xgb-av-janatahack-healthcare-analytics-tuning-blending.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import xgboost as xgb\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import os\n",
    "from matplotlib import pyplot as plt\n",
    "from sklearn import preprocessing\n",
    "from sklearn.metrics import mean_squared_log_error, mean_squared_error\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import roc_auc_score\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.metrics import log_loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_train = pd.read_csv('data/Train/Train.csv', header=0)\n",
    "df_camps = pd.read_csv('data/Train/Health_Camp_Detail.csv', header=0)\n",
    "df_patients = pd.read_csv('data/Train/Patient_Profile.csv', header=0)\n",
    "\n",
    "df_first_camp = pd.read_csv('data/Train/First_Health_Camp_Attended.csv', header=0)\n",
    "df_second_camp = pd.read_csv('data/Train/Second_Health_Camp_Attended.csv', header=0)\n",
    "df_third_camp = pd.read_csv('data/Train/Third_Health_Camp_Attended.csv', header=0)\n",
    "\n",
    "df_test = pd.read_csv('data/test.csv', header=0)\n",
    "submission = pd.read_csv('data/sample_submission.csv', header=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((75278, 8),\n",
       " (65, 6),\n",
       " (37633, 11),\n",
       " (6218, 5),\n",
       " (7819, 3),\n",
       " (6515, 4),\n",
       " (35249, 8),\n",
       " (35249, 3))"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train.shape, df_camps.shape, df_patients.shape, df_first_camp.shape, df_second_camp.shape, df_third_camp.shape, df_test.shape, submission.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Patient_ID</th>\n",
       "      <th>Health_Camp_ID</th>\n",
       "      <th>Registration_Date</th>\n",
       "      <th>Var1</th>\n",
       "      <th>Var2</th>\n",
       "      <th>Var3</th>\n",
       "      <th>Var4</th>\n",
       "      <th>Var5</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>489652</td>\n",
       "      <td>6578</td>\n",
       "      <td>10-Sep-05</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>507246</td>\n",
       "      <td>6578</td>\n",
       "      <td>18-Aug-05</td>\n",
       "      <td>45</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>523729</td>\n",
       "      <td>6534</td>\n",
       "      <td>29-Apr-06</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>524931</td>\n",
       "      <td>6535</td>\n",
       "      <td>07-Feb-04</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>521364</td>\n",
       "      <td>6529</td>\n",
       "      <td>28-Feb-06</td>\n",
       "      <td>15</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Patient_ID  Health_Camp_ID Registration_Date  Var1  Var2  Var3  Var4  Var5\n",
       "0      489652            6578         10-Sep-05     4     0     0     0     2\n",
       "1      507246            6578         18-Aug-05    45     5     0     0     7\n",
       "2      523729            6534         29-Apr-06     0     0     0     0     0\n",
       "3      524931            6535         07-Feb-04     0     0     0     0     0\n",
       "4      521364            6529         28-Feb-06    15     1     0     0     7"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Patient_ID</th>\n",
       "      <th>Health_Camp_ID</th>\n",
       "      <th>Registration_Date</th>\n",
       "      <th>Var1</th>\n",
       "      <th>Var2</th>\n",
       "      <th>Var3</th>\n",
       "      <th>Var4</th>\n",
       "      <th>Var5</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>505701</td>\n",
       "      <td>6548</td>\n",
       "      <td>21-May-06</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>500633</td>\n",
       "      <td>6584</td>\n",
       "      <td>02-Jun-06</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>506945</td>\n",
       "      <td>6582</td>\n",
       "      <td>10-Aug-06</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>497447</td>\n",
       "      <td>6551</td>\n",
       "      <td>27-Aug-06</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>496446</td>\n",
       "      <td>6533</td>\n",
       "      <td>19-Sep-06</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Patient_ID  Health_Camp_ID Registration_Date  Var1  Var2  Var3  Var4  Var5\n",
       "0      505701            6548         21-May-06     1     0     0     0     2\n",
       "1      500633            6584         02-Jun-06     0     0     0     0     0\n",
       "2      506945            6582         10-Aug-06     0     0     0     0     0\n",
       "3      497447            6551         27-Aug-06     0     0     0     0     0\n",
       "4      496446            6533         19-Sep-06     0     0     0     0     0"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_test.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Health_Camp_ID</th>\n",
       "      <th>Camp_Start_Date</th>\n",
       "      <th>Camp_End_Date</th>\n",
       "      <th>Category1</th>\n",
       "      <th>Category2</th>\n",
       "      <th>Category3</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>6560</td>\n",
       "      <td>16-Aug-03</td>\n",
       "      <td>20-Aug-03</td>\n",
       "      <td>First</td>\n",
       "      <td>B</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>6530</td>\n",
       "      <td>16-Aug-03</td>\n",
       "      <td>28-Oct-03</td>\n",
       "      <td>First</td>\n",
       "      <td>C</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>6544</td>\n",
       "      <td>03-Nov-03</td>\n",
       "      <td>15-Nov-03</td>\n",
       "      <td>First</td>\n",
       "      <td>F</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>6585</td>\n",
       "      <td>22-Nov-03</td>\n",
       "      <td>05-Dec-03</td>\n",
       "      <td>First</td>\n",
       "      <td>E</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>6561</td>\n",
       "      <td>30-Nov-03</td>\n",
       "      <td>18-Dec-03</td>\n",
       "      <td>First</td>\n",
       "      <td>E</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Health_Camp_ID Camp_Start_Date Camp_End_Date Category1 Category2  Category3\n",
       "0            6560       16-Aug-03     20-Aug-03     First         B          2\n",
       "1            6530       16-Aug-03     28-Oct-03     First         C          2\n",
       "2            6544       03-Nov-03     15-Nov-03     First         F          1\n",
       "3            6585       22-Nov-03     05-Dec-03     First         E          2\n",
       "4            6561       30-Nov-03     18-Dec-03     First         E          1"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_camps.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Patient_ID</th>\n",
       "      <th>Online_Follower</th>\n",
       "      <th>LinkedIn_Shared</th>\n",
       "      <th>Twitter_Shared</th>\n",
       "      <th>Facebook_Shared</th>\n",
       "      <th>Income</th>\n",
       "      <th>Education_Score</th>\n",
       "      <th>Age</th>\n",
       "      <th>First_Interaction</th>\n",
       "      <th>City_Type</th>\n",
       "      <th>Employer_Category</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>516956</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>90</td>\n",
       "      <td>39</td>\n",
       "      <td>18-Jun-03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Software Industry</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>507733</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>None</td>\n",
       "      <td>40</td>\n",
       "      <td>20-Jul-03</td>\n",
       "      <td>H</td>\n",
       "      <td>Software Industry</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>508307</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>87</td>\n",
       "      <td>46</td>\n",
       "      <td>02-Nov-02</td>\n",
       "      <td>D</td>\n",
       "      <td>BFSI</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>512612</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>75</td>\n",
       "      <td>47</td>\n",
       "      <td>02-Nov-02</td>\n",
       "      <td>D</td>\n",
       "      <td>Education</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>521075</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>None</td>\n",
       "      <td>80</td>\n",
       "      <td>24-Nov-02</td>\n",
       "      <td>H</td>\n",
       "      <td>Others</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Patient_ID  Online_Follower  LinkedIn_Shared  Twitter_Shared  \\\n",
       "0      516956                0                0               0   \n",
       "1      507733                0                0               0   \n",
       "2      508307                0                0               0   \n",
       "3      512612                0                0               0   \n",
       "4      521075                0                0               0   \n",
       "\n",
       "   Facebook_Shared Income Education_Score Age First_Interaction City_Type  \\\n",
       "0                0      1              90  39         18-Jun-03       NaN   \n",
       "1                0      1            None  40         20-Jul-03         H   \n",
       "2                0      3              87  46         02-Nov-02         D   \n",
       "3                0      1              75  47         02-Nov-02         D   \n",
       "4                0      3            None  80         24-Nov-02         H   \n",
       "\n",
       "   Employer_Category  \n",
       "0  Software Industry  \n",
       "1  Software Industry  \n",
       "2               BFSI  \n",
       "3          Education  \n",
       "4             Others  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_patients.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Patient_ID</th>\n",
       "      <th>Health_Camp_ID</th>\n",
       "      <th>Outcome</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>505701</td>\n",
       "      <td>6548</td>\n",
