s\tremaining: 4m 57s\n",
      "584:\tlearn: 0.2403779\ttotal: 2m 2s\tremaining: 4m 56s\n",
      "585:\tlearn: 0.2403664\ttotal: 2m 2s\tremaining: 4m 56s\n",
      "586:\tlearn: 0.2403581\ttotal: 2m 3s\tremaining: 4m 56s\n",
      "587:\tlearn: 0.2403439\ttotal: 2m 3s\tremaining: 4m 56s\n",
      "588:\tlearn: 0.2403296\ttotal: 2m 3s\tremaining: 4m 55s\n",
      "589:\tlearn: 0.2403163\ttotal: 2m 3s\tremaining: 4m 55s\n",
      "590:\tlearn: 0.2403011\ttotal: 2m 3s\tremaining: 4m 55s\n",
      "591:\tlearn: 0.2402884\ttotal: 2m 4s\tremaining: 4m 55s\n",
      "592:\tlearn: 0.2402758\ttotal: 2m 4s\tremaining: 4m 55s\n",
      "593:\tlearn: 0.2402619\ttotal: 2m 4s\tremaining: 4m 54s\n",
      "594:\tlearn: 0.2402525\ttotal: 2m 4s\tremaining: 4m 54s\n",
      "595:\tlearn: 0.2402359\ttotal: 2m 5s\tremaining: 4m 54s\n",
      "596:\tlearn: 0.2402288\ttotal: 2m 5s\tremaining: 4m 54s\n",
      "597:\tlearn: 0.2402221\ttotal: 2m 5s\tremaining: 4m 54s\n",
      "598:\tlearn: 0.2402063\ttotal: 2m 5s\tremaining: 4m 53s\n",
      "599:\tlearn: 0.2401970\ttotal: 2m 5s\tremaining: 4m 53s\n",
      "600:\tlearn: 0.2401821\ttotal: 2m 6s\tremaining: 4m 53s\n",
      "601:\tlearn: 0.2401684\ttotal: 2m 6s\tremaining: 4m 53s\n",
      "602:\tlearn: 0.2401612\ttotal: 2m 6s\tremaining: 4m 53s\n",
      "603:\tlearn: 0.2401464\ttotal: 2m 6s\tremaining: 4m 52s\n",
      "604:\tlearn: 0.2401356\ttotal: 2m 6s\tremaining: 4m 52s\n",
      "605:\tlearn: 0.2401182\ttotal: 2m 7s\tremaining: 4m 52s\n",
      "606:\tlearn: 0.2401047\ttotal: 2m 7s\tremaining: 4m 52s\n",
      "607:\tlearn: 0.2400929\ttotal: 2m 7s\tremaining: 4m 52s\n",
      "608:\tlearn: 0.2400791\ttotal: 2m 7s\tremaining: 4m 51s\n",
      "609:\tlearn: 0.2400721\ttotal: 2m 7s\tremaining: 4m 51s\n",
      "610:\tlearn: 0.2400595\ttotal: 2m 8s\tremaining: 4m 51s\n",
      "611:\tlearn: 0.2400414\ttotal: 2m 8s\tremaining: 4m 51s\n",
      "612:\tlearn: 0.2400310\ttotal: 2m 8s\tremaining: 4m 51s\n",
      "613:\tlearn: 0.2400184\ttotal: 2m 8s\tremaining: 4m 50s\n",
      "614:\tlearn: 0.2400036\ttotal: 2m 9s\tremaining: 4m 50s\n",
      "615:\tlearn: 0.2399938\ttotal: 2m 9s\tremaining: 4m 50s\n",
      "616:\tlearn: 0.2399830\ttotal: 2m 9s\tremaining: 4m 50s\n",
      "617:\tlearn: 0.2399645\ttotal: 2m 9s\tremaining: 4m 50s\n",
      "618:\tlearn: 0.2399553\ttotal: 2m 9s\tremaining: 4m 49s\n",
      "619:\tlearn: 0.2399396\ttotal: 2m 10s\tremaining: 4m 49s\n",
      "620:\tlearn: 0.2399232\ttotal: 2m 10s\tremaining: 4m 49s\n",
      "621:\tlearn: 0.2399138\ttotal: 2m 10s\tremaining: 4m 49s\n",
      "622:\tlearn: 0.2399034\ttotal: 2m 10s\tremaining: 4m 49s\n",
      "623:\tlearn: 0.2398902\ttotal: 2m 10s\tremaining: 4m 48s\n",
      "624:\tlearn: 0.2398761\ttotal: 2m 11s\tremaining: 4m 48s\n",
      "625:\tlearn: 0.2398643\ttotal: 2m 11s\tremaining: 4m 48s\n",
      "626:\tlearn: 0.2398579\ttotal: 2m 11s\tremaining: 4m 48s\n",
      "627:\tlearn: 0.2398424\ttotal: 2m 11s\tremaining: 4m 48s\n",
      "628:\tlearn: 0.2398329\ttotal: 2m 12s\tremaining: 4m 47s\n",
      "629:\tlearn: 0.2398260\ttotal: 2m 12s\tremaining: 4m 47s\n",
      "630:\tlearn: 0.2398142\ttotal: 2m 12s\tremaining: 4m 47s\n",
      "631:\tlearn: 0.2397980\ttotal: 2m 12s\tremaining: 4m 47s\n",
      "632:\tlearn: 0.2397863\ttotal: 2m 12s\tremaining: 4m 46s\n",
      "633:\tlearn: 0.2397781\ttotal: 2m 13s\tremaining: 4m 46s\n",
      "634:\tlearn: 0.2397637\ttotal: 2m 13s\tremaining: 4m 46s\n",
      "635:\tlearn: 0.2397443\ttotal: 2m 13s\tremaining: 4m 46s\n",
      "636:\tlearn: 0.2397270\ttotal: 2m 13s\tremaining: 4m 46s\n",
      "637:\tlearn: 0.2397053\ttotal: 2m 13s\tremaining: 4m 45s\n",
      "638:\tlearn: 0.2396940\ttotal: 2m 14s\tremaining: 4m 45s\n",
      "639:\tlearn: 0.2396803\ttotal: 2m 14s\tremaining: 4m 45s\n",
      "640:\tlearn: 0.2396688\ttotal: 2m 14s\tremaining: 4m 45s\n",
      "641:\tlearn: 0.2396591\ttotal: 2m 14s\tremaining: 4m 45s\n",
      "642:\tlearn: 0.2396382\ttotal: 2m 15s\tremaining: 4m 44s\n",
      "643:\tlearn: 0.2396284\ttotal: 2m 15s\tremaining: 4m 44s\n",
      "644:\tlearn: 0.2396170\ttotal: 2m 15s\tremaining: 4m 44s\n",
      "645:\tlearn: 0.2396054\ttotal: 2m 15s\tremaining: 4m 44s\n",
      "646:\tlearn: 0.2395902\ttotal: 2m 15s\tremaining: 4m 44s\n",
      "647:\tlearn: 0.2395738\ttotal: 2m 16s\tremaining: 4m 43s\n",
      "648:\tlearn: 0.2395626\ttotal: 2m 16s\tremaining: 4m 43s\n",
      "649:\tlearn: 0.2395503\ttotal: 2m 16s\tremaining: 4m 43s\n",
      "650:\tlearn: 0.2395424\ttotal: 2m 16s\tremaining: 4m 43s\n",
      "651:\tlearn: 0.2395328\ttotal: 2m 16s\tremaining: 4m 43s\n",
      "652:\tlearn: 0.2395181\ttotal: 2m 17s\tremaining: 4m 42s\n",
      "653:\tlearn: 0.2395050\ttotal: 2m 17s\tremaining: 4m 42s\n",
      "654:\tlearn: 0.2394936\ttotal: 2m 17s\tremaining: 4m 42s\n",
      "655:\tlearn: 0.2394827\ttotal: 2m 17s\tremaining: 4m 42s\n",
      "656:\tlearn: 0.2394669\ttotal: 2m 18s\tremaining: 4m 42s\n",
      "657:\tlearn: 0.2394539\ttotal: 2m 18s\tremaining: 4m 41s\n",
      "658:\tlearn: 0.2394449\ttotal: 2m 18s\tremaining: 4m 41s\n",
      "659:\tlearn: 0.2394324\ttotal: 2m 18s\tremaining: 4m 41s\n",
      "660:\tlearn: 0.2394250\ttotal: 2m 18s\tremaining: 4m 41s\n",
      "661:\tlearn: 0.2394160\ttotal: 2m 19s\tremaining: 4m 41s\n",
      "662:\tlearn: 0.2394030\ttotal: 2m 19s\tremaining: 4m 40s\n",
      "663:\tlearn: 0.2393894\ttotal: 2m 19s\tremaining: 4m 40s\n",
      "664:\tlearn: 0.2393719\ttotal: 2m 19s\tremaining: 4m 40s\n",
      "665:\tlearn: 0.2393584\ttotal: 2m 19s\tremaining: 4m 40s\n",
      "666:\tlearn: 0.2393499\ttotal: 2m 20s\tremaining: 4m 40s\n",
      "667:\tlearn: 0.2393386\ttotal: 2m 20s\tremaining: 4m 39s\n",
      "668:\tlearn: 0.2393309\ttotal: 2m 20s\tremaining: 4m 39s\n",
      "669:\tlearn: 0.2393152\ttotal: 2m 20s\tremaining: 4m 39s\n",
      "670:\tlearn: 0.2393030\ttotal: 2m 20s\tremaining: 4m 39s\n",
      "671:\tlearn: 0.2392858\ttotal: 2m 21s\tremaining: 4m 39s\n",
      "672:\tlearn: 0.2392742\ttotal: 2m 21s\tremaining: 4m 38s\n",
      "673:\tlearn: 0.2392607\ttotal: 2m 21s\tremaining: 4m 38s\n",
      "674:\tlearn: 0.2392484\ttotal: 2m 21s\tremaining: 4m 38s\n",
      "675:\tlearn: 0.2392376\ttotal: 2m 22s\tremaining: 4m 38s\n",
      "676:\tlearn: 0.2392266\ttotal: 2m 22s\tremaining: 4m 38s\n",
      "677:\tlearn: 0.2392121\ttotal: 2m 22s\tremaining: 4m 37s\n",
      "678:\tlearn: 0.2391977\ttotal: 2m 22s\tremaining: 4m 37s\n",
      "679:\tlearn: 0.2391780\ttotal: 2m 22s\tremaining: 4m 37s\n",
      "680:\tlearn: 0.2391676\ttotal: 2m 23s\tremaining: 4m 37s\n",
      "681:\tlearn: 0.2391553\ttotal: 2m 23s\tremaining: 4m 36s\n",
      "682:\tlearn: 0.2391423\ttotal: 2m 23s\tremaining: 4m 36s\n",
      "683:\tlearn: 0.2391318\ttotal: 2m 23s\tremaining: 4m 36s\n",
      "684:\tlearn: 0.2391172\ttotal: 2m 23s\tremaining: 4m 36s\n",
      "685:\tlearn: 0.2391067\ttotal: 2m 24s\tremaining: 4m 36s\n",
      "686:\tlearn: 0.2390959\ttotal: 2m 24s\tremaining: 4m 35s\n",
      "687:\tlearn: 0.2390845\ttotal: 2m 24s\tremaining: 4m 35s\n",
      "688:\tlearn: 0.2390699\ttotal: 2m 24s\tremaining: 4m 35s\n",
      "689:\tlearn: 0.2390587\ttotal: 2m 25s\tremaining: 4m 35s\n",
      "690:\tlearn: 0.2390457\ttotal: 2m 25s\tremaining: 4m 35s\n",
      "691:\tlearn: 0.2390346\ttotal: 2m 25s\tremaining: 4m 34s\n",
      "692:\tlearn: 0.2390245\ttotal: 2m 25s\tremaining: 4m 34s\n",
      "693:\tlearn: 0.2390068\ttotal: 2m 25s\tremaining: 4m 34s\n",
      "694:\tlearn: 0.2389947\ttotal: 2m 26s\tremaining: 4m 34s\n",
      "695:\tlearn: 0.2389814\ttotal: 2m 26s\tremaining: 4m 34s\n",
      "696:\tlearn: 0.2389657\ttotal: 2m 26s\tremaining: 4m 33s\n",
      "697:\tlearn: 0.2389539\ttotal: 2m 26s\tremaining: 4m 33s\n",
      "698:\tlearn: 0.2389407\ttotal: 2m 26s\tremaining: 4m 33s\n",
      "699:\tlearn: 0.2389262\ttotal: 2m 27s\tremaining: 4m 33s\n",
      "700:\tlearn: 0.2389119\ttotal: 2m 27s\tremaining: 4m 33s\n",
      "701:\tlearn: 0.2389043\ttotal: 2m 27s\tremaining: 4m 32s\n",
      "702:\tlearn: 0.2388965\ttotal: 2m 27s\tremaining: 4m 32s\n",
      "703:\tlearn: 0.2388838\ttotal: 2m 27s\tremaining: 4m 32s\n",
      "704:\tlearn: 0.2388772\ttotal: 2m 28s\tremaining: 4m 32s\n",
      "705:\tlearn: 0.2388676\ttotal: 2m 28s\tremaining: 4m 31s\n",
      "706:\tlearn: 0.2388551\ttotal: 2m 28s\tremaining: 4m 31s\n",
      "707:\tlearn: 0.2388418\ttotal: 2m 28s\tremaining: 4m 31s\n",
      "708:\tlearn: 0.2388316\ttotal: 2m 29s\tremaining: 4m 31s\n",
      "709:\tlearn: 0.2388128\ttotal: 2m 29s\tremaining: 4m 31s\n",
      "710:\tlearn: 0.2387967\ttotal: 2m 29s\tremaining: 4m 30s\n",
      "711:\tlearn: 0.2387807\ttotal: 2m 29s\tremaining: 4m 30s\n",
      "712:\tlearn: 0.2387717\ttotal: 2m 29s\tremaining: 4m 30s\n",
      "713:\tlearn: 0.2387591\ttotal: 2m 30s\tremaining: 4m 30s\n",
      "714:\tlearn: 0.2387467\ttotal: 2m 30s\tremaining: 4m 30s\n",
      "715:\tlearn: 0.2387383\ttotal: 2m 30s\tremaining: 4m 29s\n",
      "716:\tlearn: 0.2387290\ttotal: 2m 30s\tremaining: 4m 29s\n",
      "717:\tlearn: 0.2387139\ttotal: 2m 30s\tremaining: 4m 29s\n",
      "718:\tlearn: 0.2387035\ttotal: 2m 31s\tremaining: 4m 29s\n",
      "719:\tlearn: 0.2386860\ttotal: 2m 31s\tremaining: 4m 29s\n",
      "720:\tlearn: 0.2386708\ttotal: 2m 31s\tremaining: 4m 28s\n",
      "721:\tlearn: 0.2386602\ttotal: 2m 31s\tremaining: 4m 28s\n",
      "722:\tlearn: 0.2386493\ttotal: 2m 32s\tremaining: 4m 28s\n",
      "723:\tlearn: 0.2386397\ttotal: 2m 32s\tremaining: 4m 28s\n",
      "724:\tlearn: 0.2386258\ttotal: 2m 32s\tremaining: 4m 28s\n",
      "725:\tlearn: 0.2386112\ttotal: 2m 32s\tremaining: 4m 27s\n",
      "726:\tlearn: 0.2386033\ttotal: 2m 32s\tremaining: 4m 27s\n",
      "727:\tlearn: 0.2385928\ttotal: 2m 33s\tremaining: 4m 27s\n",
      "728:\tlearn: 0.2385804\ttotal: 2m 33s\tremaining: 4m 27s\n",
      "729:\tlearn: 0.2385670\ttotal: 2m 33s\tremaining: 4m 27s\n",
      "730:\tlearn: 0.2385581\ttotal: 2m 33s\tremaining: 4m 26s\n",
      "731:\tlearn: 0.2385445\ttotal: 2m 33s\tremaining: 4m 26s\n",
      "732:\tlearn: 0.2385273\ttotal: 2m 34s\tremaining: 4m 26s\n",
      "733:\tlearn: 0.2385166\ttotal: 2m 34s\tremaining: 4m 26s\n",
      "734:\tlearn: 0.2385054\ttotal: 2m 34s\tremaining: 4m 26s\n",
      "735:\tlearn: 0.2384925\ttotal: 2m 34s\tremaining: 4m 25s\n",
      "736:\tlearn: 0.2384792\ttotal: 2m 35s\tremaining: 4m 25s\n",
      "737:\tlearn: 0.2384715\ttotal: 2m 35s\tremaining: 4m 25s\n",
      "738:\tlearn: 0.2384650\ttotal: 2m 35s\tremaining: 4m 25s\n",
      "739:\tlearn: 0.2384521\ttotal: 2m 35s\tremaining: 4m 24s\n",
      "740:\tlearn: 0.2384347\ttotal: 2m 35s\tremaining: 4m 24s\n",
      "741:\tlearn: 0.2384193\ttotal: 2m 36s\tremaining: 4m 24s\n",
      "742:\tlearn: 0.2384056\ttotal: 2m 36s\tremaining: 4m 24s\n",
      "743:\tlearn: 0.2383934\ttotal: 2m 36s\tremaining: 4m 24s\n",
      "744:\tlearn: 0.2383812\ttotal: 2m 36s\tremaining: 4m 23s\n",
      "745:\tlearn: 0.2383673\ttotal: 2m 36s\tremaining: 4m 23s\n",
      "746:\tlearn: 0.2383575\ttotal: 2m 37s\tremaining: 4m 23s\n",
      "747:\tlearn: 0.2383482\ttotal: 2m 37s\tremaining: 4m 23s\n",
      "748:\tlearn: 0.2383330\ttotal: 2m 37s\tremaining: 4m 23s\n",
      "749:\tlearn: 0.2383254\ttotal: 2m 37s\tremaining: 4m 22s\n",
      "750:\tlearn: 0.2383144\ttotal: 2m 37s\tremaining: 4m 22s\n",
      "751:\tlearn: 0.2382995\ttotal: 2m 38s\tremaining: 4m 22s\n",
      "752:\tlearn: 0.2382864\ttotal: 2m 38s\tremaining: 4m 22s\n",
      "753:\tlearn: 0.2382812\ttotal: 2m 38s\tremaining: 4m 22s\n",
      "754:\tlearn: 0.2382731\ttotal: 2m 38s\tremaining: 4m 21s\n",
      "755:\tlearn: 0.2382635\ttotal: 2m 39s\tremaining: 4m 21s\n",
      "756:\tlearn: 0.2382511\ttotal: 2m 39s\tremaining: 4m 21s\n",
      "757:\tlearn: 0.2382346\ttotal: 2m 39s\tremaining: 4m 21s\n",
      "758:\tlearn: 0.2382206\ttotal: 2m 39s\tremaining: 4m 21s\n",
      "759:\tlearn: 0.2382098\ttotal: 2m 39s\tremaining: 4m 20s\n",
      "760:\tlearn: 0.2382025\ttotal: 2m 40s\tremaining: 4m 20s\n",
      "761:\tlearn: 0.2381907\ttotal: 2m 40s\tremaining: 4m 20s\n",
      "762:\tlearn: 0.2381761\ttotal: 2m 40s\tremaining: 4m 20s\n",
      "763:\tlearn: 0.2381643\ttotal: 2m 40s\tremaining: 4m 20s\n",
      "764:\tlearn: 0.2381497\ttotal: 2m 40s\tremaining: 4m 19s\n",
      "765:\tlearn: 0.2381344\ttotal: 2m 41s\tremaining: 4m 19s\n",
      "766:\tlearn: 0.2381182\ttotal: 2m 41s\tremaining: 4m 19s\n",
      "767:\tlearn: 0.2381091\ttotal: 2m 41s\tremaining: 4m 19s\n",
      "768:\tlearn: 0.2380956\ttotal: 2m 41s\tremaining: 4m 18s\n",
      "769:\tlearn: 0.2380882\ttotal: 2m 42s\tremaining: 4m 18s\n",
      "770:\tlearn: 0.2380749\ttotal: 2m 42s\tremaining: 4m 18s\n",
      "771:\tlearn: 0.2380600\ttotal: 2m 42s\tremaining: 4m 18s\n",
      "772:\tlearn: 0.2380481\ttotal: 2m 42s\tremaining: 4m 18s\n",
      "773:\tlearn: 0.2380352\ttotal: 2m 42s\tremaining: 4m 17s\n",