       "      <td>0.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>500633</td>\n",
       "      <td>6584</td>\n",
       "      <td>0.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>506945</td>\n",
       "      <td>6582</td>\n",
       "      <td>0.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>497447</td>\n",
       "      <td>6551</td>\n",
       "      <td>0.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>496446</td>\n",
       "      <td>6533</td>\n",
       "      <td>0.5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Patient_ID  Health_Camp_ID  Outcome\n",
       "0      505701            6548      0.5\n",
       "1      500633            6584      0.5\n",
       "2      506945            6582      0.5\n",
       "3      497447            6551      0.5\n",
       "4      496446            6533      0.5"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "submission.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Patient_ID</th>\n",
       "      <th>Health_Camp_ID</th>\n",
       "      <th>Donation</th>\n",
       "      <th>Health_Score</th>\n",
       "      <th>Unnamed: 4</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>506181</td>\n",
       "      <td>6560</td>\n",
       "      <td>40</td>\n",
       "      <td>0.439024</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>494977</td>\n",
       "      <td>6560</td>\n",
       "      <td>20</td>\n",
       "      <td>0.097561</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>518680</td>\n",
       "      <td>6560</td>\n",
       "      <td>10</td>\n",
       "      <td>0.048780</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>509916</td>\n",
       "      <td>6560</td>\n",
       "      <td>30</td>\n",
       "      <td>0.634146</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>488006</td>\n",
       "      <td>6560</td>\n",
       "      <td>20</td>\n",
       "      <td>0.024390</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Patient_ID  Health_Camp_ID  Donation  Health_Score  Unnamed: 4\n",
       "0      506181            6560        40      0.439024         NaN\n",
       "1      494977            6560        20      0.097561         NaN\n",
       "2      518680            6560        10      0.048780         NaN\n",
       "3      509916            6560        30      0.634146         NaN\n",
       "4      488006            6560        20      0.024390         NaN"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_first_camp.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Patient_ID</th>\n",
       "      <th>Health_Camp_ID</th>\n",
       "      <th>Health Score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>526631</td>\n",
       "      <td>6536</td>\n",
       "      <td>0.875136</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>509122</td>\n",
       "      <td>6536</td>\n",
       "      <td>0.755700</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>498864</td>\n",
       "      <td>6536</td>\n",
       "      <td>0.673181</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>515398</td>\n",
       "      <td>6536</td>\n",
       "      <td>0.722041</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>504624</td>\n",
       "      <td>6536</td>\n",
       "      <td>0.464712</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Patient_ID  Health_Camp_ID  Health Score\n",
       "0      526631            6536      0.875136\n",
       "1      509122            6536      0.755700\n",
       "2      498864            6536      0.673181\n",
       "3      515398            6536      0.722041\n",
       "4      504624            6536      0.464712"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_second_camp.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Patient_ID</th>\n",
       "      <th>Health_Camp_ID</th>\n",
       "      <th>Number_of_stall_visited</th>\n",
       "      <th>Last_Stall_Visited_Number</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>517875</td>\n",
       "      <td>6527</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>504692</td>\n",
       "      <td>6578</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>504692</td>\n",
       "      <td>6527</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>493167</td>\n",
       "      <td>6527</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>510954</td>\n",
       "      <td>6528</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Patient_ID  Health_Camp_ID  Number_of_stall_visited  \\\n",
       "0      517875            6527                        3   \n",
       "1      504692            6578                        1   \n",
       "2      504692            6527                        3   \n",
       "3      493167            6527                        4   \n",
       "4      510954            6528                        2   \n",
       "\n",
       "   Last_Stall_Visited_Number  \n",
       "0                          1  \n",
       "1                          1  \n",
       "2                          1  \n",
       "3                          4  \n",
       "4                          2  "
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_third_camp.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(5212, 1)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.merge(df_first_camp[['Patient_ID']], df_second_camp[['Patient_ID']], on='Patient_ID').shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((6218, 5), (7819, 3), (6515, 4))"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_first_camp.shape, df_second_camp.shape, df_third_camp.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.0016666670000000001, 1.0, 0.058992805999999995, 1.0, 0, 7)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "min(df_first_camp.Health_Score), max(df_first_camp.Health_Score), min(df_second_camp['Health Score']), max(df_second_camp['Health Score']), min(df_third_camp.Number_of_stall_visited), max(df_third_camp.Number_of_stall_visited)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(6218, 6) (7819, 4) (6515, 5) (20552, 3)\n"
     ]
    }
   ],
   "source": [
    "df_first_camp['label'] = df_first_camp['Health_Score'].apply(lambda x: 1 if x > 0 else 0)\n",
    "df_second_camp['label'] = df_second_camp['Health Score'].apply(lambda x: 1 if x > 0 else 0)\n",
    "df_third_camp['label'] = df_third_camp['Number_of_stall_visited'].apply(lambda x: 1 if x > 0 else 0)\n",
    "\n",
    "df_label = pd.concat((df_first_camp[['Patient_ID', 'Health_Camp_ID', 'label']], df_second_camp[['Patient_ID', 'Health_Camp_ID', 'label']], df_third_camp[['Patient_ID', 'Health_Camp_ID', 'label']]))\n",
    "\n",
    "print(df_first_camp.shape, df_second_camp.shape, df_third_camp.shape, df_label.shape)\n",
    "assert df_first_camp.shape[0] + df_second_camp.shape[0] + df_third_camp.shape[0] == df_label.shape[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_train = pd.merge(df_train, df_camps, on=['Health_Camp_ID'], how='left')\n",
    "df_train = pd.merge(df_train, df_patients, on=['Patient_ID'], how='left')\n",
    "\n",
    "# df_train = pd.merge(df_train, df_first_camp, on=['Patient_ID', 'Health_Camp_ID'], how='left')\n",
    "# df_train = pd.merge(df_train, df_second_camp, on=['Patient_ID', 'Health_Camp_ID'], how='left')\n",
    "# df_train = pd.merge(df_train, df_third_camp, on=['Patient_ID', 'Health_Camp_ID'], how='left')\n",
    "df_train = pd.merge(df_train, df_label, on=['Patient_ID', 'Health_Camp_ID'], how='left')\n",
    "df_train['label'] = df_train['label'].fillna(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_test = pd.merge(df_test, df_camps, on=['Health_Camp_ID'], how='left')\n",
    "df_test = pd.merge(df_test, df_patients, on=['Patient_ID'], how='left')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((75278, 24), (35249, 23))"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train.shape, df_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Patient_ID</th>\n",
       "      <th>Health_Camp_ID</th>\n",
       "      <th>Registration_Date</th>\n",
       "      <th>Var1</th>\n",
       "      <th>Var2</th>\n",
       "      <th>Var3</th>\n",
       "      <th>Var4</th>\n",
       "      <th>Var5</th>\n",
       "      <th>Camp_Start_Date</th>\n",
       "      <th>Camp_End_Date</th>\n",
       "      <th>...</th>\n",
       "      <th>LinkedIn_Shared</th>\n",
       "      <th>Twitter_Shared</th>\n",
       "      <th>Facebook_Shared</th>\n",
       "      <th>Income</th>\n",
       "      <th>Education_Score</th>\n",
       "      <th>Age</th>\n",
       "      <th>First_Interaction</th>\n",
       "      <th>City_Type</th>\n",
       "      <th>Employer_Category</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>489652</td>\n",
       "      <td>6578</td>\n",
       "      <td>10-Sep-05</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>16-Aug-05</td>\n",
       "      <td>14-Oct-05</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>06-Dec-04</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>507246</td>\n",
       "      <td>6578</td>\n",
       "      <td>18-Aug-05</td>\n",
       "      <td>45</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>7</td>\n",
       "      <td>16-Aug-05</td>\n",
       "      <td>14-Oct-05</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>75</td>\n",
       "      <td>40</td>\n",
       "      <td>08-Sep-04</td>\n",
       "      <td>C</td>\n",
       "      <td>Others</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>523729</td>\n",
       "      <td>6534</td>\n",
       "      <td>29-Apr-06</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>17-Oct-05</td>\n",
       "      <td>07-Nov-07</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>22-Jun-04</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>524931</td>\n",
       "      <td>6535</td>\n",
       "      <td>07-Feb-04</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>01-Feb-04</td>\n",
       "      <td>18-Feb-04</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>07-Feb-04</td>\n",
       "      <td>I</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>521364</td>\n",
       "      <td>6529</td>\n",
       "      <td>28-Feb-06</td>\n",
       "      <td>15</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>7</td>\n",
       "      <td>30-Mar-06</td>\n",
       "      <td>03-Apr-06</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>70</td>\n",
       "      <td>40</td>\n",
       "      <td>04-Jul-03</td>\n",
       "      <td>I</td>\n",
       "      <td>Technology</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 24 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Patient_ID  Health_Camp_ID Registration_Date  Var1  Var2  Var3  Var4  Var5  \\\n",
       "0      489652            6578         10-Sep-05     4     0     0     0     2   \n",
       "1      507246            6578         18-Aug-05    45     5     0     0     7   \n",
       "2      523729            6534         29-Apr-06     0     0     0     0     0   \n",
       "3      524931            6535         07-Feb-04     0     0     0     0     0   \n",
       "4      521364            6529         28-Feb-06    15     1     0     0     7   \n",
       "\n",
       "  Camp_Start_Date Camp_End_Date  ... LinkedIn_Shared Twitter_Shared  \\\n",
       "0       16-Aug-05     14-Oct-05  ...               0              0   \n",
       "1       16-Aug-05     14-Oct-05  ...               0              0   \n",
       "2       17-Oct-05     07-Nov-07  ...               0              0   \n",
       "3       01-Feb-04     18-Feb-04  ...               0              0   \n",
       "4       30-Mar-06     03-Apr-06  ...               0              0   \n",
       "\n",
       "   Facebook_Shared  Income  Education_Score   Age  First_Interaction  \\\n",
       "0                0    None             None  None          06-Dec-04   \n",
       "1                0       1               75    40          08-Sep-04   \n",
       "2                0    None             None  None          22-Jun-04   \n",
       "3                0    None             None  None          07-Feb-04   \n",
       "4                1       1               70    40          04-Jul-03   \n",
       "\n",
       "  City_Type Employer_Category label  \n",
       "0       NaN               NaN   1.0  \n",
       "1         C            Others   0.0  \n",
       "2       NaN               NaN   1.0  \n",
       "3         I               NaN   0.0  \n",
       "4         I        Technology   1.0  \n",
       "\n",
       "[5 rows x 24 columns]"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Patient_ID</th>\n",
       "      <th>Health_Camp_ID</th>\n",
       "      <th>Registration_Date</th>\n",
       "      <th>Var1</th>\n",
       "      <th>Var2</th>\n",
       "      <th>Var3</th>\n",
       "      <th>Var4</th>\n",
       "      <th>Var5</th>\n",
       "      <th>Camp_Start_Date</th>\n",
       "      <th>Camp_End_Date</th>\n",
       "      <th>...</th>\n",
       "      <th>Online_Follower</th>\n",
       "      <th>LinkedIn_Shared</th>\n",
       "      <th>Twitter_Shared</th>\n",
       "      <th>Facebook_Shared</th>\n",
       "      <th>Income</th>\n",
       "      <th>Education_Score</th>\n",
       "      <th>Age</th>\n",
       "      <th>First_Interaction</th>\n",
       "      <th>City_Type</th>\n",
       "      <th>Employer_Category</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>505701</td>\n",
       "      <td>6548</td>\n",
       "      <td>21-May-06</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>13-Jun-06</td>\n",
       "      <td>18-Aug-06</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>None</td>\n",
       "      <td>44</td>\n",
       "      <td>05-Feb-03</td>\n",
       "      <td>E</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>500633</td>\n",
       "      <td>6584</td>\n",
       "      <td>02-Jun-06</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>04-Aug-06</td>\n",
       "      <td>09-Aug-06</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>67</td>\n",
       "      <td>41</td>\n",
       "      <td>11-Dec-04</td>\n",
       "      <td>D</td>\n",
       "      <td>Consulting</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>506945</td>\n",
       "      <td>6582</td>\n",
       "      <td>10-Aug-06</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>06-Apr-06</td>\n",
       "      <td>07-Nov-07</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>19-Apr-06</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>497447</td>\n",
       "      <td>6551</td>\n",
       "      <td>27-Aug-06</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>13-Nov-06</td>\n",
       "      <td>18-Nov-06</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>None</td>\n",
       "      <td>47</td>\n",
       "      <td>25-Aug-06</td>\n",
       "      <td>B</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>496446</td>\n",
       "      <td>6533</td>\n",
       "      <td>19-Sep-06</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>20-Sep-06</td>\n",
       "      <td>23-Sep-06</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>19-Sep-06</td>\n",
       "      <td>B</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 23 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Patient_ID  Health_Camp_ID Registration_Date  Var1  Var2  Var3  Var4  Var5  \\\n",
       "0      505701            6548         21-May-06     1     0     0     0     2   \n",
       "1      500633            6584         02-Jun-06     0     0     0     0     0   \n",
       "2      506945            6582         10-Aug-06     0     0     0     0     0   \n",
       "3      497447            6551         27-Aug-06     0     0     0     0     0   \n",
       "4      496446            6533         19-Sep-06     0     0     0     0     0   \n",
       "\n",
       "  Camp_Start_Date Camp_End_Date  ... Online_Follower LinkedIn_Shared  \\\n",
       "0       13-Jun-06     18-Aug-06  ...               0               0   \n",
       "1       04-Aug-06     09-Aug-06  ...               0               1   \n",
       "2       06-Apr-06     07-Nov-07  ...               0               0   \n",
       "3       13-Nov-06     18-Nov-06  ...               0               0   \n",
       "4       20-Sep-06     23-Sep-06  ...               0               0   \n",
       "\n",
       "   Twitter_Shared  Facebook_Shared  Income  Education_Score   Age  \\\n",
       "0               0                0       0             None    44   \n",
       "1               0                0       1               67    41   \n",
       "2               0                0    None             None  None   \n",
       "3               0                0       0             None    47   \n",
       "4               0                0    None             None  None   \n",
       "\n",
       "  First_Interaction City_Type Employer_Category  \n",
       "0         05-Feb-03         E               NaN  \n",
       "1         11-Dec-04         D        Consulting  \n",
       "2         19-Apr-06       NaN               NaN  \n",
       "3         25-Aug-06         B               NaN  \n",
       "4         19-Sep-06         B               NaN  \n",
       "\n",
       "[5 rows x 23 columns]"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_test.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0    54744\n",
       "1.0    20534\n",
       "Name: label, dtype: int64"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train['label'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(Patient_ID             int64\n",
       " Health_Camp_ID         int64\n",
       " Registration_Date     object\n",
       " Var1                   int64\n",
       " Var2                   int64\n",
       " Var3                   int64\n",
       " Var4                   int64\n",
       " Var5                   int64\n",
       " Camp_Start_Date       object\n",
       " Camp_End_Date         object\n",
       " Category1             object\n",
       " Category2             object\n",
       " Category3              int64\n",
       " Online_Follower        int64\n",
       " LinkedIn_Shared        int64\n",
       " Twitter_Shared         int64\n",
       " Facebook_Shared        int64\n",
       " Income                object\n",
       " Education_Score       object\n",
       " Age                   object\n",
       " First_Interaction     object\n",
       " City_Type             object\n",
       " Employer_Category     object\n",
       " label                float64\n",
       " dtype: object,\n",
       " Patient_ID            int64\n",
       " Health_Camp_ID        int64\n",
       " Registration_Date    object\n",
       " Var1                  int64\n",
       " Var2                  int64\n",
       " Var3                  int64\n",
       " Var4                  int64\n",
       " Var5                  int64\n",
       " Camp_Start_Date      object\n",
       " Camp_End_Date        object\n",
       " Category1            object\n",
       " Category2            object\n",
       " Category3             int64\n",
       " Online_Follower       int64\n",
       " LinkedIn_Shared       int64\n",
       " Twitter_Shared        int64\n",
       " Facebook_Shared       int64\n",
       " Income               object\n",
       " Education_Score      object\n",
       " Age                  object\n",
       " First_Interaction    object\n",
       " City_Type            object\n",
       " Employer_Category    object\n",
       " dtype: object)"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train.dtypes, df_test.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Patient_ID               0\n",
       "Health_Camp_ID           0\n",
       "Registration_Date      334\n",
       "Var1                     0\n",
       "Var2                     0\n",
       "Var3                     0\n",
       "Var4                     0\n",
       "Var5                     0\n",