      "774:\tlearn: 0.2380251\ttotal: 2m 43s\tremaining: 4m 17s\n",
      "775:\tlearn: 0.2380130\ttotal: 2m 43s\tremaining: 4m 17s\n",
      "776:\tlearn: 0.2380004\ttotal: 2m 43s\tremaining: 4m 17s\n",
      "777:\tlearn: 0.2379819\ttotal: 2m 43s\tremaining: 4m 17s\n",
      "778:\tlearn: 0.2379729\ttotal: 2m 43s\tremaining: 4m 16s\n",
      "779:\tlearn: 0.2379600\ttotal: 2m 44s\tremaining: 4m 16s\n",
      "780:\tlearn: 0.2379465\ttotal: 2m 44s\tremaining: 4m 16s\n",
      "781:\tlearn: 0.2379373\ttotal: 2m 44s\tremaining: 4m 16s\n",
      "782:\tlearn: 0.2379318\ttotal: 2m 44s\tremaining: 4m 16s\n",
      "783:\tlearn: 0.2379238\ttotal: 2m 45s\tremaining: 4m 15s\n",
      "784:\tlearn: 0.2379125\ttotal: 2m 45s\tremaining: 4m 15s\n",
      "785:\tlearn: 0.2378989\ttotal: 2m 45s\tremaining: 4m 15s\n",
      "786:\tlearn: 0.2378874\ttotal: 2m 45s\tremaining: 4m 15s\n",
      "787:\tlearn: 0.2378794\ttotal: 2m 45s\tremaining: 4m 15s\n",
      "788:\tlearn: 0.2378630\ttotal: 2m 46s\tremaining: 4m 14s\n",
      "789:\tlearn: 0.2378484\ttotal: 2m 46s\tremaining: 4m 14s\n",
      "790:\tlearn: 0.2378324\ttotal: 2m 46s\tremaining: 4m 14s\n",
      "791:\tlearn: 0.2378180\ttotal: 2m 46s\tremaining: 4m 14s\n",
      "792:\tlearn: 0.2378065\ttotal: 2m 46s\tremaining: 4m 14s\n",
      "793:\tlearn: 0.2377944\ttotal: 2m 47s\tremaining: 4m 13s\n",
      "794:\tlearn: 0.2377775\ttotal: 2m 47s\tremaining: 4m 13s\n",
      "795:\tlearn: 0.2377706\ttotal: 2m 47s\tremaining: 4m 13s\n",
      "796:\tlearn: 0.2377586\ttotal: 2m 47s\tremaining: 4m 13s\n",
      "797:\tlearn: 0.2377438\ttotal: 2m 47s\tremaining: 4m 13s\n",
      "798:\tlearn: 0.2377297\ttotal: 2m 48s\tremaining: 4m 12s\n",
      "799:\tlearn: 0.2377180\ttotal: 2m 48s\tremaining: 4m 12s\n",
      "800:\tlearn: 0.2376999\ttotal: 2m 48s\tremaining: 4m 12s\n",
      "801:\tlearn: 0.2376903\ttotal: 2m 48s\tremaining: 4m 12s\n",
      "802:\tlearn: 0.2376753\ttotal: 2m 49s\tremaining: 4m 12s\n",
      "803:\tlearn: 0.2376616\ttotal: 2m 49s\tremaining: 4m 11s\n",
      "804:\tlearn: 0.2376509\ttotal: 2m 49s\tremaining: 4m 11s\n",
      "805:\tlearn: 0.2376419\ttotal: 2m 49s\tremaining: 4m 11s\n",
      "806:\tlearn: 0.2376351\ttotal: 2m 49s\tremaining: 4m 11s\n",
      "807:\tlearn: 0.2376225\ttotal: 2m 50s\tremaining: 4m 10s\n",
      "808:\tlearn: 0.2376160\ttotal: 2m 50s\tremaining: 4m 10s\n",
      "809:\tlearn: 0.2376019\ttotal: 2m 50s\tremaining: 4m 10s\n",
      "810:\tlearn: 0.2375864\ttotal: 2m 50s\tremaining: 4m 10s\n",
      "811:\tlearn: 0.2375789\ttotal: 2m 50s\tremaining: 4m 10s\n",
      "812:\tlearn: 0.2375652\ttotal: 2m 51s\tremaining: 4m 9s\n",
      "813:\tlearn: 0.2375519\ttotal: 2m 51s\tremaining: 4m 9s\n",
      "814:\tlearn: 0.2375421\ttotal: 2m 51s\tremaining: 4m 9s\n",
      "815:\tlearn: 0.2375318\ttotal: 2m 51s\tremaining: 4m 9s\n",
      "816:\tlearn: 0.2375210\ttotal: 2m 52s\tremaining: 4m 9s\n",
      "817:\tlearn: 0.2375106\ttotal: 2m 52s\tremaining: 4m 8s\n",
      "818:\tlearn: 0.2375025\ttotal: 2m 52s\tremaining: 4m 8s\n",
      "819:\tlearn: 0.2374956\ttotal: 2m 52s\tremaining: 4m 8s\n",
      "820:\tlearn: 0.2374817\ttotal: 2m 52s\tremaining: 4m 8s\n",
      "821:\tlearn: 0.2374618\ttotal: 2m 53s\tremaining: 4m 8s\n",
      "822:\tlearn: 0.2374455\ttotal: 2m 53s\tremaining: 4m 7s\n",
      "823:\tlearn: 0.2374371\ttotal: 2m 53s\tremaining: 4m 7s\n",
      "824:\tlearn: 0.2374248\ttotal: 2m 53s\tremaining: 4m 7s\n",
      "825:\tlearn: 0.2374093\ttotal: 2m 53s\tremaining: 4m 7s\n",
      "826:\tlearn: 0.2374004\ttotal: 2m 54s\tremaining: 4m 7s\n",
      "827:\tlearn: 0.2373860\ttotal: 2m 54s\tremaining: 4m 6s\n",
      "828:\tlearn: 0.2373767\ttotal: 2m 54s\tremaining: 4m 6s\n",
      "829:\tlearn: 0.2373654\ttotal: 2m 54s\tremaining: 4m 6s\n",
      "830:\tlearn: 0.2373516\ttotal: 2m 54s\tremaining: 4m 6s\n",
      "831:\tlearn: 0.2373368\ttotal: 2m 55s\tremaining: 4m 5s\n",
      "832:\tlearn: 0.2373242\ttotal: 2m 55s\tremaining: 4m 5s\n",
      "833:\tlearn: 0.2373133\ttotal: 2m 55s\tremaining: 4m 5s\n",
      "834:\tlearn: 0.2373047\ttotal: 2m 55s\tremaining: 4m 5s\n",
      "835:\tlearn: 0.2372875\ttotal: 2m 56s\tremaining: 4m 5s\n",
      "836:\tlearn: 0.2372788\ttotal: 2m 56s\tremaining: 4m 4s\n",
      "837:\tlearn: 0.2372688\ttotal: 2m 56s\tremaining: 4m 4s\n",
      "838:\tlearn: 0.2372617\ttotal: 2m 56s\tremaining: 4m 4s\n",
      "839:\tlearn: 0.2372531\ttotal: 2m 56s\tremaining: 4m 4s\n",
      "840:\tlearn: 0.2372392\ttotal: 2m 57s\tremaining: 4m 4s\n",
      "841:\tlearn: 0.2372274\ttotal: 2m 57s\tremaining: 4m 3s\n",
      "842:\tlearn: 0.2372124\ttotal: 2m 57s\tremaining: 4m 3s\n",
      "843:\tlearn: 0.2372053\ttotal: 2m 57s\tremaining: 4m 3s\n",
      "844:\tlearn: 0.2371918\ttotal: 2m 57s\tremaining: 4m 3s\n",
      "845:\tlearn: 0.2371784\ttotal: 2m 58s\tremaining: 4m 3s\n",
      "846:\tlearn: 0.2371720\ttotal: 2m 58s\tremaining: 4m 2s\n",
      "847:\tlearn: 0.2371592\ttotal: 2m 58s\tremaining: 4m 2s\n",
      "848:\tlearn: 0.2371438\ttotal: 2m 58s\tremaining: 4m 2s\n",
      "849:\tlearn: 0.2371357\ttotal: 2m 59s\tremaining: 4m 2s\n",
      "850:\tlearn: 0.2371291\ttotal: 2m 59s\tremaining: 4m 2s\n",
      "851:\tlearn: 0.2371132\ttotal: 2m 59s\tremaining: 4m 1s\n",
      "852:\tlearn: 0.2370953\ttotal: 2m 59s\tremaining: 4m 1s\n",
      "853:\tlearn: 0.2370801\ttotal: 2m 59s\tremaining: 4m 1s\n",
      "854:\tlearn: 0.2370670\ttotal: 3m\tremaining: 4m 1s\n",
      "855:\tlearn: 0.2370562\ttotal: 3m\tremaining: 4m\n",
      "856:\tlearn: 0.2370454\ttotal: 3m\tremaining: 4m\n",
      "857:\tlearn: 0.2370283\ttotal: 3m\tremaining: 4m\n",
      "858:\tlearn: 0.2370156\ttotal: 3m\tremaining: 4m\n",
      "859:\tlearn: 0.2370042\ttotal: 3m 1s\tremaining: 4m\n",
      "860:\tlearn: 0.2369892\ttotal: 3m 1s\tremaining: 3m 59s\n",
      "861:\tlearn: 0.2369755\ttotal: 3m 1s\tremaining: 3m 59s\n",
      "862:\tlearn: 0.2369601\ttotal: 3m 1s\tremaining: 3m 59s\n",
      "863:\tlearn: 0.2369529\ttotal: 3m 2s\tremaining: 3m 59s\n",
      "864:\tlearn: 0.2369442\ttotal: 3m 2s\tremaining: 3m 59s\n",
      "865:\tlearn: 0.2369353\ttotal: 3m 2s\tremaining: 3m 58s\n",
      "866:\tlearn: 0.2369224\ttotal: 3m 2s\tremaining: 3m 58s\n",
      "867:\tlearn: 0.2369091\ttotal: 3m 2s\tremaining: 3m 58s\n",
      "868:\tlearn: 0.2368937\ttotal: 3m 3s\tremaining: 3m 58s\n",
      "869:\tlearn: 0.2368850\ttotal: 3m 3s\tremaining: 3m 58s\n",
      "870:\tlearn: 0.2368703\ttotal: 3m 3s\tremaining: 3m 57s\n",
      "871:\tlearn: 0.2368594\ttotal: 3m 3s\tremaining: 3m 57s\n",
      "872:\tlearn: 0.2368512\ttotal: 3m 3s\tremaining: 3m 57s\n",
      "873:\tlearn: 0.2368390\ttotal: 3m 4s\tremaining: 3m 57s\n",
      "874:\tlearn: 0.2368255\ttotal: 3m 4s\tremaining: 3m 57s\n",
      "875:\tlearn: 0.2368128\ttotal: 3m 4s\tremaining: 3m 56s\n",
      "876:\tlearn: 0.2368008\ttotal: 3m 4s\tremaining: 3m 56s\n",
      "877:\tlearn: 0.2367915\ttotal: 3m 4s\tremaining: 3m 56s\n",
      "878:\tlearn: 0.2367755\ttotal: 3m 5s\tremaining: 3m 56s\n",
      "879:\tlearn: 0.2367640\ttotal: 3m 5s\tremaining: 3m 56s\n",
      "880:\tlearn: 0.2367468\ttotal: 3m 5s\tremaining: 3m 55s\n",
      "881:\tlearn: 0.2367337\ttotal: 3m 5s\tremaining: 3m 55s\n",
      "882:\tlearn: 0.2367227\ttotal: 3m 6s\tremaining: 3m 55s\n",
      "883:\tlearn: 0.2367094\ttotal: 3m 6s\tremaining: 3m 55s\n",
      "884:\tlearn: 0.2366993\ttotal: 3m 6s\tremaining: 3m 54s\n",
      "885:\tlearn: 0.2366899\ttotal: 3m 6s\tremaining: 3m 54s\n",
      "886:\tlearn: 0.2366803\ttotal: 3m 6s\tremaining: 3m 54s\n",
      "887:\tlearn: 0.2366684\ttotal: 3m 7s\tremaining: 3m 54s\n",
      "888:\tlearn: 0.2366572\ttotal: 3m 7s\tremaining: 3m 54s\n",
      "889:\tlearn: 0.2366394\ttotal: 3m 7s\tremaining: 3m 53s\n",
      "890:\tlearn: 0.2366320\ttotal: 3m 7s\tremaining: 3m 53s\n",
      "891:\tlearn: 0.2366197\ttotal: 3m 7s\tremaining: 3m 53s\n",
      "892:\tlearn: 0.2366044\ttotal: 3m 8s\tremaining: 3m 53s\n",
      "893:\tlearn: 0.2365933\ttotal: 3m 8s\tremaining: 3m 53s\n",
      "894:\tlearn: 0.2365791\ttotal: 3m 8s\tremaining: 3m 52s\n",
      "895:\tlearn: 0.2365669\ttotal: 3m 8s\tremaining: 3m 52s\n",
      "896:\tlearn: 0.2365574\ttotal: 3m 9s\tremaining: 3m 52s\n",
      "897:\tlearn: 0.2365450\ttotal: 3m 9s\tremaining: 3m 52s\n",
      "898:\tlearn: 0.2365295\ttotal: 3m 9s\tremaining: 3m 52s\n",
      "899:\tlearn: 0.2365211\ttotal: 3m 9s\tremaining: 3m 51s\n",
      "900:\tlearn: 0.2365030\ttotal: 3m 9s\tremaining: 3m 51s\n",
      "901:\tlearn: 0.2364919\ttotal: 3m 10s\tremaining: 3m 51s\n",
      "902:\tlearn: 0.2364810\ttotal: 3m 10s\tremaining: 3m 51s\n",
      "903:\tlearn: 0.2364733\ttotal: 3m 10s\tremaining: 3m 50s\n",
      "904:\tlearn: 0.2364602\ttotal: 3m 10s\tremaining: 3m 50s\n",
      "905:\tlearn: 0.2364470\ttotal: 3m 10s\tremaining: 3m 50s\n",
      "906:\tlearn: 0.2364410\ttotal: 3m 11s\tremaining: 3m 50s\n",
      "907:\tlearn: 0.2364311\ttotal: 3m 11s\tremaining: 3m 50s\n",
      "908:\tlearn: 0.2364239\ttotal: 3m 11s\tremaining: 3m 49s\n",
      "909:\tlearn: 0.2364107\ttotal: 3m 11s\tremaining: 3m 49s\n",
      "910:\tlearn: 0.2363982\ttotal: 3m 11s\tremaining: 3m 49s\n",
      "911:\tlearn: 0.2363877\ttotal: 3m 12s\tremaining: 3m 49s\n",
      "912:\tlearn: 0.2363721\ttotal: 3m 12s\tremaining: 3m 49s\n",
      "913:\tlearn: 0.2363590\ttotal: 3m 12s\tremaining: 3m 48s\n",
      "914:\tlearn: 0.2363499\ttotal: 3m 12s\tremaining: 3m 48s\n",
      "915:\tlearn: 0.2363393\ttotal: 3m 13s\tremaining: 3m 48s\n",
      "916:\tlearn: 0.2363245\ttotal: 3m 13s\tremaining: 3m 48s\n",
      "917:\tlearn: 0.2363112\ttotal: 3m 13s\tremaining: 3m 48s\n",
      "918:\tlearn: 0.2363030\ttotal: 3m 13s\tremaining: 3m 47s\n",
      "919:\tlearn: 0.2362891\ttotal: 3m 13s\tremaining: 3m 47s\n",
      "920:\tlearn: 0.2362756\ttotal: 3m 14s\tremaining: 3m 47s\n",
      "921:\tlearn: 0.2362615\ttotal: 3m 14s\tremaining: 3m 47s\n",
      "922:\tlearn: 0.2362511\ttotal: 3m 14s\tremaining: 3m 46s\n",
      "923:\tlearn: 0.2362343\ttotal: 3m 14s\tremaining: 3m 46s\n",
      "924:\tlearn: 0.2362191\ttotal: 3m 14s\tremaining: 3m 46s\n",
      "925:\tlearn: 0.2362079\ttotal: 3m 15s\tremaining: 3m 46s\n",
      "926:\tlearn: 0.2361975\ttotal: 3m 15s\tremaining: 3m 46s\n",
      "927:\tlearn: 0.2361834\ttotal: 3m 15s\tremaining: 3m 45s\n",
      "928:\tlearn: 0.2361757\ttotal: 3m 15s\tremaining: 3m 45s\n",
      "929:\tlearn: 0.2361676\ttotal: 3m 16s\tremaining: 3m 45s\n",
      "930:\tlearn: 0.2361517\ttotal: 3m 16s\tremaining: 3m 45s\n",
      "931:\tlearn: 0.2361436\ttotal: 3m 16s\tremaining: 3m 45s\n",
      "932:\tlearn: 0.2361332\ttotal: 3m 16s\tremaining: 3m 44s\n",
      "933:\tlearn: 0.2361196\ttotal: 3m 16s\tremaining: 3m 44s\n",
      "934:\tlearn: 0.2361102\ttotal: 3m 17s\tremaining: 3m 44s\n",
      "935:\tlearn: 0.2360995\ttotal: 3m 17s\tremaining: 3m 44s\n",
      "936:\tlearn: 0.2360947\ttotal: 3m 17s\tremaining: 3m 44s\n",
      "937:\tlearn: 0.2360809\ttotal: 3m 17s\tremaining: 3m 43s\n",
      "938:\tlearn: 0.2360727\ttotal: 3m 17s\tremaining: 3m 43s\n",
      "939:\tlearn: 0.2360628\ttotal: 3m 18s\tremaining: 3m 43s\n",
      "940:\tlearn: 0.2360541\ttotal: 3m 18s\tremaining: 3m 43s\n",
      "941:\tlearn: 0.2360446\ttotal: 3m 18s\tremaining: 3m 43s\n",
      "942:\tlearn: 0.2360311\ttotal: 3m 18s\tremaining: 3m 42s\n",
      "943:\tlearn: 0.2360246\ttotal: 3m 18s\tremaining: 3m 42s\n",
      "944:\tlearn: 0.2360118\ttotal: 3m 19s\tremaining: 3m 42s\n",
      "945:\tlearn: 0.2359973\ttotal: 3m 19s\tremaining: 3m 42s\n",
      "946:\tlearn: 0.2359882\ttotal: 3m 19s\tremaining: 3m 41s\n",
      "947:\tlearn: 0.2359760\ttotal: 3m 19s\tremaining: 3m 41s\n",
      "948:\tlearn: 0.2359630\ttotal: 3m 20s\tremaining: 3m 41s\n",
      "949:\tlearn: 0.2359452\ttotal: 3m 20s\tremaining: 3m 41s\n",
      "950:\tlearn: 0.2359341\ttotal: 3m 20s\tremaining: 3m 41s\n",
      "951:\tlearn: 0.2359171\ttotal: 3m 20s\tremaining: 3m 40s\n",
      "952:\tlearn: 0.2359061\ttotal: 3m 20s\tremaining: 3m 40s\n",
      "953:\tlearn: 0.2358944\ttotal: 3m 21s\tremaining: 3m 40s\n",
      "954:\tlearn: 0.2358834\ttotal: 3m 21s\tremaining: 3m 40s\n",
      "955:\tlearn: 0.2358719\ttotal: 3m 21s\tremaining: 3m 40s\n",
      "956:\tlearn: 0.2358638\ttotal: 3m 21s\tremaining: 3m 39s\n",
      "957:\tlearn: 0.2358528\ttotal: 3m 21s\tremaining: 3m 39s\n",
      "958:\tlearn: 0.2358413\ttotal: 3m 22s\tremaining: 3m 39s\n",
      "959:\tlearn: 0.2358293\ttotal: 3m 22s\tremaining: 3m 39s\n",
      "960:\tlearn: 0.2358149\ttotal: 3m 22s\tremaining: 3m 39s\n",
      "961:\tlearn: 0.2358041\ttotal: 3m 22s\tremaining: 3m 38s\n",
      "962:\tlearn: 0.2357962\ttotal: 3m 23s\tremaining: 3m 38s\n",
      "963:\tlearn: 0.2357835\ttotal: 3m 23s\tremaining: 3m 38s\n",
      "964:\tlearn: 0.2357727\ttotal: 3m 23s\tremaining: 3m 38s\n",