       "Camp_Start_Date          0\n",
       "Camp_End_Date            0\n",
       "Category1                0\n",
       "Category2                0\n",
       "Category3                0\n",
       "Online_Follower          0\n",
       "LinkedIn_Shared          0\n",
       "Twitter_Shared           0\n",
       "Facebook_Shared          0\n",
       "Income                   0\n",
       "Education_Score          0\n",
       "Age                      0\n",
       "First_Interaction        0\n",
       "City_Type            33208\n",
       "Employer_Category    60075\n",
       "label                    0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train.apply(lambda x: sum(pd.isna(x)), axis=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(None    53546\n",
       " 0        8393\n",
       " 1        5346\n",
       " 2        4108\n",
       " 3        2155\n",
       " 4        1034\n",
       " 5         440\n",
       " 6         256\n",
       " Name: Income, dtype: int64,\n",
       " None           65345\n",
       " 82               468\n",
       " 79               423\n",
       " 75               413\n",
       " 86               390\n",
       "                ...  \n",
       " 78.43333333        1\n",
       " 72.875             1\n",
       " 67.325             1\n",
       " 78.05              1\n",
       " 80.43333333        1\n",
       " Name: Education_Score, Length: 228, dtype: int64,\n",
       " None    51612\n",
       " 41       1843\n",
       " 40       1760\n",
       " 42       1676\n",
       " 43       1628\n",
       " 39       1477\n",
       " 44       1460\n",
       " 45       1094\n",
       " 37       1033\n",
       " 46       1025\n",
       " 38        996\n",
       " 47        885\n",
       " 72        792\n",
       " 48        736\n",
       " 73        710\n",
       " 49        644\n",
       " 71        593\n",
       " 74        540\n",
       " 36        477\n",
       " 50        446\n",
       " 52        418\n",
       " 51        416\n",
       " 70        394\n",
       " 35        360\n",
       " 53        315\n",
       " 54        222\n",
       " 34        191\n",
       " 56        189\n",
       " 75        167\n",
       " 55        147\n",
       " 76        131\n",
       " 69        118\n",
       " 58        111\n",
       " 59         88\n",
       " 67         87\n",
       " 33         68\n",
       " 57         68\n",
       " 77         48\n",
       " 60         46\n",
       " 68         45\n",
       " 63         37\n",
       " 64         33\n",
       " 65         30\n",
       " 66         29\n",
       " 61         27\n",
       " 32         24\n",
       " 78         17\n",
       " 80         10\n",
       " 31          8\n",
       " 62          7\n",
       " Name: Age, dtype: int64,\n",
       " 08-Sep-04    1497\n",
       " 01-May-05    1236\n",
       " 08-Feb-03     987\n",
       " 21-May-05     934\n",
       " 25-Oct-04     906\n",
       "              ... \n",
       " 16-Dec-02       1\n",
       " 13-Mar-03       1\n",
       " 21-May-04       1\n",
       " 15-Apr-04       1\n",
       " 18-Mar-03       1\n",
       " Name: First_Interaction, Length: 1474, dtype: int64,\n",
       " B    8273\n",
       " H    6139\n",
       " D    5451\n",
       " G    4360\n",
       " C    4259\n",
       " E    4025\n",
       " A    3441\n",
       " I    3312\n",
       " F    2810\n",
       " Name: City_Type, dtype: int64,\n",
       " Technology           3533\n",
       " Software Industry    2358\n",
       " Others               2239\n",
       " Consulting           2128\n",
       " Education            1125\n",
       " BFSI                  845\n",
       " Retail                535\n",
       " Manufacturing         510\n",
       " Health                446\n",
       " Transport             364\n",
       " Real Estate           361\n",
       " Food                  302\n",
       " Telecom               272\n",
       " Broadcasting          185\n",
       " Name: Employer_Category, dtype: int64)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train.Income.value_counts(), df_train.Education_Score.value_counts(), df_train.Age.value_counts(), df_train.First_Interaction.value_counts(), df_train.City_Type.value_counts(), df_train.Employer_Category.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# np.median(df_train.loc[pd.notna(df_train['Income']), 'Income'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_train['Income'] = df_train['Income'].apply(lambda x: np.nan if x == 'None' else x).astype('float')\n",
    "df_test['Income'] = df_test['Income'].apply(lambda x: np.nan if x == 'None' else x).astype('float')\n",
    "\n",
    "df_train['Education_Score'] = df_train['Education_Score'].apply(lambda x: np.nan if x == 'None' else x).astype('float')\n",
    "df_test['Education_Score'] = df_test['Education_Score'].apply(lambda x: np.nan if x == 'None' else x).astype('float')\n",
    "\n",
    "df_train['Age'] = df_train['Age'].apply(lambda x: np.nan if x == 'None' else x).astype('float')\n",
    "df_test['Age'] = df_test['Age'].apply(lambda x: np.nan if x == 'None' else x).astype('float')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "le_Category1, le_Category2, le_City_Type, le_Employer_Category = preprocessing.LabelEncoder(), preprocessing.LabelEncoder(), preprocessing.LabelEncoder(), preprocessing.LabelEncoder()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LabelEncoder()"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "le_Category1.fit(df_train.loc[pd.notna(df_train['Category1']), 'Category1'])\n",
    "le_Category2.fit(df_train.loc[pd.notna(df_train['Category2']), 'Category2'])\n",
    "le_City_Type.fit(df_train.loc[pd.notna(df_train['City_Type']), 'City_Type'])\n",
    "le_Employer_Category.fit(df_train.loc[pd.notna(df_train['Employer_Category']), 'Employer_Category'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "missing_impute = -999"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_train['Category1'] = df_train['Category1'].apply(lambda x: missing_impute if pd.isna(x) else le_Category1.transform([x])[0])\n",
    "df_train['Category2'] = df_train['Category2'].apply(lambda x: missing_impute if pd.isna(x) else le_Category2.transform([x])[0])\n",
    "df_train['City_Type'] = df_train['City_Type'].apply(lambda x: missing_impute if pd.isna(x) else le_City_Type.transform([x])[0])\n",
    "df_train['Employer_Category'] = df_train['Employer_Category'].apply(lambda x: missing_impute if pd.isna(x) else le_Employer_Category.transform([x])[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_test['Category1'] = df_test['Category1'].apply(lambda x: missing_impute if pd.isna(x) else le_Category1.transform([x])[0])\n",
    "df_test['Category2'] = df_test['Category2'].apply(lambda x: missing_impute if pd.isna(x) else le_Category2.transform([x])[0])\n",
    "df_test['City_Type'] = df_test['City_Type'].apply(lambda x: missing_impute if pd.isna(x) else le_City_Type.transform([x])[0])\n",
    "df_test['Employer_Category'] = df_test['Employer_Category'].apply(lambda x: missing_impute if pd.isna(x) else le_Employer_Category.transform([x])[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "#dates\n",
    "# Registration_Date, Camp_Start_Date, Camp_End_Date, First_Interaction\n",
    "\n",
    "df_train['Registration_Date'] = pd.to_datetime(df_train['Registration_Date'], format='%d-%b-%y')\n",
    "df_train['Camp_Start_Date'] = pd.to_datetime(df_train['Camp_Start_Date'], format='%d-%b-%y')\n",
    "df_train['Camp_End_Date'] = pd.to_datetime(df_train['Camp_End_Date'], format='%d-%b-%y')\n",
    "df_train['First_Interaction'] = pd.to_datetime(df_train['First_Interaction'], format='%d-%b-%y')\n",
    "\n",
    "df_test['Registration_Date'] = pd.to_datetime(df_test['Registration_Date'], format='%d-%b-%y')\n",
    "df_test['Camp_Start_Date'] = pd.to_datetime(df_test['Camp_Start_Date'], format='%d-%b-%y')\n",
    "df_test['Camp_End_Date'] = pd.to_datetime(df_test['Camp_End_Date'], format='%d-%b-%y')\n",
    "df_test['First_Interaction'] = pd.to_datetime(df_test['First_Interaction'], format='%d-%b-%y')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_train['Registration_weekday'] = df_train['Registration_Date'].apply(lambda x: x.weekday())\n",
    "df_train['Registration_weekofyear'] = df_train['Registration_Date'].apply(lambda x: x.weekofyear)\n",
    "df_train['Registration_month'] = df_train['Registration_Date'].apply(lambda x: x.month)\n",
    "df_train['Registration_quarter'] = df_train['Registration_Date'].apply(lambda x: x.quarter)\n",
    "df_train['Registration_year'] = df_train['Registration_Date'].apply(lambda x: x.year)\n",
    "\n",
    "df_test['Registration_weekday'] = df_test['Registration_Date'].apply(lambda x: x.weekday())\n",
    "df_test['Registration_weekofyear'] = df_test['Registration_Date'].apply(lambda x: x.weekofyear)\n",
    "df_test['Registration_month'] = df_test['Registration_Date'].apply(lambda x: x.month)\n",
    "df_test['Registration_quarter'] = df_test['Registration_Date'].apply(lambda x: x.quarter)\n",
    "df_test['Registration_year'] = df_test['Registration_Date'].apply(lambda x: x.year)\n",
    "\n",
    "\n",
    "\n",
    "df_train['Camp_Start_Date_weekday'] = df_train['Camp_Start_Date'].apply(lambda x: x.weekday())\n",
    "df_train['Camp_Start_Date_weekofyear'] = df_train['Camp_Start_Date'].apply(lambda x: x.weekofyear)\n",
    "df_train['Camp_Start_Date_month'] = df_train['Camp_Start_Date'].apply(lambda x: x.month)\n",
    "df_train['Camp_Start_Date_quarter'] = df_train['Camp_Start_Date'].apply(lambda x: x.quarter)\n",
    "df_train['Camp_Start_Date_year'] = df_train['Camp_Start_Date'].apply(lambda x: x.year)\n",
    "\n",
    "df_test['Camp_Start_Date_weekday'] = df_test['Camp_Start_Date'].apply(lambda x: x.weekday())\n",
    "df_test['Camp_Start_Date_weekofyear'] = df_test['Camp_Start_Date'].apply(lambda x: x.weekofyear)\n",
    "df_test['Camp_Start_Date_month'] = df_test['Camp_Start_Date'].apply(lambda x: x.month)\n",
    "df_test['Camp_Start_Date_quarter'] = df_test['Camp_Start_Date'].apply(lambda x: x.quarter)\n",
    "df_test['Camp_Start_Date_year'] = df_test['Camp_Start_Date'].apply(lambda x: x.year)\n",
    "\n",
    "\n",
    "\n",
    "df_train['Camp_End_Date_weekday'] = df_train['Camp_End_Date'].apply(lambda x: x.weekday())\n",
    "df_train['Camp_End_Date_weekofyear'] = df_train['Camp_End_Date'].apply(lambda x: x.weekofyear)\n",
    "df_train['Camp_End_Date_month'] = df_train['Camp_End_Date'].apply(lambda x: x.month)\n",
    "df_train['Camp_End_Date_quarter'] = df_train['Camp_End_Date'].apply(lambda x: x.quarter)\n",
    "df_train['Camp_End_Date_year'] = df_train['Camp_End_Date'].apply(lambda x: x.year)\n",
    "\n",
    "df_test['Camp_End_Date_weekday'] = df_test['Camp_End_Date'].apply(lambda x: x.weekday())\n",
    "df_test['Camp_End_Date_weekofyear'] = df_test['Camp_End_Date'].apply(lambda x: x.weekofyear)\n",
    "df_test['Camp_End_Date_month'] = df_test['Camp_End_Date'].apply(lambda x: x.month)\n",
    "df_test['Camp_End_Date_quarter'] = df_test['Camp_End_Date'].apply(lambda x: x.quarter)\n",
    "df_test['Camp_End_Date_year'] = df_test['Camp_End_Date'].apply(lambda x: x.year)\n",
    "\n",
    "\n",
    "\n",
    "df_train['First_Interaction_weekday'] = df_train['First_Interaction'].apply(lambda x: x.weekday())\n",
    "df_train['First_Interaction_weekofyear'] = df_train['First_Interaction'].apply(lambda x: x.weekofyear)\n",
    "df_train['First_Interaction_month'] = df_train['First_Interaction'].apply(lambda x: x.month)\n",
    "df_train['First_Interaction_quarter'] = df_train['First_Interaction'].apply(lambda x: x.quarter)\n",
    "df_train['First_Interaction_year'] = df_train['First_Interaction'].apply(lambda x: x.year)\n",
    "\n",
    "df_test['First_Interaction_weekday'] = df_test['First_Interaction'].apply(lambda x: x.weekday())\n",
    "df_test['First_Interaction_weekofyear'] = df_test['First_Interaction'].apply(lambda x: x.weekofyear)\n",
    "df_test['First_Interaction_month'] = df_test['First_Interaction'].apply(lambda x: x.month)\n",
    "df_test['First_Interaction_quarter'] = df_test['First_Interaction'].apply(lambda x: x.quarter)\n",
    "df_test['First_Interaction_year'] = df_test['First_Interaction'].apply(lambda x: x.year)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_train['interation_to_registration'] = (df_train['Registration_Date'] - df_train['First_Interaction']).apply(lambda x: x.days)\n",
    "df_train['camp_duration'] = (df_train['Camp_End_Date'] - df_train['Camp_Start_Date']).apply(lambda x: x.days)\n",
    "df_train['camp_weeks'] = df_train['camp_duration']/7\n",
    "df_train['registration_to_camp_start'] = (df_train['Camp_Start_Date'] - df_train['Registration_Date']).apply(lambda x: x.days)\n",
    "\n",
    "\n",
    "df_test['interation_to_registration'] = (df_test['Registration_Date'] - df_test['First_Interaction']).apply(lambda x: x.days)\n",
    "df_test['camp_duration'] = (df_test['Camp_End_Date'] - df_test['Camp_Start_Date']).apply(lambda x: x.days)\n",
    "df_test['camp_weeks'] = df_test['camp_duration']/7\n",
    "df_test['registration_to_camp_start'] = (df_test['Camp_Start_Date'] - df_test['Registration_Date']).apply(lambda x: x.days)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(Index(['Patient_ID', 'Health_Camp_ID', 'Registration_Date', 'Var1', 'Var2',\n",
       "        'Var3', 'Var4', 'Var5', 'Camp_Start_Date', 'Camp_End_Date', 'Category1',\n",
       "        'Category2', 'Category3', 'Online_Follower', 'LinkedIn_Shared',\n",
       "        'Twitter_Shared', 'Facebook_Shared', 'Income', 'Education_Score', 'Age',\n",
       "        'First_Interaction', 'City_Type', 'Employer_Category', 'label',\n",
       "        'Registration_weekday', 'Registration_weekofyear', 'Registration_month',\n",
       "        'Registration_quarter', 'Registration_year', 'Camp_Start_Date_weekday',\n",
       "        'Camp_Start_Date_weekofyear', 'Camp_Start_Date_month',\n",
       "        'Camp_Start_Date_quarter', 'Camp_Start_Date_year',\n",
       "        'Camp_End_Date_weekday', 'Camp_End_Date_weekofyear',\n",
       "        'Camp_End_Date_month', 'Camp_End_Date_quarter', 'Camp_End_Date_year',\n",
       "        'First_Interaction_weekday', 'First_Interaction_weekofyear',\n",
       "        'First_Interaction_month', 'First_Interaction_quarter',\n",
       "        'First_Interaction_year', 'interation_to_registration', 'camp_duration',\n",
       "        'camp_weeks', 'registration_to_camp_start'],\n",
       "       dtype='object'),\n",
       " Index(['Patient_ID', 'Health_Camp_ID', 'Registration_Date', 'Var1', 'Var2',\n",
       "        'Var3', 'Var4', 'Var5', 'Camp_Start_Date', 'Camp_End_Date', 'Category1',\n",
       "        'Category2', 'Category3', 'Online_Follower', 'LinkedIn_Shared',\n",
       "        'Twitter_Shared', 'Facebook_Shared', 'Income', 'Education_Score', 'Age',\n",
       "        'First_Interaction', 'City_Type', 'Employer_Category',\n",
       "        'Registration_weekday', 'Registration_weekofyear', 'Registration_month',\n",
       "        'Registration_quarter', 'Registration_year', 'Camp_Start_Date_weekday',\n",
       "        'Camp_Start_Date_weekofyear', 'Camp_Start_Date_month',\n",
       "        'Camp_Start_Date_quarter', 'Camp_Start_Date_year',\n",
       "        'Camp_End_Date_weekday', 'Camp_End_Date_weekofyear',\n",
       "        'Camp_End_Date_month', 'Camp_End_Date_quarter', 'Camp_End_Date_year',\n",
       "        'First_Interaction_weekday', 'First_Interaction_weekofyear',\n",
       "        'First_Interaction_month', 'First_Interaction_quarter',\n",
       "        'First_Interaction_year', 'interation_to_registration', 'camp_duration',\n",
       "        'camp_weeks', 'registration_to_camp_start'],\n",
       "       dtype='object'))"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train.columns, df_test.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(75278, 49) (35249, 48) (110527, 49)\n"
     ]
    }
   ],
   "source": [
    "df_train['train_data'] = 1\n",
    "df_test['train_data'] = 0\n",
    "\n",
    "df_train_test = pd.concat((df_train, df_test))\n",
    "df_train_test = df_train_test.sort_values(['Patient_ID', 'Camp_Start_Date'])\n",
    "\n",
    "print(df_train.shape, df_test.shape, df_train_test.shape)\n",
    "df_train_test['attended_before'] = df_train_test.sort_values(['Patient_ID', 'Camp_Start_Date']).groupby(['Patient_ID'])['label'].apply(lambda x: x.shift().rolling(365, min_periods=1).sum()).fillna(0).values\n",
    "df_train_test['attended_before_flag'] = df_train_test['attended_before'].apply(lambda x: 1 if x > 0 else 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "# df_train_test[df_train_test['Patient_ID'] == 513633]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_train_test['registered_before'] = df_train_test.sort_values(['Patient_ID', 'Registration_Date']).groupby(['Patient_ID'])['label'].apply(lambda x: x.rolling(365, min_periods=1).count()).values\n",
    "\n",
    "df_train_test['interaction_before'] = df_train_test.sort_values(['Patient_ID', 'First_Interaction']).groupby(['Patient_ID'])['label'].apply(lambda x: x.rolling(365, min_periods=1).count()).values\n",
    "\n",
    "df_train_test['days_since_last_registration'] = df_train_test.sort_values(['Patient_ID', 'Registration_Date']).groupby('Patient_ID')['Registration_Date'].diff().apply(lambda x: x.days).fillna(missing_impute).values\n",
    "\n",
    "df_train_test = pd.merge(df_train_test, df_train_test.groupby('Health_Camp_ID')['Health_Camp_ID'].agg(['count']).reset_index().rename(columns={'count':'camp_popularity'}), on=['Health_Camp_ID'])\n",
    "\n",
    "# df_train_test['interaction_before'] = df_train_test.sort_values(['Patient_ID', 'First_Interaction']).groupby(['Patient_ID'])['label'].rolling(365, min_periods=1).count().values\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "# df_train_test[df_train_test['Patient_ID'] == 513633]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_train, df_test = df_train_test[df_train_test.train_data == 1], df_train_test[df_train_test.train_data == 0]\n",
    "\n",
    "del df_train_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.random.seed(123)\n",
    "patient_assignment = {}\n",
    "train_frac = 0.70\n",
    "\n",
    "def train_flag(x, train_frac):\n",
    "    res = 0\n",
    "    if x in patient_assignment:\n",
    "        res = patient_assignment[x]\n",
    "    else:\n",
    "        if np.random.random() < train_frac:\n",
    "            res = 1\n",
    "        patient_assignment[x] = res\n",
    "    return res\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "cat_cols = ['Category1', 'Category2', 'Category3', 'City_Type', 'Employer_Category']\n",
    "feature_cols = ['Var1', 'Var2', 'Var3', 'Var4', 'Var5', 'Category1', 'Category2', 'Category3', 'Online_Follower', 'LinkedIn_Shared', 'Twitter_Shared', 'Facebook_Shared', 'Income', 'Education_Score', 'Age', 'City_Type', 'Employer_Category', 'Registration_weekday', 'Registration_weekofyear', 'Registration_month',\n",
    "        'Registration_quarter', 'Registration_year', 'Camp_Start_Date_weekday',\n",
    "        'Camp_Start_Date_weekofyear', 'Camp_Start_Date_month',\n",
    "        'Camp_Start_Date_quarter', 'Camp_Start_Date_year',\n",
    "        'Camp_End_Date_weekday', 'Camp_End_Date_weekofyear',\n",
    "        'Camp_End_Date_month', 'Camp_End_Date_quarter', 'Camp_End_Date_year',\n",
    "        'First_Interaction_weekday', 'First_Interaction_weekofyear',\n",
    "        'First_Interaction_month', 'First_Interaction_quarter',\n",
    "        'First_Interaction_year', 'interation_to_registration', 'camp_duration',\n",
    "        'camp_weeks', 'registration_to_camp_start', 'attended_before', 'attended_before_flag', 'registered_before', 'interaction_before', 'days_since_last_registration', 'camp_popularity']\n",
    "label_col = 'label'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "#approach 1\n",