      "965:\tlearn: 0.2357631\ttotal: 3m 23s\tremaining: 3m 38s\n",
      "966:\tlearn: 0.2357571\ttotal: 3m 23s\tremaining: 3m 37s\n",
      "967:\tlearn: 0.2357467\ttotal: 3m 24s\tremaining: 3m 37s\n",
      "968:\tlearn: 0.2357357\ttotal: 3m 24s\tremaining: 3m 37s\n",
      "969:\tlearn: 0.2357247\ttotal: 3m 24s\tremaining: 3m 37s\n",
      "970:\tlearn: 0.2357103\ttotal: 3m 24s\tremaining: 3m 36s\n",
      "971:\tlearn: 0.2357000\ttotal: 3m 24s\tremaining: 3m 36s\n",
      "972:\tlearn: 0.2356939\ttotal: 3m 25s\tremaining: 3m 36s\n",
      "973:\tlearn: 0.2356850\ttotal: 3m 25s\tremaining: 3m 36s\n",
      "974:\tlearn: 0.2356722\ttotal: 3m 25s\tremaining: 3m 36s\n",
      "975:\tlearn: 0.2356636\ttotal: 3m 25s\tremaining: 3m 35s\n",
      "976:\tlearn: 0.2356466\ttotal: 3m 26s\tremaining: 3m 35s\n",
      "977:\tlearn: 0.2356351\ttotal: 3m 26s\tremaining: 3m 35s\n",
      "978:\tlearn: 0.2356238\ttotal: 3m 26s\tremaining: 3m 35s\n",
      "979:\tlearn: 0.2356112\ttotal: 3m 26s\tremaining: 3m 35s\n",
      "980:\tlearn: 0.2355985\ttotal: 3m 26s\tremaining: 3m 34s\n",
      "981:\tlearn: 0.2355866\ttotal: 3m 27s\tremaining: 3m 34s\n",
      "982:\tlearn: 0.2355760\ttotal: 3m 27s\tremaining: 3m 34s\n",
      "983:\tlearn: 0.2355682\ttotal: 3m 27s\tremaining: 3m 34s\n",
      "984:\tlearn: 0.2355587\ttotal: 3m 27s\tremaining: 3m 34s\n",
      "985:\tlearn: 0.2355457\ttotal: 3m 27s\tremaining: 3m 33s\n",
      "986:\tlearn: 0.2355332\ttotal: 3m 28s\tremaining: 3m 33s\n",
      "987:\tlearn: 0.2355179\ttotal: 3m 28s\tremaining: 3m 33s\n",
      "988:\tlearn: 0.2355060\ttotal: 3m 28s\tremaining: 3m 33s\n",
      "989:\tlearn: 0.2354887\ttotal: 3m 28s\tremaining: 3m 32s\n",
      "990:\tlearn: 0.2354750\ttotal: 3m 28s\tremaining: 3m 32s\n",
      "991:\tlearn: 0.2354629\ttotal: 3m 29s\tremaining: 3m 32s\n",
      "992:\tlearn: 0.2354524\ttotal: 3m 29s\tremaining: 3m 32s\n",
      "993:\tlearn: 0.2354410\ttotal: 3m 29s\tremaining: 3m 32s\n",
      "994:\tlearn: 0.2354314\ttotal: 3m 29s\tremaining: 3m 31s\n",
      "995:\tlearn: 0.2354164\ttotal: 3m 30s\tremaining: 3m 31s\n",
      "996:\tlearn: 0.2354039\ttotal: 3m 30s\tremaining: 3m 31s\n",
      "997:\tlearn: 0.2353942\ttotal: 3m 30s\tremaining: 3m 31s\n",
      "998:\tlearn: 0.2353837\ttotal: 3m 30s\tremaining: 3m 31s\n",
      "999:\tlearn: 0.2353688\ttotal: 3m 30s\tremaining: 3m 30s\n",
      "1000:\tlearn: 0.2353582\ttotal: 3m 31s\tremaining: 3m 30s\n",
      "1001:\tlearn: 0.2353469\ttotal: 3m 31s\tremaining: 3m 30s\n",
      "1002:\tlearn: 0.2353291\ttotal: 3m 31s\tremaining: 3m 30s\n",
      "1003:\tlearn: 0.2353159\ttotal: 3m 31s\tremaining: 3m 30s\n",
      "1004:\tlearn: 0.2353098\ttotal: 3m 31s\tremaining: 3m 29s\n",
      "1005:\tlearn: 0.2353021\ttotal: 3m 32s\tremaining: 3m 29s\n",
      "1006:\tlearn: 0.2352913\ttotal: 3m 32s\tremaining: 3m 29s\n",
      "1007:\tlearn: 0.2352839\ttotal: 3m 32s\tremaining: 3m 29s\n",
      "1008:\tlearn: 0.2352765\ttotal: 3m 32s\tremaining: 3m 28s\n",
      "1009:\tlearn: 0.2352641\ttotal: 3m 32s\tremaining: 3m 28s\n",
      "1010:\tlearn: 0.2352496\ttotal: 3m 33s\tremaining: 3m 28s\n",
      "1011:\tlearn: 0.2352399\ttotal: 3m 33s\tremaining: 3m 28s\n",
      "1012:\tlearn: 0.2352215\ttotal: 3m 33s\tremaining: 3m 28s\n",
      "1013:\tlearn: 0.2352076\ttotal: 3m 33s\tremaining: 3m 27s\n",
      "1014:\tlearn: 0.2352020\ttotal: 3m 34s\tremaining: 3m 27s\n",
      "1015:\tlearn: 0.2351906\ttotal: 3m 34s\tremaining: 3m 27s\n",
      "1016:\tlearn: 0.2351793\ttotal: 3m 34s\tremaining: 3m 27s\n",
      "1017:\tlearn: 0.2351680\ttotal: 3m 34s\tremaining: 3m 27s\n",
      "1018:\tlearn: 0.2351549\ttotal: 3m 34s\tremaining: 3m 26s\n",
      "1019:\tlearn: 0.2351478\ttotal: 3m 35s\tremaining: 3m 26s\n",
      "1020:\tlearn: 0.2351304\ttotal: 3m 35s\tremaining: 3m 26s\n",
      "1021:\tlearn: 0.2351192\ttotal: 3m 35s\tremaining: 3m 26s\n",
      "1022:\tlearn: 0.2351083\ttotal: 3m 35s\tremaining: 3m 26s\n",
      "1023:\tlearn: 0.2351003\ttotal: 3m 35s\tremaining: 3m 25s\n",
      "1024:\tlearn: 0.2350915\ttotal: 3m 36s\tremaining: 3m 25s\n",
      "1025:\tlearn: 0.2350730\ttotal: 3m 36s\tremaining: 3m 25s\n",
      "1026:\tlearn: 0.2350640\ttotal: 3m 36s\tremaining: 3m 25s\n",
      "1027:\tlearn: 0.2350513\ttotal: 3m 36s\tremaining: 3m 25s\n",
      "1028:\tlearn: 0.2350434\ttotal: 3m 37s\tremaining: 3m 24s\n",
      "1029:\tlearn: 0.2350337\ttotal: 3m 37s\tremaining: 3m 24s\n",
      "1030:\tlearn: 0.2350267\ttotal: 3m 37s\tremaining: 3m 24s\n",
      "1031:\tlearn: 0.2350075\ttotal: 3m 37s\tremaining: 3m 24s\n",
      "1032:\tlearn: 0.2349964\ttotal: 3m 37s\tremaining: 3m 24s\n",
      "1033:\tlearn: 0.2349891\ttotal: 3m 38s\tremaining: 3m 23s\n",
      "1034:\tlearn: 0.2349771\ttotal: 3m 38s\tremaining: 3m 23s\n",
      "1035:\tlearn: 0.2349671\ttotal: 3m 38s\tremaining: 3m 23s\n",
      "1036:\tlearn: 0.2349542\ttotal: 3m 38s\tremaining: 3m 23s\n",
      "1037:\tlearn: 0.2349424\ttotal: 3m 38s\tremaining: 3m 22s\n",
      "1038:\tlearn: 0.2349328\ttotal: 3m 39s\tremaining: 3m 22s\n",
      "1039:\tlearn: 0.2349240\ttotal: 3m 39s\tremaining: 3m 22s\n",
      "1040:\tlearn: 0.2349129\ttotal: 3m 39s\tremaining: 3m 22s\n",
      "1041:\tlearn: 0.2348958\ttotal: 3m 39s\tremaining: 3m 22s\n",
      "1042:\tlearn: 0.2348833\ttotal: 3m 40s\tremaining: 3m 21s\n",
      "1043:\tlearn: 0.2348719\ttotal: 3m 40s\tremaining: 3m 21s\n",
      "1044:\tlearn: 0.2348598\ttotal: 3m 40s\tremaining: 3m 21s\n",
      "1045:\tlearn: 0.2348522\ttotal: 3m 40s\tremaining: 3m 21s\n",
      "1046:\tlearn: 0.2348381\ttotal: 3m 40s\tremaining: 3m 21s\n",
      "1047:\tlearn: 0.2348242\ttotal: 3m 41s\tremaining: 3m 20s\n",
      "1048:\tlearn: 0.2348151\ttotal: 3m 41s\tremaining: 3m 20s\n",
      "1049:\tlearn: 0.2348043\ttotal: 3m 41s\tremaining: 3m 20s\n",
      "1050:\tlearn: 0.2347921\ttotal: 3m 41s\tremaining: 3m 20s\n",
      "1051:\tlearn: 0.2347816\ttotal: 3m 41s\tremaining: 3m 19s\n",
      "1052:\tlearn: 0.2347687\ttotal: 3m 42s\tremaining: 3m 19s\n",
      "1053:\tlearn: 0.2347573\ttotal: 3m 42s\tremaining: 3m 19s\n",
      "1054:\tlearn: 0.2347426\ttotal: 3m 42s\tremaining: 3m 19s\n",
      "1055:\tlearn: 0.2347310\ttotal: 3m 42s\tremaining: 3m 19s\n",
      "1056:\tlearn: 0.2347206\ttotal: 3m 43s\tremaining: 3m 18s\n",
      "1057:\tlearn: 0.2347074\ttotal: 3m 43s\tremaining: 3m 18s\n",
      "1058:\tlearn: 0.2346976\ttotal: 3m 43s\tremaining: 3m 18s\n",
      "1059:\tlearn: 0.2346888\ttotal: 3m 43s\tremaining: 3m 18s\n",
      "1060:\tlearn: 0.2346793\ttotal: 3m 43s\tremaining: 3m 18s\n",
      "1061:\tlearn: 0.2346645\ttotal: 3m 44s\tremaining: 3m 17s\n",
      "1062:\tlearn: 0.2346527\ttotal: 3m 44s\tremaining: 3m 17s\n",
      "1063:\tlearn: 0.2346403\ttotal: 3m 44s\tremaining: 3m 17s\n",
      "1064:\tlearn: 0.2346261\ttotal: 3m 44s\tremaining: 3m 17s\n",
      "1065:\tlearn: 0.2346172\ttotal: 3m 44s\tremaining: 3m 17s\n",
      "1066:\tlearn: 0.2346044\ttotal: 3m 45s\tremaining: 3m 16s\n",
      "1067:\tlearn: 0.2345927\ttotal: 3m 45s\tremaining: 3m 16s\n",
      "1068:\tlearn: 0.2345807\ttotal: 3m 45s\tremaining: 3m 16s\n",
      "1069:\tlearn: 0.2345688\ttotal: 3m 45s\tremaining: 3m 16s\n",
      "1070:\tlearn: 0.2345538\ttotal: 3m 45s\tremaining: 3m 16s\n",
      "1071:\tlearn: 0.2345400\ttotal: 3m 46s\tremaining: 3m 15s\n",
      "1072:\tlearn: 0.2345212\ttotal: 3m 46s\tremaining: 3m 15s\n",
      "1073:\tlearn: 0.2345076\ttotal: 3m 46s\tremaining: 3m 15s\n",
      "1074:\tlearn: 0.2345022\ttotal: 3m 46s\tremaining: 3m 15s\n",
      "1075:\tlearn: 0.2344917\ttotal: 3m 47s\tremaining: 3m 14s\n",
      "1076:\tlearn: 0.2344791\ttotal: 3m 47s\tremaining: 3m 14s\n",
      "1077:\tlearn: 0.2344703\ttotal: 3m 47s\tremaining: 3m 14s\n",
      "1078:\tlearn: 0.2344569\ttotal: 3m 47s\tremaining: 3m 14s\n",
      "1079:\tlearn: 0.2344447\ttotal: 3m 47s\tremaining: 3m 14s\n",
      "1080:\tlearn: 0.2344342\ttotal: 3m 48s\tremaining: 3m 13s\n",
      "1081:\tlearn: 0.2344260\ttotal: 3m 48s\tremaining: 3m 13s\n",
      "1082:\tlearn: 0.2344153\ttotal: 3m 48s\tremaining: 3m 13s\n",
      "1083:\tlearn: 0.2344043\ttotal: 3m 48s\tremaining: 3m 13s\n",
      "1084:\tlearn: 0.2343956\ttotal: 3m 48s\tremaining: 3m 13s\n",
      "1085:\tlearn: 0.2343803\ttotal: 3m 49s\tremaining: 3m 12s\n",
      "1086:\tlearn: 0.2343695\ttotal: 3m 49s\tremaining: 3m 12s\n",
      "1087:\tlearn: 0.2343604\ttotal: 3m 49s\tremaining: 3m 12s\n",
      "1088:\tlearn: 0.2343548\ttotal: 3m 49s\tremaining: 3m 12s\n",
      "1089:\tlearn: 0.2343401\ttotal: 3m 50s\tremaining: 3m 12s\n",
      "1090:\tlearn: 0.2343290\ttotal: 3m 50s\tremaining: 3m 11s\n",
      "1091:\tlearn: 0.2343204\ttotal: 3m 50s\tremaining: 3m 11s\n",
      "1092:\tlearn: 0.2343051\ttotal: 3m 50s\tremaining: 3m 11s\n",
      "1093:\tlearn: 0.2342929\ttotal: 3m 50s\tremaining: 3m 11s\n",
      "1094:\tlearn: 0.2342818\ttotal: 3m 51s\tremaining: 3m 10s\n",
      "1095:\tlearn: 0.2342732\ttotal: 3m 51s\tremaining: 3m 10s\n",
      "1096:\tlearn: 0.2342627\ttotal: 3m 51s\tremaining: 3m 10s\n",
      "1097:\tlearn: 0.2342460\ttotal: 3m 51s\tremaining: 3m 10s\n",
      "1098:\tlearn: 0.2342354\ttotal: 3m 51s\tremaining: 3m 10s\n",
      "1099:\tlearn: 0.2342234\ttotal: 3m 52s\tremaining: 3m 9s\n",
      "1100:\tlearn: 0.2342102\ttotal: 3m 52s\tremaining: 3m 9s\n",
      "1101:\tlearn: 0.2341978\ttotal: 3m 52s\tremaining: 3m 9s\n",
      "1102:\tlearn: 0.2341806\ttotal: 3m 52s\tremaining: 3m 9s\n",
      "1103:\tlearn: 0.2341685\ttotal: 3m 52s\tremaining: 3m 9s\n",
      "1104:\tlearn: 0.2341602\ttotal: 3m 53s\tremaining: 3m 8s\n",
      "1105:\tlearn: 0.2341483\ttotal: 3m 53s\tremaining: 3m 8s\n",
      "1106:\tlearn: 0.2341394\ttotal: 3m 53s\tremaining: 3m 8s\n",
      "1107:\tlearn: 0.2341303\ttotal: 3m 53s\tremaining: 3m 8s\n",
      "1108:\tlearn: 0.2341229\ttotal: 3m 54s\tremaining: 3m 8s\n",
      "1109:\tlearn: 0.2341145\ttotal: 3m 54s\tremaining: 3m 7s\n",
      "1110:\tlearn: 0.2341030\ttotal: 3m 54s\tremaining: 3m 7s\n",
      "1111:\tlearn: 0.2340915\ttotal: 3m 54s\tremaining: 3m 7s\n",
      "1112:\tlearn: 0.2340777\ttotal: 3m 54s\tremaining: 3m 7s\n",
      "1113:\tlearn: 0.2340630\ttotal: 3m 55s\tremaining: 3m 6s\n",
      "1114:\tlearn: 0.2340541\ttotal: 3m 55s\tremaining: 3m 6s\n",
      "1115:\tlearn: 0.2340413\ttotal: 3m 55s\tremaining: 3m 6s\n",
      "1116:\tlearn: 0.2340331\ttotal: 3m 55s\tremaining: 3m 6s\n",
      "1117:\tlearn: 0.2340269\ttotal: 3m 55s\tremaining: 3m 6s\n",
      "1118:\tlearn: 0.2340183\ttotal: 3m 56s\tremaining: 3m 5s\n",
      "1119:\tlearn: 0.2340056\ttotal: 3m 56s\tremaining: 3m 5s\n",
      "1120:\tlearn: 0.2339945\ttotal: 3m 56s\tremaining: 3m 5s\n",
      "1121:\tlearn: 0.2339825\ttotal: 3m 56s\tremaining: 3m 5s\n",
      "1122:\tlearn: 0.2339691\ttotal: 3m 56s\tremaining: 3m 5s\n",
      "1123:\tlearn: 0.2339563\ttotal: 3m 57s\tremaining: 3m 4s\n",
      "1124:\tlearn: 0.2339416\ttotal: 3m 57s\tremaining: 3m 4s\n",
      "1125:\tlearn: 0.2339358\ttotal: 3m 57s\tremaining: 3m 4s\n",
      "1126:\tlearn: 0.2339246\ttotal: 3m 57s\tremaining: 3m 4s\n",
      "1127:\tlearn: 0.2339123\ttotal: 3m 58s\tremaining: 3m 4s\n",
      "1128:\tlearn: 0.2339013\ttotal: 3m 58s\tremaining: 3m 3s\n",
      "1129:\tlearn: 0.2338861\ttotal: 3m 58s\tremaining: 3m 3s\n",
      "1130:\tlearn: 0.2338723\ttotal: 3m 58s\tremaining: 3m 3s\n",
      "1131:\tlearn: 0.2338627\ttotal: 3m 58s\tremaining: 3m 3s\n",
      "1132:\tlearn: 0.2338514\ttotal: 3m 59s\tremaining: 3m 2s\n",
      "1133:\tlearn: 0.2338437\ttotal: 3m 59s\tremaining: 3m 2s\n",
      "1134:\tlearn: 0.2338345\ttotal: 3m 59s\tremaining: 3m 2s\n",
      "1135:\tlearn: 0.2338246\ttotal: 3m 59s\tremaining: 3m 2s\n",
      "1136:\tlearn: 0.2338079\ttotal: 3m 59s\tremaining: 3m 2s\n",
      "1137:\tlearn: 0.2337976\ttotal: 4m\tremaining: 3m 1s\n",
      "1138:\tlearn: 0.2337924\ttotal: 4m\tremaining: 3m 1s\n",
      "1139:\tlearn: 0.2337802\ttotal: 4m\tremaining: 3m 1s\n",
      "1140:\tlearn: 0.2337655\ttotal: 4m\tremaining: 3m 1s\n",
      "1141:\tlearn: 0.2337590\ttotal: 4m\tremaining: 3m 1s\n",
      "1142:\tlearn: 0.2337489\ttotal: 4m 1s\tremaining: 3m\n",
      "1143:\tlearn: 0.2337347\ttotal: 4m 1s\tremaining: 3m\n",
      "1144:\tlearn: 0.2337236\ttotal: 4m 1s\tremaining: 3m\n",
      "1145:\tlearn: 0.2337110\ttotal: 4m 1s\tremaining: 3m\n",
      "1146:\tlearn: 0.2336999\ttotal: 4m 2s\tremaining: 3m\n",
      "1147:\tlearn: 0.2336898\ttotal: 4m 2s\tremaining: 2m 59s\n",
      "1148:\tlearn: 0.2336734\ttotal: 4m 2s\tremaining: 2m 59s\n",
      "1149:\tlearn: 0.2336593\ttotal: 4m 2s\tremaining: 2m 59s\n",
      "1150:\tlearn: 0.2336497\ttotal: 4m 2s\tremaining: 2m 59s\n",
      "1151:\tlearn: 0.2336430\ttotal: 4m 3s\tremaining: 2m 58s\n",
      "1152:\tlearn: 0.2336309\ttotal: 4m 3s\tremaining: 2m 58s\n",
      "1153:\tlearn: 0.2336191\ttotal: 4m 3s\tremaining: 2m 58s\n",