    "# df_train['train_flag'] = df_train['Patient_ID'].apply(train_flag , train_frac=train_frac)\n",
    "# df_train, df_eval = df_train[df_train.train_flag == 1], df_train[df_train.train_flag == 0]\n",
    "\n",
    "#approach 2\n",
    "df_train, df_eval = df_train[df_train['Camp_Start_Date'] <'2005-11-01'], df_train[df_train['Camp_Start_Date'] >'2005-10-30']\n",
    "\n",
    "#approach 3\n",
    "# stratify_var = [str(x1) + '_' + str(x2) for (x1, x2) in zip(df_train['Category1'], df_train['label'])]\n",
    "# df_train, df_eval = train_test_split(df_train, test_size=0.25, random_state=42, shuffle=True, stratify=stratify_var)\n",
    "\n",
    "#approach 4\n",
    "# df_train, df_eval = train_test_split(df_train, test_size=0.25, random_state=42, shuffle=True, stratify=df_train['Category1'])\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((68164, 55), (7114, 55))"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train.shape, df_eval.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "from bayes_opt import BayesianOptimization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "# https://stats.stackexchange.com/questions/317073/explanation-of-min-child-weight-in-xgboost-algorithm\n",
    "\n",
    "pbounds = {\n",
    "    'learning_rate': (0.005, 0.05),\n",
    "    'n_estimators': (100, 1500),\n",
    "    'max_depth': (3,19),\n",
    "    'colsample_bytree': (0.5, 0.8),  # Change for datasets with lots of features\n",
    "    'colsample_bylevel': (0.5, 0.8),\n",
    "    'colsample_bynode': (0.5, 0.8),\n",
    "    'min_child_weight': (9, 500),\n",
    "    'reg_alpha': (0, 5),\n",
    "    'reg_lambda': (0, 5),\n",
    "    'random_state': (0, 10000)\n",
    "}\n",
    "\n",
    "def gbm_hyper_param(learning_rate, n_estimators, max_depth, min_child_weight, reg_alpha, reg_lambda, colsample_bytree, colsample_bylevel, colsample_bynode, random_state):\n",
    "    max_depth = int(max_depth)\n",
    "    n_estimators = int(n_estimators)\n",
    "    subsample=0.7\n",
    "    random_state=int(random_state)\n",
    "#     colsample_bytree=0.7\n",
    "#     reg_alpha=2\n",
    "#     reg_lambda=2\n",
    "    clf = xgb.XGBClassifier(objective='binary:logistic', booster ='gbtree', max_depth=max_depth, learning_rate=learning_rate, n_estimators=n_estimators, min_child_weight=min_child_weight, subsample = subsample, colsample_bytree = colsample_bytree, colsample_bylevel=colsample_bylevel, colsample_bynode=colsample_bynode, random_state=random_state, reg_lambda=reg_lambda, reg_alpha=reg_alpha, missing=missing_impute)\n",
    "    \n",
    "    clf.fit(df_train[feature_cols], df_train[label_col], early_stopping_rounds=100, eval_set=[(df_train[feature_cols], df_train[label_col]), (df_eval[feature_cols], df_eval[label_col])], eval_metric='auc', verbose=False)\n",
    "    eval_score = roc_auc_score(df_eval[label_col], clf.predict_proba(df_eval[feature_cols])[:,1])\n",
    "#     eval_score = accuracy_score(df_eval[label_col], clf.predict(df_eval[feature_cols]))\n",
    "#     eval_score = -log_loss(df_eval[label_col], clf.predict_proba(df_eval[feature_cols])[:,1])\n",
    "    return eval_score\n",
    "\n",
    "\n",
    "optimizer = BayesianOptimization(\n",
    "    f=gbm_hyper_param,\n",
    "    pbounds=pbounds,\n",
    "    random_state=1,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "|   iter    |  target   | colsam... | colsam... | colsam... | learni... | max_depth | min_ch... | n_esti... | random... | reg_alpha | reg_la... |\n",
      "-------------------------------------------------------------------------------------------------------------------------------------------------\n",
      "| \u001b[0m 1       \u001b[0m | \u001b[0m 0.703   \u001b[0m | \u001b[0m 0.6251  \u001b[0m | \u001b[0m 0.7161  \u001b[0m | \u001b[0m 0.5     \u001b[0m | \u001b[0m 0.0186  \u001b[0m | \u001b[0m 5.348   \u001b[0m | \u001b[0m 54.34   \u001b[0m | \u001b[0m 360.8   \u001b[0m | \u001b[0m 3.456e+0\u001b[0m | \u001b[0m 1.984   \u001b[0m | \u001b[0m 2.694   \u001b[0m |\n",
      "| \u001b[0m 2       \u001b[0m | \u001b[0m 0.7028  \u001b[0m | \u001b[0m 0.6258  \u001b[0m | \u001b[0m 0.7056  \u001b[0m | \u001b[0m 0.5613  \u001b[0m | \u001b[0m 0.04452 \u001b[0m | \u001b[0m 3.438   \u001b[0m | \u001b[0m 338.2   \u001b[0m | \u001b[0m 684.2   \u001b[0m | \u001b[0m 5.587e+0\u001b[0m | \u001b[0m 0.7019  \u001b[0m | \u001b[0m 0.9905  \u001b[0m |\n",
      "| \u001b[0m 3       \u001b[0m | \u001b[0m 0.6979  \u001b[0m | \u001b[0m 0.7402  \u001b[0m | \u001b[0m 0.7905  \u001b[0m | \u001b[0m 0.594   \u001b[0m | \u001b[0m 0.03615 \u001b[0m | \u001b[0m 17.02   \u001b[0m | \u001b[0m 448.3   \u001b[0m | \u001b[0m 219.1   \u001b[0m | \u001b[0m 390.5   \u001b[0m | \u001b[0m 0.8492  \u001b[0m | \u001b[0m 4.391   \u001b[0m |\n",
      "| \u001b[0m 4       \u001b[0m | \u001b[0m 0.6636  \u001b[0m | \u001b[0m 0.5295  \u001b[0m | \u001b[0m 0.6263  \u001b[0m | \u001b[0m 0.7874  \u001b[0m | \u001b[0m 0.02899 \u001b[0m | \u001b[0m 14.07   \u001b[0m | \u001b[0m 163.9   \u001b[0m | \u001b[0m 1.061e+0\u001b[0m | \u001b[0m 8.346e+0\u001b[0m | \u001b[0m 0.09144 \u001b[0m | \u001b[0m 3.751   \u001b[0m |\n",
      "| \u001b[95m 5       \u001b[0m | \u001b[95m 0.7169  \u001b[0m | \u001b[95m 0.7967  \u001b[0m | \u001b[95m 0.7244  \u001b[0m | \u001b[95m 0.5841  \u001b[0m | \u001b[95m 0.04052 \u001b[0m | \u001b[95m 4.652   \u001b[0m | \u001b[95m 228.9   \u001b[0m | \u001b[95m 1.372e+0\u001b[0m | \u001b[95m 2.936e+0\u001b[0m | \u001b[95m 1.439   \u001b[0m | \u001b[95m 0.6501  \u001b[0m |\n",
      "| \u001b[0m 6       \u001b[0m | \u001b[0m 0.7009  \u001b[0m | \u001b[0m 0.5058  \u001b[0m | \u001b[0m 0.7037  \u001b[0m | \u001b[0m 0.5635  \u001b[0m | \u001b[0m 0.01695 \u001b[0m | \u001b[0m 10.87   \u001b[0m | \u001b[0m 35.2    \u001b[0m | \u001b[0m 903.8   \u001b[0m | \u001b[0m 1.467e+0\u001b[0m | \u001b[0m 2.947   \u001b[0m | \u001b[0m 3.499   \u001b[0m |\n",
      "| \u001b[0m 7       \u001b[0m | \u001b[0m 0.7126  \u001b[0m | \u001b[0m 0.5307  \u001b[0m | \u001b[0m 0.6242  \u001b[0m | \u001b[0m 0.7083  \u001b[0m | \u001b[0m 0.02364 \u001b[0m | \u001b[0m 3.799   \u001b[0m | \u001b[0m 272.1   \u001b[0m | \u001b[0m 1.029e+0\u001b[0m | \u001b[0m 5.149e+0\u001b[0m | \u001b[0m 4.723   \u001b[0m | \u001b[0m 2.933   \u001b[0m |\n",
      "| \u001b[0m 8       \u001b[0m | \u001b[0m 0.7006  \u001b[0m | \u001b[0m 0.771   \u001b[0m | \u001b[0m 0.5412  \u001b[0m | \u001b[0m 0.5418  \u001b[0m | \u001b[0m 0.04133 \u001b[0m | \u001b[0m 9.363   \u001b[0m | \u001b[0m 90.19   \u001b[0m | \u001b[0m 1.399e+0\u001b[0m | \u001b[0m 3.478e+0\u001b[0m | \u001b[0m 3.754   \u001b[0m | \u001b[0m 3.63    \u001b[0m |\n",
      "| \u001b[0m 9       \u001b[0m | \u001b[0m 0.7041  \u001b[0m | \u001b[0m 0.765   \u001b[0m | \u001b[0m 0.6871  \u001b[0m | \u001b[0m 0.7253  \u001b[0m | \u001b[0m 0.0207  \u001b[0m | \u001b[0m 7.319   \u001b[0m | \u001b[0m 448.9   \u001b[0m | \u001b[0m 699.3   \u001b[0m | \u001b[0m 9.648e+0\u001b[0m | \u001b[0m 3.317   \u001b[0m | \u001b[0m 3.108   \u001b[0m |\n",
      "| \u001b[0m 10      \u001b[0m | \u001b[0m 0.6926  \u001b[0m | \u001b[0m 0.5344  \u001b[0m | \u001b[0m 0.7848  \u001b[0m | \u001b[0m 0.635   \u001b[0m | \u001b[0m 0.03103 \u001b[0m | \u001b[0m 9.53    \u001b[0m | \u001b[0m 125.4   \u001b[0m | \u001b[0m 1.365e+0\u001b[0m | \u001b[0m 5.737e+0\u001b[0m | \u001b[0m 0.01435 \u001b[0m | \u001b[0m 3.086   \u001b[0m |\n",
      "| \u001b[0m 11      \u001b[0m | \u001b[0m 0.7052  \u001b[0m | \u001b[0m 0.598   \u001b[0m | \u001b[0m 0.6581  \u001b[0m | \u001b[0m 0.7658  \u001b[0m | \u001b[0m 0.02108 \u001b[0m | \u001b[0m 17.54   \u001b[0m | \u001b[0m 315.1   \u001b[0m | \u001b[0m 122.1   \u001b[0m | \u001b[0m 9.294e+0\u001b[0m | \u001b[0m 3.454   \u001b[0m | \u001b[0m 4.987   \u001b[0m |\n",
      "| \u001b[0m 12      \u001b[0m | \u001b[0m 0.6928  \u001b[0m | \u001b[0m 0.5517  \u001b[0m | \u001b[0m 0.5411  \u001b[0m | \u001b[0m 0.7798  \u001b[0m | \u001b[0m 0.03636 \u001b[0m | \u001b[0m 4.056   \u001b[0m | \u001b[0m 379.9   \u001b[0m | \u001b[0m 1.155e+0\u001b[0m | \u001b[0m 9.23e+03\u001b[0m | \u001b[0m 3.558   \u001b[0m | \u001b[0m 0.6214  \u001b[0m |\n",
      "| \u001b[0m 13      \u001b[0m | \u001b[0m 0.6273  \u001b[0m | \u001b[0m 0.506   \u001b[0m | \u001b[0m 0.5079  \u001b[0m | \u001b[0m 0.5085  \u001b[0m | \u001b[0m 0.01608 \u001b[0m | \u001b[0m 16.76   \u001b[0m | \u001b[0m 273.6   \u001b[0m | \u001b[0m 874.0   \u001b[0m | \u001b[0m 8.42e+03\u001b[0m | \u001b[0m 0.6209  \u001b[0m | \u001b[0m 1.396   \u001b[0m |\n",
      "| \u001b[0m 14      \u001b[0m | \u001b[0m 0.701   \u001b[0m | \u001b[0m 0.6757  \u001b[0m | \u001b[0m 0.7909  \u001b[0m | \u001b[0m 0.6683  \u001b[0m | \u001b[0m 0.005839\u001b[0m | \u001b[0m 15.81   \u001b[0m | \u001b[0m 123.4   \u001b[0m | \u001b[0m 1.23e+03\u001b[0m | \u001b[0m 3.879e+0\u001b[0m | \u001b[0m 4.318   \u001b[0m | \u001b[0m 3.736   \u001b[0m |\n",
      "| \u001b[0m 15      \u001b[0m | \u001b[0m 0.6823  \u001b[0m | \u001b[0m 0.6669  \u001b[0m | \u001b[0m 0.5409  \u001b[0m | \u001b[0m 0.518   \u001b[0m | \u001b[0m 0.01046 \u001b[0m | \u001b[0m 3.713   \u001b[0m | \u001b[0m 61.78   \u001b[0m | \u001b[0m 416.0   \u001b[0m | \u001b[0m 7.13e+03\u001b[0m | \u001b[0m 2.799   \u001b[0m | \u001b[0m 0.06278 \u001b[0m |\n",
      "| \u001b[0m 16      \u001b[0m | \u001b[0m 0.698   \u001b[0m | \u001b[0m 0.5216  \u001b[0m | \u001b[0m 0.7902  \u001b[0m | \u001b[0m 0.6704  \u001b[0m | \u001b[0m 0.01415 \u001b[0m | \u001b[0m 7.037   \u001b[0m | \u001b[0m 374.2   \u001b[0m | \u001b[0m 373.6   \u001b[0m | \u001b[0m 5.814e+0\u001b[0m | \u001b[0m 4.85    \u001b[0m | \u001b[0m 4.234   \u001b[0m |\n",
      "| \u001b[0m 17      \u001b[0m | \u001b[0m 0.7118  \u001b[0m | \u001b[0m 0.572   \u001b[0m | \u001b[0m 0.6481  \u001b[0m | \u001b[0m 0.686   \u001b[0m | \u001b[0m 0.0423  \u001b[0m | \u001b[0m 5.509   \u001b[0m | \u001b[0m 18.12   \u001b[0m | \u001b[0m 198.0   \u001b[0m | \u001b[0m 4.863e+0\u001b[0m | \u001b[0m 3.032   \u001b[0m | \u001b[0m 2.844   \u001b[0m |\n",