      "1154:\tlearn: 0.2336074\ttotal: 4m 3s\tremaining: 2m 58s\n",
      "1155:\tlearn: 0.2335940\ttotal: 4m 3s\tremaining: 2m 58s\n",
      "1156:\tlearn: 0.2335854\ttotal: 4m 4s\tremaining: 2m 57s\n",
      "1157:\tlearn: 0.2335761\ttotal: 4m 4s\tremaining: 2m 57s\n",
      "1158:\tlearn: 0.2335686\ttotal: 4m 4s\tremaining: 2m 57s\n",
      "1159:\tlearn: 0.2335577\ttotal: 4m 4s\tremaining: 2m 57s\n",
      "1160:\tlearn: 0.2335468\ttotal: 4m 5s\tremaining: 2m 57s\n",
      "1161:\tlearn: 0.2335333\ttotal: 4m 5s\tremaining: 2m 56s\n",
      "1162:\tlearn: 0.2335199\ttotal: 4m 5s\tremaining: 2m 56s\n",
      "1163:\tlearn: 0.2335102\ttotal: 4m 5s\tremaining: 2m 56s\n",
      "1164:\tlearn: 0.2334979\ttotal: 4m 5s\tremaining: 2m 56s\n",
      "1165:\tlearn: 0.2334892\ttotal: 4m 6s\tremaining: 2m 56s\n",
      "1166:\tlearn: 0.2334809\ttotal: 4m 6s\tremaining: 2m 55s\n",
      "1167:\tlearn: 0.2334673\ttotal: 4m 6s\tremaining: 2m 55s\n",
      "1168:\tlearn: 0.2334531\ttotal: 4m 6s\tremaining: 2m 55s\n",
      "1169:\tlearn: 0.2334450\ttotal: 4m 6s\tremaining: 2m 55s\n",
      "1170:\tlearn: 0.2334341\ttotal: 4m 7s\tremaining: 2m 54s\n",
      "1171:\tlearn: 0.2334193\ttotal: 4m 7s\tremaining: 2m 54s\n",
      "1172:\tlearn: 0.2334060\ttotal: 4m 7s\tremaining: 2m 54s\n",
      "1173:\tlearn: 0.2333975\ttotal: 4m 7s\tremaining: 2m 54s\n",
      "1174:\tlearn: 0.2333837\ttotal: 4m 7s\tremaining: 2m 54s\n",
      "1175:\tlearn: 0.2333677\ttotal: 4m 8s\tremaining: 2m 53s\n",
      "1176:\tlearn: 0.2333563\ttotal: 4m 8s\tremaining: 2m 53s\n",
      "1177:\tlearn: 0.2333426\ttotal: 4m 8s\tremaining: 2m 53s\n",
      "1178:\tlearn: 0.2333359\ttotal: 4m 8s\tremaining: 2m 53s\n",
      "1179:\tlearn: 0.2333253\ttotal: 4m 9s\tremaining: 2m 53s\n",
      "1180:\tlearn: 0.2333135\ttotal: 4m 9s\tremaining: 2m 52s\n",
      "1181:\tlearn: 0.2333015\ttotal: 4m 9s\tremaining: 2m 52s\n",
      "1182:\tlearn: 0.2332959\ttotal: 4m 9s\tremaining: 2m 52s\n",
      "1183:\tlearn: 0.2332855\ttotal: 4m 9s\tremaining: 2m 52s\n",
      "1184:\tlearn: 0.2332771\ttotal: 4m 10s\tremaining: 2m 52s\n",
      "1185:\tlearn: 0.2332661\ttotal: 4m 10s\tremaining: 2m 51s\n",
      "1186:\tlearn: 0.2332584\ttotal: 4m 10s\tremaining: 2m 51s\n",
      "1187:\tlearn: 0.2332491\ttotal: 4m 10s\tremaining: 2m 51s\n",
      "1188:\tlearn: 0.2332360\ttotal: 4m 10s\tremaining: 2m 51s\n",
      "1189:\tlearn: 0.2332228\ttotal: 4m 11s\tremaining: 2m 50s\n",
      "1190:\tlearn: 0.2332166\ttotal: 4m 11s\tremaining: 2m 50s\n",
      "1191:\tlearn: 0.2332015\ttotal: 4m 11s\tremaining: 2m 50s\n",
      "1192:\tlearn: 0.2331905\ttotal: 4m 11s\tremaining: 2m 50s\n",
      "1193:\tlearn: 0.2331776\ttotal: 4m 12s\tremaining: 2m 50s\n",
      "1194:\tlearn: 0.2331635\ttotal: 4m 12s\tremaining: 2m 49s\n",
      "1195:\tlearn: 0.2331478\ttotal: 4m 12s\tremaining: 2m 49s\n",
      "1196:\tlearn: 0.2331366\ttotal: 4m 12s\tremaining: 2m 49s\n",
      "1197:\tlearn: 0.2331186\ttotal: 4m 12s\tremaining: 2m 49s\n",
      "1198:\tlearn: 0.2331095\ttotal: 4m 13s\tremaining: 2m 49s\n",
      "1199:\tlearn: 0.2331029\ttotal: 4m 13s\tremaining: 2m 48s\n",
      "1200:\tlearn: 0.2330908\ttotal: 4m 13s\tremaining: 2m 48s\n",
      "1201:\tlearn: 0.2330801\ttotal: 4m 13s\tremaining: 2m 48s\n",
      "1202:\tlearn: 0.2330635\ttotal: 4m 13s\tremaining: 2m 48s\n",
      "1203:\tlearn: 0.2330497\ttotal: 4m 14s\tremaining: 2m 48s\n",
      "1204:\tlearn: 0.2330377\ttotal: 4m 14s\tremaining: 2m 47s\n",
      "1205:\tlearn: 0.2330288\ttotal: 4m 14s\tremaining: 2m 47s\n",
      "1206:\tlearn: 0.2330180\ttotal: 4m 14s\tremaining: 2m 47s\n",
      "1207:\tlearn: 0.2330094\ttotal: 4m 14s\tremaining: 2m 47s\n",
      "1208:\tlearn: 0.2329993\ttotal: 4m 15s\tremaining: 2m 46s\n",
      "1209:\tlearn: 0.2329913\ttotal: 4m 15s\tremaining: 2m 46s\n",
      "1210:\tlearn: 0.2329804\ttotal: 4m 15s\tremaining: 2m 46s\n",
      "1211:\tlearn: 0.2329652\ttotal: 4m 15s\tremaining: 2m 46s\n",
      "1212:\tlearn: 0.2329523\ttotal: 4m 16s\tremaining: 2m 46s\n",
      "1213:\tlearn: 0.2329426\ttotal: 4m 16s\tremaining: 2m 45s\n",
      "1214:\tlearn: 0.2329315\ttotal: 4m 16s\tremaining: 2m 45s\n",
      "1215:\tlearn: 0.2329209\ttotal: 4m 16s\tremaining: 2m 45s\n",
      "1216:\tlearn: 0.2329104\ttotal: 4m 16s\tremaining: 2m 45s\n",
      "1217:\tlearn: 0.2328922\ttotal: 4m 17s\tremaining: 2m 45s\n",
      "1218:\tlearn: 0.2328779\ttotal: 4m 17s\tremaining: 2m 44s\n",
      "1219:\tlearn: 0.2328673\ttotal: 4m 17s\tremaining: 2m 44s\n",
      "1220:\tlearn: 0.2328560\ttotal: 4m 17s\tremaining: 2m 44s\n",
      "1221:\tlearn: 0.2328511\ttotal: 4m 17s\tremaining: 2m 44s\n",
      "1222:\tlearn: 0.2328404\ttotal: 4m 18s\tremaining: 2m 44s\n",
      "1223:\tlearn: 0.2328278\ttotal: 4m 18s\tremaining: 2m 43s\n",
      "1224:\tlearn: 0.2328155\ttotal: 4m 18s\tremaining: 2m 43s\n",
      "1225:\tlearn: 0.2328073\ttotal: 4m 18s\tremaining: 2m 43s\n",
      "1226:\tlearn: 0.2327928\ttotal: 4m 18s\tremaining: 2m 43s\n",
      "1227:\tlearn: 0.2327839\ttotal: 4m 19s\tremaining: 2m 42s\n",
      "1228:\tlearn: 0.2327726\ttotal: 4m 19s\tremaining: 2m 42s\n",
      "1229:\tlearn: 0.2327639\ttotal: 4m 19s\tremaining: 2m 42s\n",
      "1230:\tlearn: 0.2327525\ttotal: 4m 19s\tremaining: 2m 42s\n",
      "1231:\tlearn: 0.2327427\ttotal: 4m 20s\tremaining: 2m 42s\n",
      "1232:\tlearn: 0.2327316\ttotal: 4m 20s\tremaining: 2m 41s\n",
      "1233:\tlearn: 0.2327167\ttotal: 4m 20s\tremaining: 2m 41s\n",
      "1234:\tlearn: 0.2327023\ttotal: 4m 20s\tremaining: 2m 41s\n",
      "1235:\tlearn: 0.2326909\ttotal: 4m 20s\tremaining: 2m 41s\n",
      "1236:\tlearn: 0.2326783\ttotal: 4m 21s\tremaining: 2m 41s\n",
      "1237:\tlearn: 0.2326626\ttotal: 4m 21s\tremaining: 2m 40s\n",
      "1238:\tlearn: 0.2326472\ttotal: 4m 21s\tremaining: 2m 40s\n",
      "1239:\tlearn: 0.2326381\ttotal: 4m 21s\tremaining: 2m 40s\n",
      "1240:\tlearn: 0.2326260\ttotal: 4m 21s\tremaining: 2m 40s\n",
      "1241:\tlearn: 0.2326109\ttotal: 4m 22s\tremaining: 2m 40s\n",
      "1242:\tlearn: 0.2325973\ttotal: 4m 22s\tremaining: 2m 39s\n",
      "1243:\tlearn: 0.2325874\ttotal: 4m 22s\tremaining: 2m 39s\n",
      "1244:\tlearn: 0.2325719\ttotal: 4m 22s\tremaining: 2m 39s\n",
      "1245:\tlearn: 0.2325601\ttotal: 4m 23s\tremaining: 2m 39s\n",
      "1246:\tlearn: 0.2325453\ttotal: 4m 23s\tremaining: 2m 38s\n",
      "1247:\tlearn: 0.2325357\ttotal: 4m 23s\tremaining: 2m 38s\n",
      "1248:\tlearn: 0.2325252\ttotal: 4m 23s\tremaining: 2m 38s\n",
      "1249:\tlearn: 0.2325153\ttotal: 4m 23s\tremaining: 2m 38s\n",
      "1250:\tlearn: 0.2325045\ttotal: 4m 24s\tremaining: 2m 38s\n",
      "1251:\tlearn: 0.2324917\ttotal: 4m 24s\tremaining: 2m 37s\n",
      "1252:\tlearn: 0.2324795\ttotal: 4m 24s\tremaining: 2m 37s\n",
      "1253:\tlearn: 0.2324662\ttotal: 4m 24s\tremaining: 2m 37s\n",
      "1254:\tlearn: 0.2324551\ttotal: 4m 24s\tremaining: 2m 37s\n",
      "1255:\tlearn: 0.2324440\ttotal: 4m 25s\tremaining: 2m 37s\n",
      "1256:\tlearn: 0.2324361\ttotal: 4m 25s\tremaining: 2m 36s\n",
      "1257:\tlearn: 0.2324242\ttotal: 4m 25s\tremaining: 2m 36s\n",
      "1258:\tlearn: 0.2324117\ttotal: 4m 25s\tremaining: 2m 36s\n",
      "1259:\tlearn: 0.2323994\ttotal: 4m 26s\tremaining: 2m 36s\n",
      "1260:\tlearn: 0.2323876\ttotal: 4m 26s\tremaining: 2m 36s\n",
      "1261:\tlearn: 0.2323737\ttotal: 4m 26s\tremaining: 2m 35s\n",
      "1262:\tlearn: 0.2323631\ttotal: 4m 26s\tremaining: 2m 35s\n",
      "1263:\tlearn: 0.2323509\ttotal: 4m 26s\tremaining: 2m 35s\n",
      "1264:\tlearn: 0.2323371\ttotal: 4m 27s\tremaining: 2m 35s\n",
      "1265:\tlearn: 0.2323257\ttotal: 4m 27s\tremaining: 2m 34s\n",
      "1266:\tlearn: 0.2323139\ttotal: 4m 27s\tremaining: 2m 34s\n",
      "1267:\tlearn: 0.2323038\ttotal: 4m 27s\tremaining: 2m 34s\n",
      "1268:\tlearn: 0.2322870\ttotal: 4m 27s\tremaining: 2m 34s\n",
      "1269:\tlearn: 0.2322783\ttotal: 4m 28s\tremaining: 2m 34s\n",
      "1270:\tlearn: 0.2322682\ttotal: 4m 28s\tremaining: 2m 33s\n",
      "1271:\tlearn: 0.2322602\ttotal: 4m 28s\tremaining: 2m 33s\n",
      "1272:\tlearn: 0.2322531\ttotal: 4m 28s\tremaining: 2m 33s\n",
      "1273:\tlearn: 0.2322411\ttotal: 4m 28s\tremaining: 2m 33s\n",
      "1274:\tlearn: 0.2322320\ttotal: 4m 29s\tremaining: 2m 33s\n",
      "1275:\tlearn: 0.2322203\ttotal: 4m 29s\tremaining: 2m 32s\n",
      "1276:\tlearn: 0.2322054\ttotal: 4m 29s\tremaining: 2m 32s\n",
      "1277:\tlearn: 0.2321932\ttotal: 4m 29s\tremaining: 2m 32s\n",
      "1278:\tlearn: 0.2321860\ttotal: 4m 30s\tremaining: 2m 32s\n",
      "1279:\tlearn: 0.2321776\ttotal: 4m 30s\tremaining: 2m 32s\n",
      "1280:\tlearn: 0.2321660\ttotal: 4m 30s\tremaining: 2m 31s\n",
      "1281:\tlearn: 0.2321557\ttotal: 4m 30s\tremaining: 2m 31s\n",
      "1282:\tlearn: 0.2321446\ttotal: 4m 30s\tremaining: 2m 31s\n",
      "1283:\tlearn: 0.2321362\ttotal: 4m 31s\tremaining: 2m 31s\n",
      "1284:\tlearn: 0.2321248\ttotal: 4m 31s\tremaining: 2m 30s\n",
      "1285:\tlearn: 0.2321121\ttotal: 4m 31s\tremaining: 2m 30s\n",
      "1286:\tlearn: 0.2321007\ttotal: 4m 31s\tremaining: 2m 30s\n",
      "1287:\tlearn: 0.2320912\ttotal: 4m 31s\tremaining: 2m 30s\n",
      "1288:\tlearn: 0.2320813\ttotal: 4m 32s\tremaining: 2m 30s\n",
      "1289:\tlearn: 0.2320671\ttotal: 4m 32s\tremaining: 2m 29s\n",
      "1290:\tlearn: 0.2320568\ttotal: 4m 32s\tremaining: 2m 29s\n",
      "1291:\tlearn: 0.2320445\ttotal: 4m 32s\tremaining: 2m 29s\n",
      "1292:\tlearn: 0.2320281\ttotal: 4m 33s\tremaining: 2m 29s\n",
      "1293:\tlearn: 0.2320157\ttotal: 4m 33s\tremaining: 2m 29s\n",
      "1294:\tlearn: 0.2320047\ttotal: 4m 33s\tremaining: 2m 28s\n",
      "1295:\tlearn: 0.2319912\ttotal: 4m 33s\tremaining: 2m 28s\n",
      "1296:\tlearn: 0.2319839\ttotal: 4m 33s\tremaining: 2m 28s\n",
      "1297:\tlearn: 0.2319751\ttotal: 4m 34s\tremaining: 2m 28s\n",
      "1298:\tlearn: 0.2319685\ttotal: 4m 34s\tremaining: 2m 28s\n",
      "1299:\tlearn: 0.2319565\ttotal: 4m 34s\tremaining: 2m 27s\n",
      "1300:\tlearn: 0.2319444\ttotal: 4m 34s\tremaining: 2m 27s\n",
      "1301:\tlearn: 0.2319370\ttotal: 4m 34s\tremaining: 2m 27s\n",
      "1302:\tlearn: 0.2319270\ttotal: 4m 35s\tremaining: 2m 27s\n",
      "1303:\tlearn: 0.2319131\ttotal: 4m 35s\tremaining: 2m 26s\n",
      "1304:\tlearn: 0.2319039\ttotal: 4m 35s\tremaining: 2m 26s\n",
      "1305:\tlearn: 0.2318930\ttotal: 4m 35s\tremaining: 2m 26s\n",
      "1306:\tlearn: 0.2318830\ttotal: 4m 35s\tremaining: 2m 26s\n",
      "1307:\tlearn: 0.2318729\ttotal: 4m 36s\tremaining: 2m 26s\n",
      "1308:\tlearn: 0.2318697\ttotal: 4m 36s\tremaining: 2m 25s\n",
      "1309:\tlearn: 0.2318538\ttotal: 4m 36s\tremaining: 2m 25s\n",
      "1310:\tlearn: 0.2318424\ttotal: 4m 36s\tremaining: 2m 25s\n",
      "1311:\tlearn: 0.2318336\ttotal: 4m 37s\tremaining: 2m 25s\n",
      "1312:\tlearn: 0.2318181\ttotal: 4m 37s\tremaining: 2m 25s\n",
      "1313:\tlearn: 0.2318105\ttotal: 4m 37s\tremaining: 2m 24s\n",
      "1314:\tlearn: 0.2318010\ttotal: 4m 37s\tremaining: 2m 24s\n",
      "1315:\tlearn: 0.2317887\ttotal: 4m 37s\tremaining: 2m 24s\n",
      "1316:\tlearn: 0.2317749\ttotal: 4m 38s\tremaining: 2m 24s\n",
      "1317:\tlearn: 0.2317652\ttotal: 4m 38s\tremaining: 2m 24s\n",
      "1318:\tlearn: 0.2317533\ttotal: 4m 38s\tremaining: 2m 23s\n",
      "1319:\tlearn: 0.2317407\ttotal: 4m 38s\tremaining: 2m 23s\n",
      "1320:\tlearn: 0.2317313\ttotal: 4m 38s\tremaining: 2m 23s\n",
      "1321:\tlearn: 0.2317245\ttotal: 4m 39s\tremaining: 2m 23s\n",
      "1322:\tlearn: 0.2317178\ttotal: 4m 39s\tremaining: 2m 22s\n",
      "1323:\tlearn: 0.2317026\ttotal: 4m 39s\tremaining: 2m 22s\n",
      "1324:\tlearn: 0.2316916\ttotal: 4m 39s\tremaining: 2m 22s\n",
      "1325:\tlearn: 0.2316826\ttotal: 4m 40s\tremaining: 2m 22s\n",
      "1326:\tlearn: 0.2316661\ttotal: 4m 40s\tremaining: 2m 22s\n",
      "1327:\tlearn: 0.2316564\ttotal: 4m 40s\tremaining: 2m 21s\n",
      "1328:\tlearn: 0.2316407\ttotal: 4m 40s\tremaining: 2m 21s\n",
      "1329:\tlearn: 0.2316273\ttotal: 4m 40s\tremaining: 2m 21s\n",
      "1330:\tlearn: 0.2316163\ttotal: 4m 41s\tremaining: 2m 21s\n",
      "1331:\tlearn: 0.2316075\ttotal: 4m 41s\tremaining: 2m 21s\n",
      "1332:\tlearn: 0.2315984\ttotal: 4m 41s\tremaining: 2m 20s\n",
      "1333:\tlearn: 0.2315911\ttotal: 4m 41s\tremaining: 2m 20s\n",
      "1334:\tlearn: 0.2315773\ttotal: 4m 41s\tremaining: 2m 20s\n",
      "1335:\tlearn: 0.2315672\ttotal: 4m 42s\tremaining: 2m 20s\n",
      "1336:\tlearn: 0.2315567\ttotal: 4m 42s\tremaining: 2m 20s\n",
      "1337:\tlearn: 0.2315442\ttotal: 4m 42s\tremaining: 2m 19s\n",
      "1338:\tlearn: 0.2315356\ttotal: 4m 42s\tremaining: 2m 19s\n",
      "1339:\tlearn: 0.2315228\ttotal: 4m 43s\tremaining: 2m 19s\n",
      "1340:\tlearn: 0.2315118\ttotal: 4m 43s\tremaining: 2m 19s\n",
      "1341:\tlearn: 0.2314993\ttotal: 4m 43s\tremaining: 2m 18s\n",
      "1342:\tlearn: 0.2314873\ttotal: 4m 43s\tremaining: 2m 18s\n",
      "1343:\tlearn: 0.2314747\ttotal: 4m 43s\tremaining: 2m 18s\n",
      "1344:\tlearn: 0.2314608\ttotal: 4m 44s\tremaining: 2m 18s\n",
      "1345:\tlearn: 0.2314525\ttotal: 4m 44s\tremaining: 2m 18s\n",
      "1346:\tlearn: 0.2314400\ttotal: 4m 44s\tremaining: 2m 17s\n",
      "1347:\tlearn: 0.2314260\ttotal: 4m 44s\tremaining: 2m 17s\n",
      "1348:\tlearn: 0.2314109\ttotal: 4m 44s\tremaining: 2m 17s\n",
      "1349:\tlearn: 0.2313997\ttotal: 4m 45s\tremaining: 2m 17s\n",
      "1350:\tlearn: 0.2313871\ttotal: 4m 45s\tremaining: 2m 17s\n",
      "1351:\tlearn: 0.2313777\ttotal: 4m 45s\tremaining: 2m 16s\n",
      "1352:\tlearn: 0.2313666\ttotal: 4m 45s\tremaining: 2m 16s\n",
      "1353:\tlearn: 0.2313509\ttotal: 4m 46s\tremaining: 2m 16s\n",
      "1354:\tlearn: 0.2313365\ttotal: 4m 46s\tremaining: 2m 16s\n",
      "1355:\tlearn: 0.2313189\ttotal: 4m 46s\tremaining: 2m 16s\n",
      "1356:\tlearn: 0.2313111\ttotal: 4m 46s\tremaining: 2m 15s\n",
      "1357:\tlearn: 0.2312990\ttotal: 4m 46s\tremaining: 2m 15s\n",
      "1358:\tlearn: 0.2312906\ttotal: 4m 47s\tremaining: 2m 15s\n",
      "1359:\tlearn: 0.2312863\ttotal: 4m 47s\tremaining: 2m 15s\n",
      "1360:\tlearn: 0.2312770\ttotal: 4m 47s\tremaining: 2m 14s\n",
      "1361:\tlearn: 0.2312635\ttotal: 4m 47s\tremaining: 2m 14s\n",
      "1362:\tlearn: 0.2312532\ttotal: 4m 47s\tremaining: 2m 14s\n",
      "1363:\tlearn: 0.2312440\ttotal: 4m 48s\tremaining: 2m 14s\n",
      "1364:\tlearn: 0.2312339\ttotal: 4m 48s\tremaining: 2m 14s\n",
      "1365:\tlearn: 0.2312215\ttotal: 4m 48s\tremaining: 2m 13s\n",
      "1366:\tlearn: 0.2312081\ttotal: 4m 48s\tremaining: 2m 13s\n",
      "1367:\tlearn: 0.2312006\ttotal: 4m 48s\tremaining: 2m 13s\n",
      "1368:\tlearn: 0.2311925\ttotal: 4m 49s\tremaining: 2m 13s\n",
      "1369:\tlearn: 0.2311839\ttotal: 4m 49s\tremaining: 2m 13s\n",
      "1370:\tlearn: 0.2311728\ttotal: 4m 49s\tremaining: 2m 12s\n",
      "1371:\tlearn: 0.2311614\ttotal: 4m 49s\tremaining: 2m 12s\n",
      "1372:\tlearn: 0.2311475\ttotal: 4m 50s\tremaining: 2m 12s\n",
      "1373:\tlearn: 0.2311375\ttotal: 4m 50s\tremaining: 2m 12s\n",
      "1374:\tlearn: 0.2311282\ttotal: 4m 50s\tremaining: 2m 12s\n",
      "1375:\tlearn: 0.2311187\ttotal: 4m 50s\tremaining: 2m 11s\n",
      "1376:\tlearn: 0.2311076\ttotal: 4m 50s\tremaining: 2m 11s\n",
      "1377:\tlearn: 0.2310973\ttotal: 4m 51s\tremaining: 2m 11s\n",
      "1378:\tlearn: 0.2310879\ttotal: 4m 51s\tremaining: 2m 11s\n",
      "1379:\tlearn: 0.2310764\ttotal: 4m 51s\tremaining: 2m 10s\n",
      "1380:\tlearn: 0.2310667\ttotal: 4m 51s\tremaining: 2m 10s\n",
      "1381:\tlearn: 0.2310593\ttotal: 4m 51s\tremaining: 2m 10s\n",
      "1382:\tlearn: 0.2310510\ttotal: 4m 52s\tremaining: 2m 10s\n",
      "1383:\tlearn: 0.2310421\ttotal: 4m 52s\tremaining: 2m 10s\n",
      "1384:\tlearn: 0.2310314\ttotal: 4m 52s\tremaining: 2m 9s\n",
      "1385:\tlearn: 0.2310202\ttotal: 4m 52s\tremaining: 2m 9s\n",
      "1386:\tlearn: 0.2310112\ttotal: 4m 52s\tremaining: 2m 9s\n",
      "1387:\tlearn: 0.2309990\ttotal: 4m 53s\tremaining: 2m 9s\n",
      "1388:\tlearn: 0.2309884\ttotal: 4m 53s\tremaining: 2m 9s\n",
      "1389:\tlearn: 0.2309765\ttotal: 4m 53s\tremaining: 2m 8s\n",
      "1390:\tlearn: 0.2309654\ttotal: 4m 53s\tremaining: 2m 8s\n",
      "1391:\tlearn: 0.2309564\ttotal: 4m 54s\tremaining: 2m 8s\n",
      "1392:\tlearn: 0.2309482\ttotal: 4m 54s\tremaining: 2m 8s\n",
      "1393:\tlearn: 0.2309377\ttotal: 4m 54s\tremaining: 2m 8s\n",
      "1394:\tlearn: 0.2309282\ttotal: 4m 54s\tremaining: 2m 7s\n",
      "1395:\tlearn: 0.2309163\ttotal: 4m 54s\tremaining: 2m 7s\n",
      "1396:\tlearn: 0.2309078\ttotal: 4m 55s\tremaining: 2m 7s\n",
      "1397:\tlearn: 0.2308921\ttotal: 4m 55s\tremaining: 2m 7s\n",
      "1398:\tlearn: 0.2308810\ttotal: 4m 55s\tremaining: 2m 6s\n",
      "1399:\tlearn: 0.2308698\ttotal: 4m 55s\tremaining: 2m 6s\n",
      "1400:\tlearn: 0.2308583\ttotal: 4m 55s\tremaining: 2m 6s\n",
      "1401:\tlearn: 0.2308453\ttotal: 4m 56s\tremaining: 2m 6s\n",
      "1402:\tlearn: 0.2308359\ttotal: 4m 56s\tremaining: 2m 6s\n",
      "1403:\tlearn: 0.2308279\ttotal: 4m 56s\tremaining: 2m 5s\n",
      "1404:\tlearn: 0.2308187\ttotal: 4m 56s\tremaining: 2m 5s\n",
      "1405:\tlearn: 0.2308125\ttotal: 4m 57s\tremaining: 2m 5s\n",
      "1406:\tlearn: 0.2308017\ttotal: 4m 57s\tremaining: 2m 5s\n",
      "1407:\tlearn: 0.2307910\ttotal: 4m 57s\tremaining: 2m 5s\n",
      "1408:\tlearn: 0.2307803\ttotal: 4m 57s\tremaining: 2m 4s\n",
      "1409:\tlearn: 0.2307678\ttotal: 4m 57s\tremaining: 2m 4s\n",
      "1410:\tlearn: 0.2307548\ttotal: 4m 58s\tremaining: 2m 4s\n",
      "1411:\tlearn: 0.2307417\ttotal: 4m 58s\tremaining: 2m 4s\n",
      "1412:\tlearn: 0.2307232\ttotal: 4m 58s\tremaining: 2m 4s\n",
      "1413:\tlearn: 0.2307105\ttotal: 4m 58s\tremaining: 2m 3s\n",
      "1414:\tlearn: 0.2307010\ttotal: 4m 58s\tremaining: 2m 3s\n",
      "1415:\tlearn: 0.2306900\ttotal: 4m 59s\tremaining: 2m 3s\n",
      "1416:\tlearn: 0.2306809\ttotal: 4m 59s\tremaining: 2m 3s\n",
      "1417:\tlearn: 0.2306673\ttotal: 4m 59s\tremaining: 2m 2s\n",
      "1418:\tlearn: 0.2306562\ttotal: 4m 59s\tremaining: 2m 2s\n",
      "1419:\tlearn: 0.2306495\ttotal: 4m 59s\tremaining: 2m 2s\n",
      "1420:\tlearn: 0.2306399\ttotal: 5m\tremaining: 2m 2s\n",
      "1421:\tlearn: 0.2306290\ttotal: 5m\tremaining: 2m 2s\n",
      "1422:\tlearn: 0.2306160\ttotal: 5m\tremaining: 2m 1s\n",
      "1423:\tlearn: 0.2305925\ttotal: 5m\tremaining: 2m 1s\n",
      "1424:\tlearn: 0.2305863\ttotal: 5m 1s\tremaining: 2m 1s\n",
      "1425:\tlearn: 0.2305733\ttotal: 5m 1s\tremaining: 2m 1s\n",
      "1426:\tlearn: 0.2305640\ttotal: 5m 1s\tremaining: 2m 1s\n",
      "1427:\tlearn: 0.2305546\ttotal: 5m 1s\tremaining: 2m\n",
      "1428:\tlearn: 0.2305480\ttotal: 5m 1s\tremaining: 2m\n",
      "1429:\tlearn: 0.2305336\ttotal: 5m 2s\tremaining: 2m\n",
      "1430:\tlearn: 0.2305215\ttotal: 5m 2s\tremaining: 2m\n",
      "1431:\tlearn: 0.2305123\ttotal: 5m 2s\tremaining: 2m\n",
      "1432:\tlearn: 0.2304957\ttotal: 5m 2s\tremaining: 1m 59s\n",
      "1433:\tlearn: 0.2304855\ttotal: 5m 2s\tremaining: 1m 59s\n",
      "1434:\tlearn: 0.2304773\ttotal: 5m 3s\tremaining: 1m 59s\n",
      "1435:\tlearn: 0.2304685\ttotal: 5m 3s\tremaining: 1m 59s\n",
      "1436:\tlearn: 0.2304582\ttotal: 5m 3s\tremaining: 1m 58s\n",
      "1437:\tlearn: 0.2304463\ttotal: 5m 3s\tremaining: 1m 58s\n",
      "1438:\tlearn: 0.2304312\ttotal: 5m 4s\tremaining: 1m 58s\n",
      "1439:\tlearn: 0.2304231\ttotal: 5m 4s\tremaining: 1m 58s\n",
      "1440:\tlearn: 0.2304128\ttotal: 5m 4s\tremaining: 1m 58s\n",
      "1441:\tlearn: 0.2304018\ttotal: 5m 4s\tremaining: 1m 57s\n",
      "1442:\tlearn: 0.2303916\ttotal: 5m 4s\tremaining: 1m 57s\n",
      "1443:\tlearn: 0.2303808\ttotal: 5m 5s\tremaining: 1m 57s\n",
      "1444:\tlearn: 0.2303723\ttotal: 5m 5s\tremaining: 1m 57s\n",
      "1445:\tlearn: 0.2303625\ttotal: 5m 5s\tremaining: 1m 57s\n",
      "1446:\tlearn: 0.2303458\ttotal: 5m 5s\tremaining: 1m 56s\n",
      "1447:\tlearn: 0.2303341\ttotal: 5m 5s\tremaining: 1m 56s\n",
      "1448:\tlearn: 0.2303223\ttotal: 5m 6s\tremaining: 1m 56s\n",
      "1449:\tlearn: 0.2303138\ttotal: 5m 6s\tremaining: 1m 56s\n",
      "1450:\tlearn: 0.2303084\ttotal: 5m 6s\tremaining: 1m 56s\n",
      "1451:\tlearn: 0.2302981\ttotal: 5m 6s\tremaining: 1m 55s\n",
      "1452:\tlearn: 0.2302899\ttotal: 5m 7s\tremaining: 1m 55s\n",
      "1453:\tlearn: 0.2302768\ttotal: 5m 7s\tremaining: 1m 55s\n",
      "1454:\tlearn: 0.2302653\ttotal: 5m 7s\tremaining: 1m 55s\n",
      "1455:\tlearn: 0.2302547\ttotal: 5m 7s\tremaining: 1m 54s\n",
      "1456:\tlearn: 0.2302429\ttotal: 5m 7s\tremaining: 1m 54s\n",
      "1457:\tlearn: 0.2302309\ttotal: 5m 8s\tremaining: 1m 54s\n",
      "1458:\tlearn: 0.2302171\ttotal: 5m 8s\tremaining: 1m 54s\n",
      "1459:\tlearn: 0.2302043\ttotal: 5m 8s\tremaining: 1m 54s\n",
      "1460:\tlearn: 0.2301918\ttotal: 5m 8s\tremaining: 1m 53s\n",
      "1461:\tlearn: 0.2301824\ttotal: 5m 8s\tremaining: 1m 53s\n",
      "1462:\tlearn: 0.2301737\ttotal: 5m 9s\tremaining: 1m 53s\n",
      "1463:\tlearn: 0.2301607\ttotal: 5m 9s\tremaining: 1m 53s\n",
      "1464:\tlearn: 0.2301492\ttotal: 5m 9s\tremaining: 1m 53s\n",
      "1465:\tlearn: 0.2301351\ttotal: 5m 9s\tremaining: 1m 52s\n",
      "1466:\tlearn: 0.2301250\ttotal: 5m 9s\tremaining: 1m 52s\n",
      "1467:\tlearn: 0.2301153\ttotal: 5m 10s\tremaining: 1m 52s\n",
      "1468:\tlearn: 0.2301024\ttotal: 5m 10s\tremaining: 1m 52s\n",
      "1469:\tlearn: 0.2300928\ttotal: 5m 10s\tremaining: 1m 51s\n",
      "1470:\tlearn: 0.2300830\ttotal: 5m 10s\tremaining: 1m 51s\n",
      "1471:\tlearn: 0.2300688\ttotal: 5m 11s\tremaining: 1m 51s\n",
      "1472:\tlearn: 0.2300543\ttotal: 5m 11s\tremaining: 1m 51s\n",
      "1473:\tlearn: 0.2300473\ttotal: 5m 11s\tremaining: 1m 51s\n",
      "1474:\tlearn: 0.2300361\ttotal: 5m 11s\tremaining: 1m 50s\n",
      "1475:\tlearn: 0.2300272\ttotal: 5m 11s\tremaining: 1m 50s\n",
      "1476:\tlearn: 0.2300154\ttotal: 5m 12s\tremaining: 1m 50s\n",
      "1477:\tlearn: 0.2300009\ttotal: 5m 12s\tremaining: 1m 50s\n",
      "1478:\tlearn: 0.2299919\ttotal: 5m 12s\tremaining: 1m 50s\n",
      "1479:\tlearn: 0.2299840\ttotal: 5m 12s\tremaining: 1m 49s\n",
      "1480:\tlearn: 0.2299727\ttotal: 5m 12s\tremaining: 1m 49s\n",
      "1481:\tlearn: 0.2299608\ttotal: 5m 13s\tremaining: 1m 49s\n",
      "1482:\tlearn: 0.2299461\ttotal: 5m 13s\tremaining: 1m 49s\n",
      "1483:\tlearn: 0.2299346\ttotal: 5m 13s\tremaining: 1m 49s\n",
      "1484:\tlearn: 0.2299279\ttotal: 5m 13s\tremaining: 1m 48s\n",
      "1485:\tlearn: 0.2299177\ttotal: 5m 13s\tremaining: 1m 48s\n",
      "1486:\tlearn: 0.2299042\ttotal: 5m 14s\tremaining: 1m 48s\n",
      "1487:\tlearn: 0.2298934\ttotal: 5m 14s\tremaining: 1m 48s\n",
      "1488:\tlearn: 0.2298841\ttotal: 5m 14s\tremaining: 1m 47s\n",
      "1489:\tlearn: 0.2298727\ttotal: 5m 14s\tremaining: 1m 47s\n",
      "1490:\tlearn: 0.2298623\ttotal: 5m 15s\tremaining: 1m 47s\n",
      "1491:\tlearn: 0.2298542\ttotal: 5m 15s\tremaining: 1m 47s\n",
      "1492:\tlearn: 0.2298393\ttotal: 5m 15s\tremaining: 1m 47s\n",
      "1493:\tlearn: 0.2298294\ttotal: 5m 15s\tremaining: 1m 46s\n",
      "1494:\tlearn: 0.2298196\ttotal: 5m 15s\tremaining: 1m 46s\n",
      "1495:\tlearn: 0.2298084\ttotal: 5m 16s\tremaining: 1m 46s\n",
      "1496:\tlearn: 0.2297975\ttotal: 5m 16s\tremaining: 1m 46s\n",
      "1497:\tlearn: 0.2297858\ttotal: 5m 16s\tremaining: 1m 46s\n",
      "1498:\tlearn: 0.2297723\ttotal: 5m 16s\tremaining: 1m 45s\n",
      "1499:\tlearn: 0.2297606\ttotal: 5m 16s\tremaining: 1m 45s\n",
      "1500:\tlearn: 0.2297484\ttotal: 5m 17s\tremaining: 1m 45s\n",
      "1501:\tlearn: 0.2297396\ttotal: 5m 17s\tremaining: 1m 45s\n",
      "1502:\tlearn: 0.2297281\ttotal: 5m 17s\tremaining: 1m 45s\n",
      "1503:\tlearn: 0.2297196\ttotal: 5m 17s\tremaining: 1m 44s\n",
      "1504:\tlearn: 0.2297059\ttotal: 5m 18s\tremaining: 1m 44s\n",
      "1505:\tlearn: 0.2296979\ttotal: 5m 18s\tremaining: 1m 44s\n",
      "1506:\tlearn: 0.2296895\ttotal: 5m 18s\tremaining: 1m 44s\n",
      "1507:\tlearn: 0.2296728\ttotal: 5m 18s\tremaining: 1m 43s\n",
      "1508:\tlearn: 0.2296595\ttotal: 5m 18s\tremaining: 1m 43s\n",
      "1509:\tlearn: 0.2296475\ttotal: 5m 19s\tremaining: 1m 43s\n",
      "1510:\tlearn: 0.2296349\ttotal: 5m 19s\tremaining: 1m 43s\n",
      "1511:\tlearn: 0.2296269\ttotal: 5m 19s\tremaining: 1m 43s\n",
      "1512:\tlearn: 0.2296157\ttotal: 5m 19s\tremaining: 1m 42s\n",
      "1513:\tlearn: 0.2296053\ttotal: 5m 19s\tremaining: 1m 42s\n",
      "1514:\tlearn: 0.2295960\ttotal: 5m 20s\tremaining: 1m 42s\n",
      "1515:\tlearn: 0.2295868\ttotal: 5m 20s\tremaining: 1m 42s\n",
      "1516:\tlearn: 0.2295744\ttotal: 5m 20s\tremaining: 1m 42s\n",
      "1517:\tlearn: 0.2295610\ttotal: 5m 20s\tremaining: 1m 41s\n",
      "1518:\tlearn: 0.2295519\ttotal: 5m 20s\tremaining: 1m 41s\n",
      "1519:\tlearn: 0.2295393\ttotal: 5m 21s\tremaining: 1m 41s\n",
      "1520:\tlearn: 0.2295275\ttotal: 5m 21s\tremaining: 1m 41s\n",
      "1521:\tlearn: 0.2295172\ttotal: 5m 21s\tremaining: 1m 41s\n",
      "1522:\tlearn: 0.2295053\ttotal: 5m 21s\tremaining: 1m 40s\n",
      "1523:\tlearn: 0.2294942\ttotal: 5m 22s\tremaining: 1m 40s\n",
      "1524:\tlearn: 0.2294866\ttotal: 5m 22s\tremaining: 1m 40s\n",
      "1525:\tlearn: 0.2294702\ttotal: 5m 22s\tremaining: 1m 40s\n",
      "1526:\tlearn: 0.2294624\ttotal: 5m 22s\tremaining: 1m 39s\n",
      "1527:\tlearn: 0.2294539\ttotal: 5m 22s\tremaining: 1m 39s\n",
      "1528:\tlearn: 0.2294442\ttotal: 5m 23s\tremaining: 1m 39s\n",