      "| \u001b[0m 18      \u001b[0m | \u001b[0m 0.7077  \u001b[0m | \u001b[0m 0.5952  \u001b[0m | \u001b[0m 0.7966  \u001b[0m | \u001b[0m 0.6739  \u001b[0m | \u001b[0m 0.02211 \u001b[0m | \u001b[0m 11.82   \u001b[0m | \u001b[0m 375.0   \u001b[0m | \u001b[0m 1.037e+0\u001b[0m | \u001b[0m 2.649e+0\u001b[0m | \u001b[0m 0.3317  \u001b[0m | \u001b[0m 1.85    \u001b[0m |\n",
      "| \u001b[0m 19      \u001b[0m | \u001b[0m 0.6651  \u001b[0m | \u001b[0m 0.6889  \u001b[0m | \u001b[0m 0.5631  \u001b[0m | \u001b[0m 0.7258  \u001b[0m | \u001b[0m 0.007994\u001b[0m | \u001b[0m 7.165   \u001b[0m | \u001b[0m 404.1   \u001b[0m | \u001b[0m 370.8   \u001b[0m | \u001b[0m 6.395e+0\u001b[0m | \u001b[0m 2.623   \u001b[0m | \u001b[0m 4.624   \u001b[0m |\n",
      "| \u001b[0m 20      \u001b[0m | \u001b[0m 0.6929  \u001b[0m | \u001b[0m 0.579   \u001b[0m | \u001b[0m 0.5198  \u001b[0m | \u001b[0m 0.7205  \u001b[0m | \u001b[0m 0.03975 \u001b[0m | \u001b[0m 17.53   \u001b[0m | \u001b[0m 466.6   \u001b[0m | \u001b[0m 119.5   \u001b[0m | \u001b[0m 2.344e+0\u001b[0m | \u001b[0m 3.084   \u001b[0m | \u001b[0m 4.745   \u001b[0m |\n",
      "| \u001b[0m 21      \u001b[0m | \u001b[0m 0.715   \u001b[0m | \u001b[0m 0.7851  \u001b[0m | \u001b[0m 0.667   \u001b[0m | \u001b[0m 0.7747  \u001b[0m | \u001b[0m 0.03387 \u001b[0m | \u001b[0m 9.24    \u001b[0m | \u001b[0m 247.6   \u001b[0m | \u001b[0m 946.0   \u001b[0m | \u001b[0m 5.495e+0\u001b[0m | \u001b[0m 4.631   \u001b[0m | \u001b[0m 4.594   \u001b[0m |\n",
      "| \u001b[0m 22      \u001b[0m | \u001b[0m 0.7116  \u001b[0m | \u001b[0m 0.6185  \u001b[0m | \u001b[0m 0.789   \u001b[0m | \u001b[0m 0.5522  \u001b[0m | \u001b[0m 0.01068 \u001b[0m | \u001b[0m 5.161   \u001b[0m | \u001b[0m 257.3   \u001b[0m | \u001b[0m 130.1   \u001b[0m | \u001b[0m 9.48e+03\u001b[0m | \u001b[0m 4.136   \u001b[0m | \u001b[0m 0.07509 \u001b[0m |\n",
      "| \u001b[0m 23      \u001b[0m | \u001b[0m 0.6829  \u001b[0m | \u001b[0m 0.5529  \u001b[0m | \u001b[0m 0.5996  \u001b[0m | \u001b[0m 0.5393  \u001b[0m | \u001b[0m 0.04143 \u001b[0m | \u001b[0m 8.516   \u001b[0m | \u001b[0m 470.6   \u001b[0m | \u001b[0m 914.8   \u001b[0m | \u001b[0m 8.788e+0\u001b[0m | \u001b[0m 4.224   \u001b[0m | \u001b[0m 4.527   \u001b[0m |\n",
      "| \u001b[0m 24      \u001b[0m | \u001b[0m 0.6968  \u001b[0m | \u001b[0m 0.638   \u001b[0m | \u001b[0m 0.6639  \u001b[0m | \u001b[0m 0.7396  \u001b[0m | \u001b[0m 0.01786 \u001b[0m | \u001b[0m 10.84   \u001b[0m | \u001b[0m 303.2   \u001b[0m | \u001b[0m 121.7   \u001b[0m | \u001b[0m 5.935e+0\u001b[0m | \u001b[0m 2.168   \u001b[0m | \u001b[0m 4.037   \u001b[0m |\n",
      "| \u001b[0m 25      \u001b[0m | \u001b[0m 0.6826  \u001b[0m | \u001b[0m 0.5946  \u001b[0m | \u001b[0m 0.7679  \u001b[0m | \u001b[0m 0.6734  \u001b[0m | \u001b[0m 0.01328 \u001b[0m | \u001b[0m 15.61   \u001b[0m | \u001b[0m 309.5   \u001b[0m | \u001b[0m 175.5   \u001b[0m | \u001b[0m 4.202e+0\u001b[0m | \u001b[0m 3.395   \u001b[0m | \u001b[0m 4.593   \u001b[0m |\n",
      "| \u001b[0m 26      \u001b[0m | \u001b[0m 0.711   \u001b[0m | \u001b[0m 0.6965  \u001b[0m | \u001b[0m 0.7663  \u001b[0m | \u001b[0m 0.6298  \u001b[0m | \u001b[0m 0.02406 \u001b[0m | \u001b[0m 18.92   \u001b[0m | \u001b[0m 407.2   \u001b[0m | \u001b[0m 1.494e+0\u001b[0m | \u001b[0m 43.42   \u001b[0m | \u001b[0m 4.792   \u001b[0m | \u001b[0m 0.3342  \u001b[0m |\n",
      "| \u001b[0m 27      \u001b[0m | \u001b[0m 0.6943  \u001b[0m | \u001b[0m 0.7836  \u001b[0m | \u001b[0m 0.7213  \u001b[0m | \u001b[0m 0.6041  \u001b[0m | \u001b[0m 0.02169 \u001b[0m | \u001b[0m 16.02   \u001b[0m | \u001b[0m 17.6    \u001b[0m | \u001b[0m 123.1   \u001b[0m | \u001b[0m 156.9   \u001b[0m | \u001b[0m 2.776   \u001b[0m | \u001b[0m 0.4354  \u001b[0m |\n",
      "| \u001b[0m 28      \u001b[0m | \u001b[0m 0.6922  \u001b[0m | \u001b[0m 0.5824  \u001b[0m | \u001b[0m 0.6792  \u001b[0m | \u001b[0m 0.7876  \u001b[0m | \u001b[0m 0.04752 \u001b[0m | \u001b[0m 10.64   \u001b[0m | \u001b[0m 486.4   \u001b[0m | \u001b[0m 1.484e+0\u001b[0m | \u001b[0m 7.855e+0\u001b[0m | \u001b[0m 1.71    \u001b[0m | \u001b[0m 1.81    \u001b[0m |\n",
      "| \u001b[0m 29      \u001b[0m | \u001b[0m 0.7046  \u001b[0m | \u001b[0m 0.6602  \u001b[0m | \u001b[0m 0.7792  \u001b[0m | \u001b[0m 0.6692  \u001b[0m | \u001b[0m 0.01096 \u001b[0m | \u001b[0m 11.87   \u001b[0m | \u001b[0m 491.6   \u001b[0m | \u001b[0m 1.45e+03\u001b[0m | \u001b[0m 1.471e+0\u001b[0m | \u001b[0m 3.445   \u001b[0m | \u001b[0m 3.225   \u001b[0m |\n",
      "| \u001b[0m 30      \u001b[0m | \u001b[0m 0.6955  \u001b[0m | \u001b[0m 0.5275  \u001b[0m | \u001b[0m 0.7299  \u001b[0m | \u001b[0m 0.6705  \u001b[0m | \u001b[0m 0.01158 \u001b[0m | \u001b[0m 3.332   \u001b[0m | \u001b[0m 488.7   \u001b[0m | \u001b[0m 159.7   \u001b[0m | \u001b[0m 9.984e+0\u001b[0m | \u001b[0m 3.188   \u001b[0m | \u001b[0m 3.28    \u001b[0m |\n",
      "| \u001b[95m 31      \u001b[0m | \u001b[95m 0.7174  \u001b[0m | \u001b[95m 0.7992  \u001b[0m | \u001b[95m 0.7095  \u001b[0m | \u001b[95m 0.7431  \u001b[0m | \u001b[95m 0.03491 \u001b[0m | \u001b[95m 12.33   \u001b[0m | \u001b[95m 262.9   \u001b[0m | \u001b[95m 930.6   \u001b[0m | \u001b[95m 5.553e+0\u001b[0m | \u001b[95m 1.452   \u001b[0m | \u001b[95m 1.367   \u001b[0m |\n",
      "| \u001b[0m 32      \u001b[0m | \u001b[0m 0.705   \u001b[0m | \u001b[0m 0.6739  \u001b[0m | \u001b[0m 0.6113  \u001b[0m | \u001b[0m 0.679   \u001b[0m | \u001b[0m 0.03209 \u001b[0m | \u001b[0m 14.1    \u001b[0m | \u001b[0m 180.4   \u001b[0m | \u001b[0m 1.363e+0\u001b[0m | \u001b[0m 2.922e+0\u001b[0m | \u001b[0m 0.7654  \u001b[0m | \u001b[0m 4.766   \u001b[0m |\n",
      "| \u001b[0m 33      \u001b[0m | \u001b[0m 0.6316  \u001b[0m | \u001b[0m 0.5039  \u001b[0m | \u001b[0m 0.5034  \u001b[0m | \u001b[0m 0.7057  \u001b[0m | \u001b[0m 0.03297 \u001b[0m | \u001b[0m 17.6    \u001b[0m | \u001b[0m 490.8   \u001b[0m | \u001b[0m 1.493e+0\u001b[0m | \u001b[0m 3.797e+0\u001b[0m | \u001b[0m 3.045   \u001b[0m | \u001b[0m 2.044   \u001b[0m |\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "| \u001b[0m 34      \u001b[0m | \u001b[0m 0.7119  \u001b[0m | \u001b[0m 0.6725  \u001b[0m | \u001b[0m 0.7656  \u001b[0m | \u001b[0m 0.7213  \u001b[0m | \u001b[0m 0.01457 \u001b[0m | \u001b[0m 4.849   \u001b[0m | \u001b[0m 16.45   \u001b[0m | \u001b[0m 1.376e+0\u001b[0m | \u001b[0m 6.965e+0\u001b[0m | \u001b[0m 2.512   \u001b[0m | \u001b[0m 3.108   \u001b[0m |\n",
      "| \u001b[0m 35      \u001b[0m | \u001b[0m 0.7051  \u001b[0m | \u001b[0m 0.7688  \u001b[0m | \u001b[0m 0.7634  \u001b[0m | \u001b[0m 0.7207  \u001b[0m | \u001b[0m 0.01053 \u001b[0m | \u001b[0m 6.225   \u001b[0m | \u001b[0m 52.08   \u001b[0m | \u001b[0m 102.7   \u001b[0m | \u001b[0m 1.34e+03\u001b[0m | \u001b[0m 4.279   \u001b[0m | \u001b[0m 0.2561  \u001b[0m |\n",
      "| \u001b[95m 36      \u001b[0m | \u001b[95m 0.7197  \u001b[0m | \u001b[95m 0.7023  \u001b[0m | \u001b[95m 0.7937  \u001b[0m | \u001b[95m 0.7896  \u001b[0m | \u001b[95m 0.04092 \u001b[0m | \u001b[95m 3.127   \u001b[0m | \u001b[95m 18.81   \u001b[0m | \u001b[95m 1.492e+0\u001b[0m | \u001b[95m 87.62   \u001b[0m | \u001b[95m 4.199   \u001b[0m | \u001b[95m 0.6669  \u001b[0m |\n",
      "| \u001b[0m 37      \u001b[0m | \u001b[0m 0.679   \u001b[0m | \u001b[0m 0.6437  \u001b[0m | \u001b[0m 0.5721  \u001b[0m | \u001b[0m 0.5211  \u001b[0m | \u001b[0m 0.01152 \u001b[0m | \u001b[0m 12.64   \u001b[0m | \u001b[0m 14.71   \u001b[0m | \u001b[0m 438.2   \u001b[0m | \u001b[0m 6.138e+0\u001b[0m | \u001b[0m 2.763   \u001b[0m | \u001b[0m 3.765   \u001b[0m |\n",
      "| \u001b[0m 38      \u001b[0m | \u001b[0m 0.6891  \u001b[0m | \u001b[0m 0.6298  \u001b[0m | \u001b[0m 0.7013  \u001b[0m | \u001b[0m 0.5447  \u001b[0m | \u001b[0m 0.01743 \u001b[0m | \u001b[0m 8.425   \u001b[0m | \u001b[0m 33.37   \u001b[0m | \u001b[0m 1.497e+0\u001b[0m | \u001b[0m 9.999e+0\u001b[0m | \u001b[0m 0.3128  \u001b[0m | \u001b[0m 0.5583  \u001b[0m |\n",
      "| \u001b[0m 39      \u001b[0m | \u001b[0m 0.7026  \u001b[0m | \u001b[0m 0.6247  \u001b[0m | \u001b[0m 0.6032  \u001b[0m | \u001b[0m 0.6987  \u001b[0m | \u001b[0m 0.04658 \u001b[0m | \u001b[0m 4.015   \u001b[0m | \u001b[0m 65.54   \u001b[0m | \u001b[0m 895.1   \u001b[0m | \u001b[0m 566.4   \u001b[0m | \u001b[0m 0.7886  \u001b[0m | \u001b[0m 0.4939  \u001b[0m |\n",
      "| \u001b[0m 40      \u001b[0m | \u001b[0m 0.6855  \u001b[0m | \u001b[0m 0.7398  \u001b[0m | \u001b[0m 0.579   \u001b[0m | \u001b[0m 0.5083  \u001b[0m | \u001b[0m 0.0304  \u001b[0m | \u001b[0m 4.55    \u001b[0m | \u001b[0m 486.7   \u001b[0m | \u001b[0m 1.499e+0\u001b[0m | \u001b[0m 9.953e+0\u001b[0m | \u001b[0m 2.01    \u001b[0m | \u001b[0m 4.553   \u001b[0m |\n",
      "| \u001b[0m 41      \u001b[0m | \u001b[0m 0.6696  \u001b[0m | \u001b[0m 0.7691  \u001b[0m | \u001b[0m 0.5088  \u001b[0m | \u001b[0m 0.5859  \u001b[0m | \u001b[0m 0.03708 \u001b[0m | \u001b[0m 7.274   \u001b[0m | \u001b[0m 482.5   \u001b[0m | \u001b[0m 607.6   \u001b[0m | \u001b[0m 1.43e+03\u001b[0m | \u001b[0m 0.1369  \u001b[0m | \u001b[0m 0.0383  \u001b[0m |\n",
      "| \u001b[0m 42      \u001b[0m | \u001b[0m 0.6231  \u001b[0m | \u001b[0m 0.6362  \u001b[0m | \u001b[0m 0.6332  \u001b[0m | \u001b[0m 0.5003  \u001b[0m | \u001b[0m 0.02051 \u001b[0m | \u001b[0m 12.68   \u001b[0m | \u001b[0m 15.38   \u001b[0m | \u001b[0m 1.489e+0\u001b[0m | \u001b[0m 8.35e+03\u001b[0m | \u001b[0m 2.417   \u001b[0m | \u001b[0m 2.259   \u001b[0m |\n",
      "| \u001b[0m 43      \u001b[0m | \u001b[0m 0.697   \u001b[0m | \u001b[0m 0.7153  \u001b[0m | \u001b[0m 0.7292  \u001b[0m | \u001b[0m 0.5852  \u001b[0m | \u001b[0m 0.04555 \u001b[0m | \u001b[0m 11.14   \u001b[0m | \u001b[0m 497.4   \u001b[0m | \u001b[0m 1.299e+0\u001b[0m | \u001b[0m 6.456e+0\u001b[0m | \u001b[0m 2.217   \u001b[0m | \u001b[0m 2.474   \u001b[0m |\n",
      "| \u001b[0m 44      \u001b[0m | \u001b[0m 0.7147  \u001b[0m | \u001b[0m 0.7498  \u001b[0m | \u001b[0m 0.719   \u001b[0m | \u001b[0m 0.7912  \u001b[0m | \u001b[0m 0.01481 \u001b[0m | \u001b[0m 7.434   \u001b[0m | \u001b[0m 45.03   \u001b[0m | \u001b[0m 1.498e+0\u001b[0m | \u001b[0m 970.9   \u001b[0m | \u001b[0m 2.771   \u001b[0m | \u001b[0m 0.8147  \u001b[0m |\n",