      "1529:\tlearn: 0.2294306\ttotal: 5m 23s\tremaining: 1m 39s\n",
      "1530:\tlearn: 0.2294184\ttotal: 5m 23s\tremaining: 1m 39s\n",
      "1531:\tlearn: 0.2294110\ttotal: 5m 23s\tremaining: 1m 38s\n",
      "1532:\tlearn: 0.2293989\ttotal: 5m 23s\tremaining: 1m 38s\n",
      "1533:\tlearn: 0.2293899\ttotal: 5m 24s\tremaining: 1m 38s\n",
      "1534:\tlearn: 0.2293778\ttotal: 5m 24s\tremaining: 1m 38s\n",
      "1535:\tlearn: 0.2293631\ttotal: 5m 24s\tremaining: 1m 38s\n",
      "1536:\tlearn: 0.2293511\ttotal: 5m 24s\tremaining: 1m 37s\n",
      "1537:\tlearn: 0.2293342\ttotal: 5m 25s\tremaining: 1m 37s\n",
      "1538:\tlearn: 0.2293250\ttotal: 5m 25s\tremaining: 1m 37s\n",
      "1539:\tlearn: 0.2293127\ttotal: 5m 25s\tremaining: 1m 37s\n",
      "1540:\tlearn: 0.2292987\ttotal: 5m 25s\tremaining: 1m 37s\n",
      "1541:\tlearn: 0.2292876\ttotal: 5m 25s\tremaining: 1m 36s\n",
      "1542:\tlearn: 0.2292737\ttotal: 5m 26s\tremaining: 1m 36s\n",
      "1543:\tlearn: 0.2292636\ttotal: 5m 26s\tremaining: 1m 36s\n",
      "1544:\tlearn: 0.2292560\ttotal: 5m 26s\tremaining: 1m 36s\n",
      "1545:\tlearn: 0.2292479\ttotal: 5m 26s\tremaining: 1m 35s\n",
      "1546:\tlearn: 0.2292360\ttotal: 5m 26s\tremaining: 1m 35s\n",
      "1547:\tlearn: 0.2292180\ttotal: 5m 27s\tremaining: 1m 35s\n",
      "1548:\tlearn: 0.2292109\ttotal: 5m 27s\tremaining: 1m 35s\n",
      "1549:\tlearn: 0.2292015\ttotal: 5m 27s\tremaining: 1m 35s\n",
      "1550:\tlearn: 0.2291949\ttotal: 5m 27s\tremaining: 1m 34s\n",
      "1551:\tlearn: 0.2291836\ttotal: 5m 27s\tremaining: 1m 34s\n",
      "1552:\tlearn: 0.2291705\ttotal: 5m 28s\tremaining: 1m 34s\n",
      "1553:\tlearn: 0.2291587\ttotal: 5m 28s\tremaining: 1m 34s\n",
      "1554:\tlearn: 0.2291436\ttotal: 5m 28s\tremaining: 1m 34s\n",
      "1555:\tlearn: 0.2291396\ttotal: 5m 28s\tremaining: 1m 33s\n",
      "1556:\tlearn: 0.2291296\ttotal: 5m 29s\tremaining: 1m 33s\n",
      "1557:\tlearn: 0.2291167\ttotal: 5m 29s\tremaining: 1m 33s\n",
      "1558:\tlearn: 0.2291048\ttotal: 5m 29s\tremaining: 1m 33s\n",
      "1559:\tlearn: 0.2290973\ttotal: 5m 29s\tremaining: 1m 32s\n",
      "1560:\tlearn: 0.2290939\ttotal: 5m 29s\tremaining: 1m 32s\n",
      "1561:\tlearn: 0.2290824\ttotal: 5m 30s\tremaining: 1m 32s\n",
      "1562:\tlearn: 0.2290685\ttotal: 5m 30s\tremaining: 1m 32s\n",
      "1563:\tlearn: 0.2290577\ttotal: 5m 30s\tremaining: 1m 32s\n",
      "1564:\tlearn: 0.2290444\ttotal: 5m 30s\tremaining: 1m 31s\n",
      "1565:\tlearn: 0.2290311\ttotal: 5m 30s\tremaining: 1m 31s\n",
      "1566:\tlearn: 0.2290205\ttotal: 5m 31s\tremaining: 1m 31s\n",
      "1567:\tlearn: 0.2290028\ttotal: 5m 31s\tremaining: 1m 31s\n",
      "1568:\tlearn: 0.2289920\ttotal: 5m 31s\tremaining: 1m 31s\n",
      "1569:\tlearn: 0.2289834\ttotal: 5m 31s\tremaining: 1m 30s\n",
      "1570:\tlearn: 0.2289736\ttotal: 5m 32s\tremaining: 1m 30s\n",
      "1571:\tlearn: 0.2289625\ttotal: 5m 32s\tremaining: 1m 30s\n",
      "1572:\tlearn: 0.2289508\ttotal: 5m 32s\tremaining: 1m 30s\n",
      "1573:\tlearn: 0.2289340\ttotal: 5m 32s\tremaining: 1m 30s\n",
      "1574:\tlearn: 0.2289248\ttotal: 5m 32s\tremaining: 1m 29s\n",
      "1575:\tlearn: 0.2289125\ttotal: 5m 33s\tremaining: 1m 29s\n",
      "1576:\tlearn: 0.2289037\ttotal: 5m 33s\tremaining: 1m 29s\n",
      "1577:\tlearn: 0.2288940\ttotal: 5m 33s\tremaining: 1m 29s\n",
      "1578:\tlearn: 0.2288863\ttotal: 5m 33s\tremaining: 1m 28s\n",
      "1579:\tlearn: 0.2288752\ttotal: 5m 33s\tremaining: 1m 28s\n",
      "1580:\tlearn: 0.2288665\ttotal: 5m 34s\tremaining: 1m 28s\n",
      "1581:\tlearn: 0.2288547\ttotal: 5m 34s\tremaining: 1m 28s\n",
      "1582:\tlearn: 0.2288446\ttotal: 5m 34s\tremaining: 1m 28s\n",
      "1583:\tlearn: 0.2288358\ttotal: 5m 34s\tremaining: 1m 27s\n",
      "1584:\tlearn: 0.2288282\ttotal: 5m 35s\tremaining: 1m 27s\n",
      "1585:\tlearn: 0.2288199\ttotal: 5m 35s\tremaining: 1m 27s\n",
      "1586:\tlearn: 0.2288070\ttotal: 5m 35s\tremaining: 1m 27s\n",
      "1587:\tlearn: 0.2287954\ttotal: 5m 35s\tremaining: 1m 27s\n",
      "1588:\tlearn: 0.2287841\ttotal: 5m 35s\tremaining: 1m 26s\n",
      "1589:\tlearn: 0.2287734\ttotal: 5m 36s\tremaining: 1m 26s\n",
      "1590:\tlearn: 0.2287619\ttotal: 5m 36s\tremaining: 1m 26s\n",
      "1591:\tlearn: 0.2287555\ttotal: 5m 36s\tremaining: 1m 26s\n",
      "1592:\tlearn: 0.2287415\ttotal: 5m 36s\tremaining: 1m 26s\n",
      "1593:\tlearn: 0.2287321\ttotal: 5m 36s\tremaining: 1m 25s\n",
      "1594:\tlearn: 0.2287213\ttotal: 5m 37s\tremaining: 1m 25s\n",
      "1595:\tlearn: 0.2287074\ttotal: 5m 37s\tremaining: 1m 25s\n",
      "1596:\tlearn: 0.2286925\ttotal: 5m 37s\tremaining: 1m 25s\n",
      "1597:\tlearn: 0.2286820\ttotal: 5m 37s\tremaining: 1m 24s\n",
      "1598:\tlearn: 0.2286733\ttotal: 5m 38s\tremaining: 1m 24s\n",
      "1599:\tlearn: 0.2286637\ttotal: 5m 38s\tremaining: 1m 24s\n",
      "1600:\tlearn: 0.2286546\ttotal: 5m 38s\tremaining: 1m 24s\n",
      "1601:\tlearn: 0.2286471\ttotal: 5m 38s\tremaining: 1m 24s\n",
      "1602:\tlearn: 0.2286352\ttotal: 5m 38s\tremaining: 1m 23s\n",
      "1603:\tlearn: 0.2286264\ttotal: 5m 39s\tremaining: 1m 23s\n",
      "1604:\tlearn: 0.2286126\ttotal: 5m 39s\tremaining: 1m 23s\n",
      "1605:\tlearn: 0.2286047\ttotal: 5m 39s\tremaining: 1m 23s\n",
      "1606:\tlearn: 0.2286003\ttotal: 5m 39s\tremaining: 1m 23s\n",
      "1607:\tlearn: 0.2285898\ttotal: 5m 39s\tremaining: 1m 22s\n",
      "1608:\tlearn: 0.2285823\ttotal: 5m 40s\tremaining: 1m 22s\n",
      "1609:\tlearn: 0.2285725\ttotal: 5m 40s\tremaining: 1m 22s\n",
      "1610:\tlearn: 0.2285647\ttotal: 5m 40s\tremaining: 1m 22s\n",
      "1611:\tlearn: 0.2285534\ttotal: 5m 40s\tremaining: 1m 22s\n",
      "1612:\tlearn: 0.2285421\ttotal: 5m 40s\tremaining: 1m 21s\n",
      "1613:\tlearn: 0.2285308\ttotal: 5m 41s\tremaining: 1m 21s\n",
      "1614:\tlearn: 0.2285173\ttotal: 5m 41s\tremaining: 1m 21s\n",
      "1615:\tlearn: 0.2285056\ttotal: 5m 41s\tremaining: 1m 21s\n",
      "1616:\tlearn: 0.2284947\ttotal: 5m 41s\tremaining: 1m 20s\n",
      "1617:\tlearn: 0.2284771\ttotal: 5m 42s\tremaining: 1m 20s\n",
      "1618:\tlearn: 0.2284671\ttotal: 5m 42s\tremaining: 1m 20s\n",
      "1619:\tlearn: 0.2284638\ttotal: 5m 42s\tremaining: 1m 20s\n",
      "1620:\tlearn: 0.2284501\ttotal: 5m 42s\tremaining: 1m 20s\n",
      "1621:\tlearn: 0.2284361\ttotal: 5m 42s\tremaining: 1m 19s\n",
      "1622:\tlearn: 0.2284227\ttotal: 5m 43s\tremaining: 1m 19s\n",
      "1623:\tlearn: 0.2284155\ttotal: 5m 43s\tremaining: 1m 19s\n",
      "1624:\tlearn: 0.2284073\ttotal: 5m 43s\tremaining: 1m 19s\n",
      "1625:\tlearn: 0.2283966\ttotal: 5m 43s\tremaining: 1m 19s\n",
      "1626:\tlearn: 0.2283818\ttotal: 5m 43s\tremaining: 1m 18s\n",
      "1627:\tlearn: 0.2283719\ttotal: 5m 44s\tremaining: 1m 18s\n",
      "1628:\tlearn: 0.2283647\ttotal: 5m 44s\tremaining: 1m 18s\n",
      "1629:\tlearn: 0.2283549\ttotal: 5m 44s\tremaining: 1m 18s\n",
      "1630:\tlearn: 0.2283409\ttotal: 5m 44s\tremaining: 1m 18s\n",
      "1631:\tlearn: 0.2283303\ttotal: 5m 45s\tremaining: 1m 17s\n",
      "1632:\tlearn: 0.2283209\ttotal: 5m 45s\tremaining: 1m 17s\n",
      "1633:\tlearn: 0.2283125\ttotal: 5m 45s\tremaining: 1m 17s\n",
      "1634:\tlearn: 0.2283042\ttotal: 5m 45s\tremaining: 1m 17s\n",
      "1635:\tlearn: 0.2282898\ttotal: 5m 45s\tremaining: 1m 16s\n",
      "1636:\tlearn: 0.2282796\ttotal: 5m 46s\tremaining: 1m 16s\n",
      "1637:\tlearn: 0.2282700\ttotal: 5m 46s\tremaining: 1m 16s\n",
      "1638:\tlearn: 0.2282597\ttotal: 5m 46s\tremaining: 1m 16s\n",
      "1639:\tlearn: 0.2282506\ttotal: 5m 46s\tremaining: 1m 16s\n",
      "1640:\tlearn: 0.2282363\ttotal: 5m 46s\tremaining: 1m 15s\n",
      "1641:\tlearn: 0.2282266\ttotal: 5m 47s\tremaining: 1m 15s\n",
      "1642:\tlearn: 0.2282181\ttotal: 5m 47s\tremaining: 1m 15s\n",
      "1643:\tlearn: 0.2282080\ttotal: 5m 47s\tremaining: 1m 15s\n",
      "1644:\tlearn: 0.2281972\ttotal: 5m 47s\tremaining: 1m 15s\n",
      "1645:\tlearn: 0.2281874\ttotal: 5m 48s\tremaining: 1m 14s\n",
      "1646:\tlearn: 0.2281759\ttotal: 5m 48s\tremaining: 1m 14s\n",
      "1647:\tlearn: 0.2281686\ttotal: 5m 48s\tremaining: 1m 14s\n",
      "1648:\tlearn: 0.2281560\ttotal: 5m 48s\tremaining: 1m 14s\n",
      "1649:\tlearn: 0.2281402\ttotal: 5m 48s\tremaining: 1m 13s\n",
      "1650:\tlearn: 0.2281303\ttotal: 5m 49s\tremaining: 1m 13s\n",
      "1651:\tlearn: 0.2281200\ttotal: 5m 49s\tremaining: 1m 13s\n",
      "1652:\tlearn: 0.2281102\ttotal: 5m 49s\tremaining: 1m 13s\n",
      "1653:\tlearn: 0.2281033\ttotal: 5m 49s\tremaining: 1m 13s\n",
      "1654:\tlearn: 0.2280925\ttotal: 5m 49s\tremaining: 1m 12s\n",
      "1655:\tlearn: 0.2280832\ttotal: 5m 50s\tremaining: 1m 12s\n",
      "1656:\tlearn: 0.2280717\ttotal: 5m 50s\tremaining: 1m 12s\n",
      "1657:\tlearn: 0.2280603\ttotal: 5m 50s\tremaining: 1m 12s\n",
      "1658:\tlearn: 0.2280514\ttotal: 5m 50s\tremaining: 1m 12s\n",
      "1659:\tlearn: 0.2280414\ttotal: 5m 50s\tremaining: 1m 11s\n",
      "1660:\tlearn: 0.2280318\ttotal: 5m 51s\tremaining: 1m 11s\n",
      "1661:\tlearn: 0.2280204\ttotal: 5m 51s\tremaining: 1m 11s\n",
      "1662:\tlearn: 0.2280108\ttotal: 5m 51s\tremaining: 1m 11s\n",
      "1663:\tlearn: 0.2280018\ttotal: 5m 51s\tremaining: 1m 11s\n",
      "1664:\tlearn: 0.2279900\ttotal: 5m 52s\tremaining: 1m 10s\n",
      "1665:\tlearn: 0.2279764\ttotal: 5m 52s\tremaining: 1m 10s\n",
      "1666:\tlearn: 0.2279680\ttotal: 5m 52s\tremaining: 1m 10s\n",
      "1667:\tlearn: 0.2279566\ttotal: 5m 52s\tremaining: 1m 10s\n",
      "1668:\tlearn: 0.2279442\ttotal: 5m 52s\tremaining: 1m 9s\n",
      "1669:\tlearn: 0.2279352\ttotal: 5m 53s\tremaining: 1m 9s\n",
      "1670:\tlearn: 0.2279250\ttotal: 5m 53s\tremaining: 1m 9s\n",
      "1671:\tlearn: 0.2279159\ttotal: 5m 53s\tremaining: 1m 9s\n",
      "1672:\tlearn: 0.2279018\ttotal: 5m 53s\tremaining: 1m 9s\n",
      "1673:\tlearn: 0.2278886\ttotal: 5m 53s\tremaining: 1m 8s\n",
      "1674:\tlearn: 0.2278804\ttotal: 5m 54s\tremaining: 1m 8s\n",
      "1675:\tlearn: 0.2278689\ttotal: 5m 54s\tremaining: 1m 8s\n",
      "1676:\tlearn: 0.2278577\ttotal: 5m 54s\tremaining: 1m 8s\n",
      "1677:\tlearn: 0.2278431\ttotal: 5m 54s\tremaining: 1m 8s\n",
      "1678:\tlearn: 0.2278327\ttotal: 5m 55s\tremaining: 1m 7s\n",
      "1679:\tlearn: 0.2278250\ttotal: 5m 55s\tremaining: 1m 7s\n",
      "1680:\tlearn: 0.2278100\ttotal: 5m 55s\tremaining: 1m 7s\n",
      "1681:\tlearn: 0.2277959\ttotal: 5m 55s\tremaining: 1m 7s\n",
      "1682:\tlearn: 0.2277854\ttotal: 5m 55s\tremaining: 1m 7s\n",
      "1683:\tlearn: 0.2277729\ttotal: 5m 56s\tremaining: 1m 6s\n",
      "1684:\tlearn: 0.2277565\ttotal: 5m 56s\tremaining: 1m 6s\n",
      "1685:\tlearn: 0.2277464\ttotal: 5m 56s\tremaining: 1m 6s\n",
      "1686:\tlearn: 0.2277384\ttotal: 5m 56s\tremaining: 1m 6s\n",
      "1687:\tlearn: 0.2277255\ttotal: 5m 56s\tremaining: 1m 5s\n",
      "1688:\tlearn: 0.2277130\ttotal: 5m 57s\tremaining: 1m 5s\n",
      "1689:\tlearn: 0.2276996\ttotal: 5m 57s\tremaining: 1m 5s\n",
      "1690:\tlearn: 0.2276940\ttotal: 5m 57s\tremaining: 1m 5s\n",
      "1691:\tlearn: 0.2276871\ttotal: 5m 57s\tremaining: 1m 5s\n",
      "1692:\tlearn: 0.2276743\ttotal: 5m 57s\tremaining: 1m 4s\n",
      "1693:\tlearn: 0.2276640\ttotal: 5m 58s\tremaining: 1m 4s\n",
      "1694:\tlearn: 0.2276549\ttotal: 5m 58s\tremaining: 1m 4s\n",
      "1695:\tlearn: 0.2276422\ttotal: 5m 58s\tremaining: 1m 4s\n",
      "1696:\tlearn: 0.2276320\ttotal: 5m 58s\tremaining: 1m 4s\n",
      "1697:\tlearn: 0.2276235\ttotal: 5m 59s\tremaining: 1m 3s\n",
      "1698:\tlearn: 0.2276117\ttotal: 5m 59s\tremaining: 1m 3s\n",
      "1699:\tlearn: 0.2276002\ttotal: 5m 59s\tremaining: 1m 3s\n",
      "1700:\tlearn: 0.2275871\ttotal: 5m 59s\tremaining: 1m 3s\n",
      "1701:\tlearn: 0.2275762\ttotal: 5m 59s\tremaining: 1m 3s\n",
      "1702:\tlearn: 0.2275640\ttotal: 6m\tremaining: 1m 2s\n",
      "1703:\tlearn: 0.2275519\ttotal: 6m\tremaining: 1m 2s\n",
      "1704:\tlearn: 0.2275382\ttotal: 6m\tremaining: 1m 2s\n",
      "1705:\tlearn: 0.2275293\ttotal: 6m\tremaining: 1m 2s\n",
      "1706:\tlearn: 0.2275164\ttotal: 6m\tremaining: 1m 1s\n",
      "1707:\tlearn: 0.2275071\ttotal: 6m 1s\tremaining: 1m 1s\n",
      "1708:\tlearn: 0.2274982\ttotal: 6m 1s\tremaining: 1m 1s\n",
      "1709:\tlearn: 0.2274829\ttotal: 6m 1s\tremaining: 1m 1s\n",
      "1710:\tlearn: 0.2274709\ttotal: 6m 1s\tremaining: 1m 1s\n",
      "1711:\tlearn: 0.2274609\ttotal: 6m 1s\tremaining: 1m\n",
      "1712:\tlearn: 0.2274512\ttotal: 6m 2s\tremaining: 1m\n",
      "1713:\tlearn: 0.2274391\ttotal: 6m 2s\tremaining: 1m\n",
      "1714:\tlearn: 0.2274259\ttotal: 6m 2s\tremaining: 1m\n",
      "1715:\tlearn: 0.2274096\ttotal: 6m 2s\tremaining: 1m\n",
      "1716:\tlearn: 0.2273999\ttotal: 6m 3s\tremaining: 59.8s\n",
      "1717:\tlearn: 0.2273886\ttotal: 6m 3s\tremaining: 59.6s\n",
      "1718:\tlearn: 0.2273762\ttotal: 6m 3s\tremaining: 59.4s\n",
      "1719:\tlearn: 0.2273654\ttotal: 6m 3s\tremaining: 59.2s\n",
      "1720:\tlearn: 0.2273554\ttotal: 6m 3s\tremaining: 59s\n",
      "1721:\tlearn: 0.2273469\ttotal: 6m 4s\tremaining: 58.8s\n",
      "1722:\tlearn: 0.2273396\ttotal: 6m 4s\tremaining: 58.6s\n",
      "1723:\tlearn: 0.2273288\ttotal: 6m 4s\tremaining: 58.4s\n",
      "1724:\tlearn: 0.2273155\ttotal: 6m 4s\tremaining: 58.1s\n",
      "1725:\tlearn: 0.2273066\ttotal: 6m 4s\tremaining: 57.9s\n",
      "1726:\tlearn: 0.2272909\ttotal: 6m 5s\tremaining: 57.7s\n",
      "1727:\tlearn: 0.2272800\ttotal: 6m 5s\tremaining: 57.5s\n",
      "1728:\tlearn: 0.2272712\ttotal: 6m 5s\tremaining: 57.3s\n",
      "1729:\tlearn: 0.2272623\ttotal: 6m 5s\tremaining: 57.1s\n",
      "1730:\tlearn: 0.2272524\ttotal: 6m 6s\tremaining: 56.9s\n",
      "1731:\tlearn: 0.2272403\ttotal: 6m 6s\tremaining: 56.7s\n",
      "1732:\tlearn: 0.2272283\ttotal: 6m 6s\tremaining: 56.5s\n",
      "1733:\tlearn: 0.2272204\ttotal: 6m 6s\tremaining: 56.2s\n",
      "1734:\tlearn: 0.2272115\ttotal: 6m 6s\tremaining: 56s\n",
      "1735:\tlearn: 0.2272023\ttotal: 6m 7s\tremaining: 55.8s\n",
      "1736:\tlearn: 0.2271873\ttotal: 6m 7s\tremaining: 55.6s\n",
      "1737:\tlearn: 0.2271759\ttotal: 6m 7s\tremaining: 55.4s\n",
      "1738:\tlearn: 0.2271657\ttotal: 6m 7s\tremaining: 55.2s\n",
      "1739:\tlearn: 0.2271579\ttotal: 6m 7s\tremaining: 55s\n",
      "1740:\tlearn: 0.2271451\ttotal: 6m 8s\tremaining: 54.8s\n",
      "1741:\tlearn: 0.2271324\ttotal: 6m 8s\tremaining: 54.6s\n",
      "1742:\tlearn: 0.2271238\ttotal: 6m 8s\tremaining: 54.3s\n",
      "1743:\tlearn: 0.2271124\ttotal: 6m 8s\tremaining: 54.1s\n",
      "1744:\tlearn: 0.2271037\ttotal: 6m 9s\tremaining: 53.9s\n",
      "1745:\tlearn: 0.2270939\ttotal: 6m 9s\tremaining: 53.7s\n",
      "1746:\tlearn: 0.2270788\ttotal: 6m 9s\tremaining: 53.5s\n",
      "1747:\tlearn: 0.2270642\ttotal: 6m 9s\tremaining: 53.3s\n",
      "1748:\tlearn: 0.2270560\ttotal: 6m 9s\tremaining: 53.1s\n",
      "1749:\tlearn: 0.2270410\ttotal: 6m 10s\tremaining: 52.9s\n",
      "1750:\tlearn: 0.2270243\ttotal: 6m 10s\tremaining: 52.7s\n",
      "1751:\tlearn: 0.2270101\ttotal: 6m 10s\tremaining: 52.4s\n",
      "1752:\tlearn: 0.2269992\ttotal: 6m 10s\tremaining: 52.2s\n",
      "1753:\tlearn: 0.2269899\ttotal: 6m 10s\tremaining: 52s\n",
      "1754:\tlearn: 0.2269807\ttotal: 6m 11s\tremaining: 51.8s\n",
      "1755:\tlearn: 0.2269711\ttotal: 6m 11s\tremaining: 51.6s\n",
      "1756:\tlearn: 0.2269613\ttotal: 6m 11s\tremaining: 51.4s\n",
      "1757:\tlearn: 0.2269480\ttotal: 6m 11s\tremaining: 51.2s\n",
      "1758:\tlearn: 0.2269329\ttotal: 6m 11s\tremaining: 51s\n",
      "1759:\tlearn: 0.2269219\ttotal: 6m 12s\tremaining: 50.8s\n",
      "1760:\tlearn: 0.2269136\ttotal: 6m 12s\tremaining: 50.5s\n",
      "1761:\tlearn: 0.2269060\ttotal: 6m 12s\tremaining: 50.3s\n",
      "1762:\tlearn: 0.2268963\ttotal: 6m 12s\tremaining: 50.1s\n",
      "1763:\tlearn: 0.2268899\ttotal: 6m 13s\tremaining: 49.9s\n",
      "1764:\tlearn: 0.2268793\ttotal: 6m 13s\tremaining: 49.7s\n",
      "1765:\tlearn: 0.2268714\ttotal: 6m 13s\tremaining: 49.5s\n",
      "1766:\tlearn: 0.2268590\ttotal: 6m 13s\tremaining: 49.3s\n",
      "1767:\tlearn: 0.2268460\ttotal: 6m 13s\tremaining: 49.1s\n",
      "1768:\tlearn: 0.2268358\ttotal: 6m 14s\tremaining: 48.8s\n",
      "1769:\tlearn: 0.2268261\ttotal: 6m 14s\tremaining: 48.6s\n",
      "1770:\tlearn: 0.2268166\ttotal: 6m 14s\tremaining: 48.4s\n",
      "1771:\tlearn: 0.2268031\ttotal: 6m 14s\tremaining: 48.2s\n",
      "1772:\tlearn: 0.2267923\ttotal: 6m 14s\tremaining: 48s\n",
      "1773:\tlearn: 0.2267818\ttotal: 6m 15s\tremaining: 47.8s\n",
      "1774:\tlearn: 0.2267707\ttotal: 6m 15s\tremaining: 47.6s\n",
      "1775:\tlearn: 0.2267610\ttotal: 6m 15s\tremaining: 47.4s\n",
      "1776:\tlearn: 0.2267531\ttotal: 6m 15s\tremaining: 47.2s\n",
      "1777:\tlearn: 0.2267466\ttotal: 6m 16s\tremaining: 46.9s\n",
      "1778:\tlearn: 0.2267324\ttotal: 6m 16s\tremaining: 46.7s\n",
      "1779:\tlearn: 0.2267227\ttotal: 6m 16s\tremaining: 46.5s\n",
      "1780:\tlearn: 0.2267159\ttotal: 6m 16s\tremaining: 46.3s\n",
      "1781:\tlearn: 0.2267061\ttotal: 6m 16s\tremaining: 46.1s\n",
      "1782:\tlearn: 0.2266928\ttotal: 6m 17s\tremaining: 45.9s\n",
      "1783:\tlearn: 0.2266787\ttotal: 6m 17s\tremaining: 45.7s\n",
      "1784:\tlearn: 0.2266685\ttotal: 6m 17s\tremaining: 45.5s\n",
      "1785:\tlearn: 0.2266592\ttotal: 6m 17s\tremaining: 45.3s\n",
      "1786:\tlearn: 0.2266473\ttotal: 6m 17s\tremaining: 45s\n",
      "1787:\tlearn: 0.2266336\ttotal: 6m 18s\tremaining: 44.8s\n",
      "1788:\tlearn: 0.2266194\ttotal: 6m 18s\tremaining: 44.6s\n",
      "1789:\tlearn: 0.2266108\ttotal: 6m 18s\tremaining: 44.4s\n",
      "1790:\tlearn: 0.2265974\ttotal: 6m 18s\tremaining: 44.2s\n",
      "1791:\tlearn: 0.2265869\ttotal: 6m 18s\tremaining: 44s\n",
      "1792:\tlearn: 0.2265770\ttotal: 6m 19s\tremaining: 43.8s\n",
      "1793:\tlearn: 0.2265656\ttotal: 6m 19s\tremaining: 43.6s\n",
      "1794:\tlearn: 0.2265565\ttotal: 6m 19s\tremaining: 43.3s\n",
      "1795:\tlearn: 0.2265445\ttotal: 6m 19s\tremaining: 43.1s\n",
      "1796:\tlearn: 0.2265366\ttotal: 6m 19s\tremaining: 42.9s\n",
      "1797:\tlearn: 0.2265247\ttotal: 6m 20s\tremaining: 42.7s\n",
      "1798:\tlearn: 0.2265168\ttotal: 6m 20s\tremaining: 42.5s\n",
      "1799:\tlearn: 0.2265062\ttotal: 6m 20s\tremaining: 42.3s\n",
      "1800:\tlearn: 0.2264975\ttotal: 6m 20s\tremaining: 42.1s\n",
      "1801:\tlearn: 0.2264836\ttotal: 6m 21s\tremaining: 41.9s\n",
      "1802:\tlearn: 0.2264707\ttotal: 6m 21s\tremaining: 41.7s\n",
      "1803:\tlearn: 0.2264588\ttotal: 6m 21s\tremaining: 41.4s\n",
      "1804:\tlearn: 0.2264506\ttotal: 6m 21s\tremaining: 41.2s\n",
      "1805:\tlearn: 0.2264442\ttotal: 6m 21s\tremaining: 41s\n",
      "1806:\tlearn: 0.2264332\ttotal: 6m 22s\tremaining: 40.8s\n",
      "1807:\tlearn: 0.2264247\ttotal: 6m 22s\tremaining: 40.6s\n",
      "1808:\tlearn: 0.2264089\ttotal: 6m 22s\tremaining: 40.4s\n",
      "1809:\tlearn: 0.2263982\ttotal: 6m 22s\tremaining: 40.2s\n",
      "1810:\tlearn: 0.2263880\ttotal: 6m 22s\tremaining: 40s\n",
      "1811:\tlearn: 0.2263797\ttotal: 6m 23s\tremaining: 39.8s\n",
      "1812:\tlearn: 0.2263685\ttotal: 6m 23s\tremaining: 39.5s\n",
      "1813:\tlearn: 0.2263563\ttotal: 6m 23s\tremaining: 39.3s\n",
      "1814:\tlearn: 0.2263454\ttotal: 6m 23s\tremaining: 39.1s\n",
      "1815:\tlearn: 0.2263339\ttotal: 6m 24s\tremaining: 38.9s\n",
      "1816:\tlearn: 0.2263260\ttotal: 6m 24s\tremaining: 38.7s\n",
      "1817:\tlearn: 0.2263180\ttotal: 6m 24s\tremaining: 38.5s\n",
      "1818:\tlearn: 0.2263103\ttotal: 6m 24s\tremaining: 38.3s\n",
      "1819:\tlearn: 0.2262991\ttotal: 6m 24s\tremaining: 38.1s\n",
      "1820:\tlearn: 0.2262889\ttotal: 6m 25s\tremaining: 37.9s\n",
      "1821:\tlearn: 0.2262783\ttotal: 6m 25s\tremaining: 37.6s\n",
      "1822:\tlearn: 0.2262709\ttotal: 6m 25s\tremaining: 37.4s\n",
      "1823:\tlearn: 0.2262595\ttotal: 6m 25s\tremaining: 37.2s\n",
      "1824:\tlearn: 0.2262516\ttotal: 6m 25s\tremaining: 37s\n",
      "1825:\tlearn: 0.2262398\ttotal: 6m 26s\tremaining: 36.8s\n",
      "1826:\tlearn: 0.2262283\ttotal: 6m 26s\tremaining: 36.6s\n",
      "1827:\tlearn: 0.2262182\ttotal: 6m 26s\tremaining: 36.4s\n",
      "1828:\tlearn: 0.2262100\ttotal: 6m 26s\tremaining: 36.2s\n",
      "1829:\tlearn: 0.2261973\ttotal: 6m 26s\tremaining: 35.9s\n",
      "1830:\tlearn: 0.2261863\ttotal: 6m 27s\tremaining: 35.7s\n",
      "1831:\tlearn: 0.2261765\ttotal: 6m 27s\tremaining: 35.5s\n",
      "1832:\tlearn: 0.2261705\ttotal: 6m 27s\tremaining: 35.3s\n",
      "1833:\tlearn: 0.2261563\ttotal: 6m 27s\tremaining: 35.1s\n",
      "1834:\tlearn: 0.2261476\ttotal: 6m 28s\tremaining: 34.9s\n",
      "1835:\tlearn: 0.2261362\ttotal: 6m 28s\tremaining: 34.7s\n",
      "1836:\tlearn: 0.2261243\ttotal: 6m 28s\tremaining: 34.5s\n",
      "1837:\tlearn: 0.2261139\ttotal: 6m 28s\tremaining: 34.3s\n",
      "1838:\tlearn: 0.2261060\ttotal: 6m 28s\tremaining: 34s\n",
      "1839:\tlearn: 0.2260927\ttotal: 6m 29s\tremaining: 33.8s\n",
      "1840:\tlearn: 0.2260810\ttotal: 6m 29s\tremaining: 33.6s\n",
      "1841:\tlearn: 0.2260740\ttotal: 6m 29s\tremaining: 33.4s\n",
      "1842:\tlearn: 0.2260607\ttotal: 6m 29s\tremaining: 33.2s\n",
      "1843:\tlearn: 0.2260503\ttotal: 6m 29s\tremaining: 33s\n",
      "1844:\tlearn: 0.2260409\ttotal: 6m 30s\tremaining: 32.8s\n",
      "1845:\tlearn: 0.2260335\ttotal: 6m 30s\tremaining: 32.6s\n",
      "1846:\tlearn: 0.2260255\ttotal: 6m 30s\tremaining: 32.4s\n",
      "1847:\tlearn: 0.2260185\ttotal: 6m 30s\tremaining: 32.1s\n",
      "1848:\tlearn: 0.2260071\ttotal: 6m 30s\tremaining: 31.9s\n",
      "1849:\tlearn: 0.2259964\ttotal: 6m 31s\tremaining: 31.7s\n",
      "1850:\tlearn: 0.2259862\ttotal: 6m 31s\tremaining: 31.5s\n",
      "1851:\tlearn: 0.2259771\ttotal: 6m 31s\tremaining: 31.3s\n",
      "1852:\tlearn: 0.2259635\ttotal: 6m 31s\tremaining: 31.1s\n",
      "1853:\tlearn: 0.2259534\ttotal: 6m 32s\tremaining: 30.9s\n",
      "1854:\tlearn: 0.2259429\ttotal: 6m 32s\tremaining: 30.7s\n",
      "1855:\tlearn: 0.2259336\ttotal: 6m 32s\tremaining: 30.4s\n",
      "1856:\tlearn: 0.2259237\ttotal: 6m 32s\tremaining: 30.2s\n",
      "1857:\tlearn: 0.2259121\ttotal: 6m 32s\tremaining: 30s\n",
      "1858:\tlearn: 0.2259052\ttotal: 6m 33s\tremaining: 29.8s\n",
      "1859:\tlearn: 0.2258983\ttotal: 6m 33s\tremaining: 29.6s\n",
      "1860:\tlearn: 0.2258853\ttotal: 6m 33s\tremaining: 29.4s\n",
      "1861:\tlearn: 0.2258749\ttotal: 6m 33s\tremaining: 29.2s\n",
      "1862:\tlearn: 0.2258642\ttotal: 6m 33s\tremaining: 29s\n",
      "1863:\tlearn: 0.2258539\ttotal: 6m 34s\tremaining: 28.8s\n",
      "1864:\tlearn: 0.2258408\ttotal: 6m 34s\tremaining: 28.5s\n",
      "1865:\tlearn: 0.2258325\ttotal: 6m 34s\tremaining: 28.3s\n",
      "1866:\tlearn: 0.2258208\ttotal: 6m 34s\tremaining: 28.1s\n",
      "1867:\tlearn: 0.2258105\ttotal: 6m 35s\tremaining: 27.9s\n",
      "1868:\tlearn: 0.2257974\ttotal: 6m 35s\tremaining: 27.7s\n",
      "1869:\tlearn: 0.2257849\ttotal: 6m 35s\tremaining: 27.5s\n",
      "1870:\tlearn: 0.2257752\ttotal: 6m 35s\tremaining: 27.3s\n",
      "1871:\tlearn: 0.2257652\ttotal: 6m 35s\tremaining: 27.1s\n",
      "1872:\tlearn: 0.2257578\ttotal: 6m 36s\tremaining: 26.9s\n",
      "1873:\tlearn: 0.2257481\ttotal: 6m 36s\tremaining: 26.6s\n",
      "1874:\tlearn: 0.2257387\ttotal: 6m 36s\tremaining: 26.4s\n",
      "1875:\tlearn: 0.2257315\ttotal: 6m 36s\tremaining: 26.2s\n",
      "1876:\tlearn: 0.2257178\ttotal: 6m 36s\tremaining: 26s\n",
      "1877:\tlearn: 0.2257054\ttotal: 6m 37s\tremaining: 25.8s\n",
      "1878:\tlearn: 0.2257020\ttotal: 6m 37s\tremaining: 25.6s\n",
      "1879:\tlearn: 0.2256931\ttotal: 6m 37s\tremaining: 25.4s\n",
      "1880:\tlearn: 0.2256818\ttotal: 6m 37s\tremaining: 25.2s\n",
      "1881:\tlearn: 0.2256718\ttotal: 6m 37s\tremaining: 25s\n",
      "1882:\tlearn: 0.2256602\ttotal: 6m 38s\tremaining: 24.7s\n",
      "1883:\tlearn: 0.2256474\ttotal: 6m 38s\tremaining: 24.5s\n",
      "1884:\tlearn: 0.2256390\ttotal: 6m 38s\tremaining: 24.3s\n",
      "1885:\tlearn: 0.2256311\ttotal: 6m 38s\tremaining: 24.1s\n",
      "1886:\tlearn: 0.2256221\ttotal: 6m 39s\tremaining: 23.9s\n",
      "1887:\tlearn: 0.2256098\ttotal: 6m 39s\tremaining: 23.7s\n",
      "1888:\tlearn: 0.2256017\ttotal: 6m 39s\tremaining: 23.5s\n",
      "1889:\tlearn: 0.2255884\ttotal: 6m 39s\tremaining: 23.3s\n",
      "1890:\tlearn: 0.2255797\ttotal: 6m 39s\tremaining: 23.1s\n",
      "1891:\tlearn: 0.2255653\ttotal: 6m 40s\tremaining: 22.8s\n",
      "1892:\tlearn: 0.2255581\ttotal: 6m 40s\tremaining: 22.6s\n",
      "1893:\tlearn: 0.2255539\ttotal: 6m 40s\tremaining: 22.4s\n",
      "1894:\tlearn: 0.2255396\ttotal: 6m 40s\tremaining: 22.2s\n",
      "1895:\tlearn: 0.2255315\ttotal: 6m 40s\tremaining: 22s\n",
      "1896:\tlearn: 0.2255203\ttotal: 6m 41s\tremaining: 21.8s\n",
      "1897:\tlearn: 0.2255148\ttotal: 6m 41s\tremaining: 21.6s\n",
      "1898:\tlearn: 0.2255012\ttotal: 6m 41s\tremaining: 21.4s\n",
      "1899:\tlearn: 0.2254899\ttotal: 6m 41s\tremaining: 21.1s\n",
      "1900:\tlearn: 0.2254765\ttotal: 6m 42s\tremaining: 20.9s\n",
      "1901:\tlearn: 0.2254635\ttotal: 6m 42s\tremaining: 20.7s\n",
      "1902:\tlearn: 0.2254510\ttotal: 6m 42s\tremaining: 20.5s\n",
      "1903:\tlearn: 0.2254396\ttotal: 6m 42s\tremaining: 20.3s\n",
      "1904:\tlearn: 0.2254283\ttotal: 6m 42s\tremaining: 20.1s\n",
      "1905:\tlearn: 0.2254150\ttotal: 6m 43s\tremaining: 19.9s\n",
      "1906:\tlearn: 0.2254072\ttotal: 6m 43s\tremaining: 19.7s\n",
      "1907:\tlearn: 0.2254006\ttotal: 6m 43s\tremaining: 19.5s\n",
      "1908:\tlearn: 0.2253887\ttotal: 6m 43s\tremaining: 19.2s\n",
      "1909:\tlearn: 0.2253802\ttotal: 6m 43s\tremaining: 19s\n",
      "1910:\tlearn: 0.2253718\ttotal: 6m 44s\tremaining: 18.8s\n",
      "1911:\tlearn: 0.2253536\ttotal: 6m 44s\tremaining: 18.6s\n",
      "1912:\tlearn: 0.2253446\ttotal: 6m 44s\tremaining: 18.4s\n",
      "1913:\tlearn: 0.2253303\ttotal: 6m 44s\tremaining: 18.2s\n",
      "1914:\tlearn: 0.2253235\ttotal: 6m 44s\tremaining: 18s\n",
      "1915:\tlearn: 0.2253122\ttotal: 6m 45s\tremaining: 17.8s\n",
      "1916:\tlearn: 0.2253042\ttotal: 6m 45s\tremaining: 17.6s\n",
      "1917:\tlearn: 0.2252949\ttotal: 6m 45s\tremaining: 17.3s\n",
      "1918:\tlearn: 0.2252838\ttotal: 6m 45s\tremaining: 17.1s\n",
      "1919:\tlearn: 0.2252718\ttotal: 6m 45s\tremaining: 16.9s\n",
      "1920:\tlearn: 0.2252639\ttotal: 6m 46s\tremaining: 16.7s\n",
      "1921:\tlearn: 0.2252572\ttotal: 6m 46s\tremaining: 16.5s\n",
      "1922:\tlearn: 0.2252510\ttotal: 6m 46s\tremaining: 16.3s\n",
      "1923:\tlearn: 0.2252380\ttotal: 6m 46s\tremaining: 16.1s\n",
      "1924:\tlearn: 0.2252284\ttotal: 6m 47s\tremaining: 15.9s\n",
      "1925:\tlearn: 0.2252203\ttotal: 6m 47s\tremaining: 15.6s\n",
      "1926:\tlearn: 0.2252096\ttotal: 6m 47s\tremaining: 15.4s\n",
      "1927:\tlearn: 0.2251968\ttotal: 6m 47s\tremaining: 15.2s\n",
      "1928:\tlearn: 0.2251807\ttotal: 6m 47s\tremaining: 15s\n",
      "1929:\tlearn: 0.2251693\ttotal: 6m 48s\tremaining: 14.8s\n",
      "1930:\tlearn: 0.2251576\ttotal: 6m 48s\tremaining: 14.6s\n",
      "1931:\tlearn: 0.2251424\ttotal: 6m 48s\tremaining: 14.4s\n",
      "1932:\tlearn: 0.2251335\ttotal: 6m 48s\tremaining: 14.2s\n",
      "1933:\tlearn: 0.2251214\ttotal: 6m 48s\tremaining: 14s\n",
      "1934:\tlearn: 0.2251127\ttotal: 6m 49s\tremaining: 13.7s\n",
      "1935:\tlearn: 0.2251055\ttotal: 6m 49s\tremaining: 13.5s\n",
      "1936:\tlearn: 0.2250948\ttotal: 6m 49s\tremaining: 13.3s\n",
      "1937:\tlearn: 0.2250824\ttotal: 6m 49s\tremaining: 13.1s\n",
      "1938:\tlearn: 0.2250754\ttotal: 6m 50s\tremaining: 12.9s\n",
      "1939:\tlearn: 0.2250659\ttotal: 6m 50s\tremaining: 12.7s\n",
      "1940:\tlearn: 0.2250572\ttotal: 6m 50s\tremaining: 12.5s\n",
      "1941:\tlearn: 0.2250476\ttotal: 6m 50s\tremaining: 12.3s\n",
      "1942:\tlearn: 0.2250330\ttotal: 6m 50s\tremaining: 12.1s\n",
      "1943:\tlearn: 0.2250242\ttotal: 6m 51s\tremaining: 11.8s\n",
      "1944:\tlearn: 0.2250167\ttotal: 6m 51s\tremaining: 11.6s\n",
      "1945:\tlearn: 0.2250086\ttotal: 6m 51s\tremaining: 11.4s\n",
      "1946:\tlearn: 0.2249946\ttotal: 6m 51s\tremaining: 11.2s\n",
      "1947:\tlearn: 0.2249833\ttotal: 6m 51s\tremaining: 11s\n",
      "1948:\tlearn: 0.2249750\ttotal: 6m 52s\tremaining: 10.8s\n",
      "1949:\tlearn: 0.2249668\ttotal: 6m 52s\tremaining: 10.6s\n",
      "1950:\tlearn: 0.2249567\ttotal: 6m 52s\tremaining: 10.4s\n",
      "1951:\tlearn: 0.2249495\ttotal: 6m 52s\tremaining: 10.2s\n",
      "1952:\tlearn: 0.2249394\ttotal: 6m 53s\tremaining: 9.94s\n",
      "1953:\tlearn: 0.2249308\ttotal: 6m 53s\tremaining: 9.73s\n",
      "1954:\tlearn: 0.2249191\ttotal: 6m 53s\tremaining: 9.52s\n",
      "1955:\tlearn: 0.2249118\ttotal: 6m 53s\tremaining: 9.3s\n",
      "1956:\tlearn: 0.2249018\ttotal: 6m 53s\tremaining: 9.09s\n",
      "1957:\tlearn: 0.2248920\ttotal: 6m 54s\tremaining: 8.88s\n",
      "1958:\tlearn: 0.2248834\ttotal: 6m 54s\tremaining: 8.67s\n",
      "1959:\tlearn: 0.2248704\ttotal: 6m 54s\tremaining: 8.46s\n",
      "1960:\tlearn: 0.2248601\ttotal: 6m 54s\tremaining: 8.25s\n",
      "1961:\tlearn: 0.2248527\ttotal: 6m 54s\tremaining: 8.04s\n",
      "1962:\tlearn: 0.2248424\ttotal: 6m 55s\tremaining: 7.83s\n",
      "1963:\tlearn: 0.2248336\ttotal: 6m 55s\tremaining: 7.61s\n",
      "1964:\tlearn: 0.2248283\ttotal: 6m 55s\tremaining: 7.4s\n",
      "1965:\tlearn: 0.2248194\ttotal: 6m 55s\tremaining: 7.19s\n",
      "1966:\tlearn: 0.2248078\ttotal: 6m 56s\tremaining: 6.98s\n",
      "1967:\tlearn: 0.2247975\ttotal: 6m 56s\tremaining: 6.77s\n",
      "1968:\tlearn: 0.2247834\ttotal: 6m 56s\tremaining: 6.56s\n",
      "1969:\tlearn: 0.2247767\ttotal: 6m 56s\tremaining: 6.34s\n",
      "1970:\tlearn: 0.2247643\ttotal: 6m 56s\tremaining: 6.13s\n",
      "1971:\tlearn: 0.2247533\ttotal: 6m 57s\tremaining: 5.92s\n",
      "1972:\tlearn: 0.2247422\ttotal: 6m 57s\tremaining: 5.71s\n",
      "1973:\tlearn: 0.2247358\ttotal: 6m 57s\tremaining: 5.5s\n",
      "1974:\tlearn: 0.2247192\ttotal: 6m 57s\tremaining: 5.29s\n",
      "1975:\tlearn: 0.2247089\ttotal: 6m 57s\tremaining: 5.08s\n",
      "1976:\tlearn: 0.2246944\ttotal: 6m 58s\tremaining: 4.86s\n",
      "1977:\tlearn: 0.2246802\ttotal: 6m 58s\tremaining: 4.65s\n",
      "1978:\tlearn: 0.2246725\ttotal: 6m 58s\tremaining: 4.44s\n",
      "1979:\tlearn: 0.2246610\ttotal: 6m 58s\tremaining: 4.23s\n",
      "1980:\tlearn: 0.2246520\ttotal: 6m 59s\tremaining: 4.02s\n",
      "1981:\tlearn: 0.2246431\ttotal: 6m 59s\tremaining: 3.81s\n",
      "1982:\tlearn: 0.2246355\ttotal: 6m 59s\tremaining: 3.6s\n",
      "1983:\tlearn: 0.2246260\ttotal: 6m 59s\tremaining: 3.38s\n",
      "1984:\tlearn: 0.2246089\ttotal: 6m 59s\tremaining: 3.17s\n",
      "1985:\tlearn: 0.2245981\ttotal: 7m\tremaining: 2.96s\n",
      "1986:\tlearn: 0.2245886\ttotal: 7m\tremaining: 2.75s\n",
      "1987:\tlearn: 0.2245803\ttotal: 7m\tremaining: 2.54s\n",
      "1988:\tlearn: 0.2245754\ttotal: 7m\tremaining: 2.33s\n",
      "1989:\tlearn: 0.2245625\ttotal: 7m\tremaining: 2.12s\n",
      "1990:\tlearn: 0.2245517\ttotal: 7m 1s\tremaining: 1.9s\n",
      "1991:\tlearn: 0.2245444\ttotal: 7m 1s\tremaining: 1.69s\n",
      "1992:\tlearn: 0.2245378\ttotal: 7m 1s\tremaining: 1.48s\n",
      "1993:\tlearn: 0.2245281\ttotal: 7m 1s\tremaining: 1.27s\n",
      "1994:\tlearn: 0.2245191\ttotal: 7m 2s\tremaining: 1.06s\n",
      "1995:\tlearn: 0.2245110\ttotal: 7m 2s\tremaining: 846ms\n",
      "1996:\tlearn: 0.2245028\ttotal: 7m 2s\tremaining: 635ms\n",
      "1997:\tlearn: 0.2244927\ttotal: 7m 2s\tremaining: 423ms\n",
      "1998:\tlearn: 0.2244819\ttotal: 7m 2s\tremaining: 212ms\n",
      "1999:\tlearn: 0.2244729\ttotal: 7m 3s\tremaining: 0us\n",
      "Training Completed \n",
      "Total Time : 0.0 hours 447.672501 seconds\n",
      "The Test log loss is: 0.24300858358388394\n",
      "Feature Importance\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "c_model, predict_y = train_catboost(X_train, X_val, y_train, y_val, plot_importance = True, save = True, file_name = 'catboost_v3.pkl' )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 299
    },
    "id": "8eEo2iC-ymfT",
    "outputId": "9f33b819-6d12-40dc-940e-66d71b5249ec"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy : 0.9107859067078719\n",
      "Precision : 0.6545585861160677\n"
     ]
    }
   ],
   "source": [
    "#plot confusion matrix\n",
    "predicted_y =np.array(predict_y>0.5,dtype=int)\n",
    "cm = confusion_matrix(y_val, predicted_y)\n",
    "sns.heatmap(cm, cmap='Blues', annot = True,  fmt=\"g\", annot_kws={\"size\": 16})\n",
    "plt.show()\n",
    "print(\"Accuracy :\",accuracy_score(y_val, predicted_y))\n",
    "print(\"Precision :\",precision_score(y_val, predicted_y))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "7stMgZswgHFM"
   },
   "source": [
    "#### Create a submission file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "uWflTL2irtZS",
    "outputId": "d9e9a0d9-c665-46aa-df14-76a2d2dedf40"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generated submission_sample_v6.1_1_catboost.csv successfully\n",
      "Generated submission_sample_v6.1_2_catboost.csv successfully\n",
      "Generated submission_sample_v6.1_3_catboost.csv successfully\n"
     ]
    }
   ],
   "source": [
    "#generate submission files based on global threshold\n",
    "globl_threshold_products(model = pickle.load(open(\"catboost_v3.pkl\", \"rb\")), model_type = 'catboost')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "5DmzHS4u3pjd",
    "outputId": "42a9f7eb-4aad-4e33-fc4e-0ccb7dd14352"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 75000/75000 [42:43<00:00, 29.26it/s]\n"
     ]
    }
   ],
   "source": [
    "#get products for order ID's and create a submission file based on local threshold\n",
    "test_products = getscores_on_testdata(model = pickle.load(open(\"catboost_v3.pkl\", \"rb\")), model_type = 'catboost')\n",
    "test_products.to_csv(\"submission_catboost_v6.csv\", index= False) #0.39498 0.39484"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "xYwXL5dstpx1"
   },
   "source": [
    "### Performance"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "3H2P6fWHeOZ4"
   },
   "source": [
    "We will compare all models for their mean f1-score on test set , on both global threshold and local threshold"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "2Cl1XbD-yiTo",
    "outputId": "57e887e3-8ffc-4890-b827-aacba244f90f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+--------------------------+---------+-----------------------------+---------------+--------------+\n",
      "|          Model           | logloss |          Threshold          | Private Score | Public Score |\n",
      "+--------------------------+---------+-----------------------------+---------------+--------------+\n",
      "|   Logistic Regression    |  0.2550 |   Global_Threshold : 0.18   |    0.35316    |   0.35485    |\n",
      "|   Logistic Regression    |  0.2550 | Local_Threshold : faron_opt |    0.37405    |   0.37687    |\n",
      "| Decision Tree Classifier |  0.2509 |   Global_Threshold : 0.20   |    0.28438    |   0.28660    |\n",
      "| Decision Tree Classifier |  0.2509 | Local_Threshold : faron_opt |    0.29532    |   0.29822    |\n",
      "| Random Forest Classifier |  0.2518 |   Global_Threshold : 0.18   |    0.35359    |   0.35456    |\n",
      "| Random Forest Classifier |  0.2518 | Local_Threshold : faron_opt |    0.37825    |   0.38067    |\n",
      "|        MLP model         |  0.2434 |   Global_Threshold : 0.18   |    0.35598    |   0.35630    |\n",
      "|        MLP model         |  0.2434 | Local_Threshold : faron_opt |    0.38327    |   0.38449    |\n",
      "|           XGB            |  0.2434 |   Global_Threshold : 0.18   |    0.36706    |   0.36815    |\n",
      "|           XGB            |  0.2434 | Local_Threshold : faron_opt |    0.39155    |   0.39303    |\n",
      "|         Catboost         |  0.2430 |   Global_Threshold : 0.18   |    0.36837    |   0.36850    |\n",
      "|         Catboost         |  0.2430 | Local_Threshold : faron_opt |    0.39498    |   0.39484    |\n",
      "+--------------------------+---------+-----------------------------+---------------+--------------+\n"
     ]
    }
   ],
   "source": [
    "from prettytable import PrettyTable\n",
    "# http://zetcode.com/python/prettytable/\n",
    "\n",
    "x = PrettyTable()\n",
    "x.field_names = [\"Model\", \"logloss\", \"Threshold \" , \"Private Score\" ,\"Public Score\"]\n",
    "x.add_row(['Logistic Regression','0.2550','Global_Threshold : 0.18','0.35316','0.35485'])\n",
    "x.add_row(['Logistic Regression','0.2550','Local_Threshold : faron_opt','0.37405','0.37687'])\n",
    "x.add_row(['Decision Tree Classifier','0.2509','Global_Threshold : 0.20','0.28438', '0.28660'])\n",
    "x.add_row(['Decision Tree Classifier','0.2509','Local_Threshold : faron_opt','0.29532', '0.29822'])\n",
    "x.add_row(['Random Forest Classifier','0.2518','Global_Threshold : 0.18','0.35359','0.35456'])\n",
    "x.add_row(['Random Forest Classifier','0.2518','Local_Threshold : faron_opt','0.37825','0.38067'])\n",
    "x.add_row(['MLP model','0.2434','Global_Threshold : 0.18','0.35598','0.35630'])\n",
    "x.add_row(['MLP model','0.2434','Local_Threshold : faron_opt','0.38327','0.38449'])\n",
    "x.add_row(['XGB','0.2434','Global_Threshold : 0.18','0.36706','0.36815'])\n",
    "x.add_row(['XGB','0.2434','Local_Threshold : faron_opt','0.39155','0.39303'])\n",
    "x.add_row(['Catboost','0.2430','Global_Threshold : 0.18','0.36837','0.36850'])\n",
    "x.add_row(['Catboost','0.2430','Local_Threshold : faron_opt','0.39498','0.39484'])\n",
    "\n",
    "print(x)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "pmQNUyxGluW9"
   },
   "source": [
    "For Local threshold :\n",
    "\n",
    "Catboost > XGB > MLP > Random forest > Logistic Regression > Decision Tree\n",
    "\n",
    "For Global threshold:\n",
    "\n",
    "Catboost > XGB > MLP > Random forest > Logistic Regression > Decision Tree"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "iWdbCQ85gHOL"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "machine_shape": "hm",
   "name": "Models_v3.ipynb",
   "provenance": []
  },
  "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": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}