      "| \u001b[0m 45      \u001b[0m | \u001b[0m 0.6826  \u001b[0m | \u001b[0m 0.5178  \u001b[0m | \u001b[0m 0.7311  \u001b[0m | \u001b[0m 0.5378  \u001b[0m | \u001b[0m 0.04708 \u001b[0m | \u001b[0m 3.087   \u001b[0m | \u001b[0m 498.0   \u001b[0m | \u001b[0m 779.9   \u001b[0m | \u001b[0m 3.432   \u001b[0m | \u001b[0m 3.453   \u001b[0m | \u001b[0m 1.929   \u001b[0m |\n",
      "| \u001b[0m 46      \u001b[0m | \u001b[0m 0.6837  \u001b[0m | \u001b[0m 0.6338  \u001b[0m | \u001b[0m 0.6031  \u001b[0m | \u001b[0m 0.7438  \u001b[0m | \u001b[0m 0.03594 \u001b[0m | \u001b[0m 18.99   \u001b[0m | \u001b[0m 22.51   \u001b[0m | \u001b[0m 146.1   \u001b[0m | \u001b[0m 9.999e+0\u001b[0m | \u001b[0m 2.765   \u001b[0m | \u001b[0m 2.106   \u001b[0m |\n",
      "| \u001b[0m 47      \u001b[0m | \u001b[0m 0.7017  \u001b[0m | \u001b[0m 0.6927  \u001b[0m | \u001b[0m 0.5084  \u001b[0m | \u001b[0m 0.5073  \u001b[0m | \u001b[0m 0.03637 \u001b[0m | \u001b[0m 3.745   \u001b[0m | \u001b[0m 436.0   \u001b[0m | \u001b[0m 1.491e+0\u001b[0m | \u001b[0m 2.315e+0\u001b[0m | \u001b[0m 3.819   \u001b[0m | \u001b[0m 3.8     \u001b[0m |\n",
      "| \u001b[0m 48      \u001b[0m | \u001b[0m 0.6609  \u001b[0m | \u001b[0m 0.7496  \u001b[0m | \u001b[0m 0.5026  \u001b[0m | \u001b[0m 0.6315  \u001b[0m | \u001b[0m 0.02415 \u001b[0m | \u001b[0m 9.762   \u001b[0m | \u001b[0m 18.74   \u001b[0m | \u001b[0m 150.8   \u001b[0m | \u001b[0m 2.535e+0\u001b[0m | \u001b[0m 0.8797  \u001b[0m | \u001b[0m 1.695   \u001b[0m |\n",
      "| \u001b[0m 49      \u001b[0m | \u001b[0m 0.7153  \u001b[0m | \u001b[0m 0.7634  \u001b[0m | \u001b[0m 0.6344  \u001b[0m | \u001b[0m 0.7784  \u001b[0m | \u001b[0m 0.01195 \u001b[0m | \u001b[0m 4.088   \u001b[0m | \u001b[0m 484.9   \u001b[0m | \u001b[0m 854.2   \u001b[0m | \u001b[0m 3.351e+0\u001b[0m | \u001b[0m 2.756   \u001b[0m | \u001b[0m 1.677   \u001b[0m |\n",
      "| \u001b[0m 50      \u001b[0m | \u001b[0m 0.7024  \u001b[0m | \u001b[0m 0.7551  \u001b[0m | \u001b[0m 0.6058  \u001b[0m | \u001b[0m 0.5971  \u001b[0m | \u001b[0m 0.007927\u001b[0m | \u001b[0m 13.27   \u001b[0m | \u001b[0m 11.38   \u001b[0m | \u001b[0m 1.169e+0\u001b[0m | \u001b[0m 7.449e+0\u001b[0m | \u001b[0m 3.944   \u001b[0m | \u001b[0m 3.692   \u001b[0m |\n",
      "| \u001b[0m 51      \u001b[0m | \u001b[0m 0.7018  \u001b[0m | \u001b[0m 0.5206  \u001b[0m | \u001b[0m 0.5537  \u001b[0m | \u001b[0m 0.7204  \u001b[0m | \u001b[0m 0.04879 \u001b[0m | \u001b[0m 10.18   \u001b[0m | \u001b[0m 15.78   \u001b[0m | \u001b[0m 884.0   \u001b[0m | \u001b[0m 4.624e+0\u001b[0m | \u001b[0m 4.929   \u001b[0m | \u001b[0m 1.565   \u001b[0m |\n",
      "| \u001b[0m 52      \u001b[0m | \u001b[0m 0.6845  \u001b[0m | \u001b[0m 0.6259  \u001b[0m | \u001b[0m 0.5394  \u001b[0m | \u001b[0m 0.5402  \u001b[0m | \u001b[0m 0.009536\u001b[0m | \u001b[0m 5.394   \u001b[0m | \u001b[0m 103.8   \u001b[0m | \u001b[0m 1.397e+0\u001b[0m | \u001b[0m 3.473e+0\u001b[0m | \u001b[0m 0.222   \u001b[0m | \u001b[0m 1.875   \u001b[0m |\n",
      "| \u001b[0m 53      \u001b[0m | \u001b[0m 0.6893  \u001b[0m | \u001b[0m 0.6433  \u001b[0m | \u001b[0m 0.7383  \u001b[0m | \u001b[0m 0.5426  \u001b[0m | \u001b[0m 0.04948 \u001b[0m | \u001b[0m 10.89   \u001b[0m | \u001b[0m 15.44   \u001b[0m | \u001b[0m 1.484e+0\u001b[0m | \u001b[0m 4.726e+0\u001b[0m | \u001b[0m 4.244   \u001b[0m | \u001b[0m 2.356   \u001b[0m |\n",
      "| \u001b[0m 54      \u001b[0m | \u001b[0m 0.6997  \u001b[0m | \u001b[0m 0.6831  \u001b[0m | \u001b[0m 0.6211  \u001b[0m | \u001b[0m 0.6903  \u001b[0m | \u001b[0m 0.0492  \u001b[0m | \u001b[0m 12.73   \u001b[0m | \u001b[0m 468.7   \u001b[0m | \u001b[0m 102.0   \u001b[0m | \u001b[0m 3.055e+0\u001b[0m | \u001b[0m 4.988   \u001b[0m | \u001b[0m 0.7586  \u001b[0m |\n",
      "| \u001b[0m 55      \u001b[0m | \u001b[0m 0.7169  \u001b[0m | \u001b[0m 0.7448  \u001b[0m | \u001b[0m 0.7798  \u001b[0m | \u001b[0m 0.6661  \u001b[0m | \u001b[0m 0.02351 \u001b[0m | \u001b[0m 12.81   \u001b[0m | \u001b[0m 493.0   \u001b[0m | \u001b[0m 114.8   \u001b[0m | \u001b[0m 4.994e+0\u001b[0m | \u001b[0m 3.122   \u001b[0m | \u001b[0m 2.704   \u001b[0m |\n",
      "| \u001b[0m 56      \u001b[0m | \u001b[0m 0.7134  \u001b[0m | \u001b[0m 0.7956  \u001b[0m | \u001b[0m 0.5126  \u001b[0m | \u001b[0m 0.7747  \u001b[0m | \u001b[0m 0.04989 \u001b[0m | \u001b[0m 8.004   \u001b[0m | \u001b[0m 495.5   \u001b[0m | \u001b[0m 1.428e+0\u001b[0m | \u001b[0m 5.481e+0\u001b[0m | \u001b[0m 2.015   \u001b[0m | \u001b[0m 4.389   \u001b[0m |\n",
      "| \u001b[0m 57      \u001b[0m | \u001b[0m 0.7181  \u001b[0m | \u001b[0m 0.7521  \u001b[0m | \u001b[0m 0.5708  \u001b[0m | \u001b[0m 0.6855  \u001b[0m | \u001b[0m 0.02933 \u001b[0m | \u001b[0m 12.33   \u001b[0m | \u001b[0m 473.6   \u001b[0m | \u001b[0m 1.463e+0\u001b[0m | \u001b[0m 2.984e+0\u001b[0m | \u001b[0m 3.985   \u001b[0m | \u001b[0m 4.863   \u001b[0m |\n",
      "| \u001b[0m 58      \u001b[0m | \u001b[0m 0.6981  \u001b[0m | \u001b[0m 0.6526  \u001b[0m | \u001b[0m 0.6741  \u001b[0m | \u001b[0m 0.522   \u001b[0m | \u001b[0m 0.04277 \u001b[0m | \u001b[0m 8.767   \u001b[0m | \u001b[0m 499.8   \u001b[0m | \u001b[0m 147.7   \u001b[0m | \u001b[0m 7.897   \u001b[0m | \u001b[0m 4.743   \u001b[0m | \u001b[0m 1.995   \u001b[0m |\n",
      "| \u001b[0m 59      \u001b[0m | \u001b[0m 0.7084  \u001b[0m | \u001b[0m 0.738   \u001b[0m | \u001b[0m 0.7423  \u001b[0m | \u001b[0m 0.5749  \u001b[0m | \u001b[0m 0.01907 \u001b[0m | \u001b[0m 12.79   \u001b[0m | \u001b[0m 448.3   \u001b[0m | \u001b[0m 1.475e+0\u001b[0m | \u001b[0m 596.1   \u001b[0m | \u001b[0m 1.572   \u001b[0m | \u001b[0m 2.69    \u001b[0m |\n",
      "| \u001b[0m 60      \u001b[0m | \u001b[0m 0.6938  \u001b[0m | \u001b[0m 0.7388  \u001b[0m | \u001b[0m 0.6424  \u001b[0m | \u001b[0m 0.508   \u001b[0m | \u001b[0m 0.04635 \u001b[0m | \u001b[0m 10.55   \u001b[0m | \u001b[0m 25.21   \u001b[0m | \u001b[0m 615.0   \u001b[0m | \u001b[0m 9.275e+0\u001b[0m | \u001b[0m 2.136   \u001b[0m | \u001b[0m 1.463   \u001b[0m |\n",
      "| \u001b[0m 61      \u001b[0m | \u001b[0m 0.6944  \u001b[0m | \u001b[0m 0.5549  \u001b[0m | \u001b[0m 0.5705  \u001b[0m | \u001b[0m 0.7252  \u001b[0m | \u001b[0m 0.03128 \u001b[0m | \u001b[0m 16.58   \u001b[0m | \u001b[0m 20.96   \u001b[0m | \u001b[0m 832.2   \u001b[0m | \u001b[0m 6.773e+0\u001b[0m | \u001b[0m 4.455   \u001b[0m | \u001b[0m 3.604   \u001b[0m |\n",
      "| \u001b[0m 62      \u001b[0m | \u001b[0m 0.7181  \u001b[0m | \u001b[0m 0.6431  \u001b[0m | \u001b[0m 0.5282  \u001b[0m | \u001b[0m 0.7873  \u001b[0m | \u001b[0m 0.02996 \u001b[0m | \u001b[0m 18.73   \u001b[0m | \u001b[0m 12.63   \u001b[0m | \u001b[0m 106.1   \u001b[0m | \u001b[0m 7.793e+0\u001b[0m | \u001b[0m 2.291   \u001b[0m | \u001b[0m 2.266   \u001b[0m |\n",
      "| \u001b[0m 63      \u001b[0m | \u001b[0m 0.7139  \u001b[0m | \u001b[0m 0.7414  \u001b[0m | \u001b[0m 0.6583  \u001b[0m | \u001b[0m 0.7686  \u001b[0m | \u001b[0m 0.01593 \u001b[0m | \u001b[0m 7.835   \u001b[0m | \u001b[0m 11.96   \u001b[0m | \u001b[0m 102.3   \u001b[0m | \u001b[0m 6.96e+03\u001b[0m | \u001b[0m 1.485   \u001b[0m | \u001b[0m 1.453   \u001b[0m |\n",
      "| \u001b[0m 64      \u001b[0m | \u001b[0m 0.6946  \u001b[0m | \u001b[0m 0.6197  \u001b[0m | \u001b[0m 0.6043  \u001b[0m | \u001b[0m 0.6148  \u001b[0m | \u001b[0m 0.04814 \u001b[0m | \u001b[0m 17.72   \u001b[0m | \u001b[0m 495.0   \u001b[0m | \u001b[0m 112.4   \u001b[0m | \u001b[0m 7.688e+0\u001b[0m | \u001b[0m 0.7212  \u001b[0m | \u001b[0m 0.8996  \u001b[0m |\n",
      "| \u001b[0m 65      \u001b[0m | \u001b[0m 0.713   \u001b[0m | \u001b[0m 0.7875  \u001b[0m | \u001b[0m 0.7883  \u001b[0m | \u001b[0m 0.6003  \u001b[0m | \u001b[0m 0.04204 \u001b[0m | \u001b[0m 3.131   \u001b[0m | \u001b[0m 9.823   \u001b[0m | \u001b[0m 864.6   \u001b[0m | \u001b[0m 9.984e+0\u001b[0m | \u001b[0m 0.493   \u001b[0m | \u001b[0m 2.914   \u001b[0m |\n",
      "| \u001b[0m 66      \u001b[0m | \u001b[0m 0.71    \u001b[0m | \u001b[0m 0.5665  \u001b[0m | \u001b[0m 0.7798  \u001b[0m | \u001b[0m 0.6221  \u001b[0m | \u001b[0m 0.03459 \u001b[0m | \u001b[0m 13.0    \u001b[0m | \u001b[0m 497.2   \u001b[0m | \u001b[0m 1.253e+0\u001b[0m | \u001b[0m 7.185e+0\u001b[0m | \u001b[0m 4.211   \u001b[0m | \u001b[0m 2.514   \u001b[0m |\n",
      "| \u001b[0m 67      \u001b[0m | \u001b[0m 0.7128  \u001b[0m | \u001b[0m 0.7486  \u001b[0m | \u001b[0m 0.6384  \u001b[0m | \u001b[0m 0.6767  \u001b[0m | \u001b[0m 0.01362 \u001b[0m | \u001b[0m 17.6    \u001b[0m | \u001b[0m 495.0   \u001b[0m | \u001b[0m 1.176e+0\u001b[0m | \u001b[0m 4.614e+0\u001b[0m | \u001b[0m 3.332   \u001b[0m | \u001b[0m 3.0     \u001b[0m |\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "| \u001b[0m 68      \u001b[0m | \u001b[0m 0.698   \u001b[0m | \u001b[0m 0.7675  \u001b[0m | \u001b[0m 0.6731  \u001b[0m | \u001b[0m 0.6363  \u001b[0m | \u001b[0m 0.03725 \u001b[0m | \u001b[0m 14.11   \u001b[0m | \u001b[0m 499.3   \u001b[0m | \u001b[0m 1.06e+03\u001b[0m | \u001b[0m 5.889e+0\u001b[0m | \u001b[0m 3.118   \u001b[0m | \u001b[0m 1.198   \u001b[0m |\n",
      "| \u001b[0m 69      \u001b[0m | \u001b[0m 0.7123  \u001b[0m | \u001b[0m 0.738   \u001b[0m | \u001b[0m 0.7876  \u001b[0m | \u001b[0m 0.6717  \u001b[0m | \u001b[0m 0.03863 \u001b[0m | \u001b[0m 5.647   \u001b[0m | \u001b[0m 10.68   \u001b[0m | \u001b[0m 100.5   \u001b[0m | \u001b[0m 9.17e+03\u001b[0m | \u001b[0m 4.612   \u001b[0m | \u001b[0m 4.202   \u001b[0m |\n",
      "| \u001b[0m 70      \u001b[0m | \u001b[0m 0.6967  \u001b[0m | \u001b[0m 0.6352  \u001b[0m | \u001b[0m 0.5356  \u001b[0m | \u001b[0m 0.7095  \u001b[0m | \u001b[0m 0.0478  \u001b[0m | \u001b[0m 14.83   \u001b[0m | \u001b[0m 89.46   \u001b[0m | \u001b[0m 101.8   \u001b[0m | \u001b[0m 5.381e+0\u001b[0m | \u001b[0m 2.717   \u001b[0m | \u001b[0m 2.772   \u001b[0m |\n",
      "| \u001b[0m 71      \u001b[0m | \u001b[0m 0.6839  \u001b[0m | \u001b[0m 0.5124  \u001b[0m | \u001b[0m 0.5715  \u001b[0m | \u001b[0m 0.7429  \u001b[0m | \u001b[0m 0.01917 \u001b[0m | \u001b[0m 16.94   \u001b[0m | \u001b[0m 475.1   \u001b[0m | \u001b[0m 107.5   \u001b[0m | \u001b[0m 7.05e+03\u001b[0m | \u001b[0m 3.229   \u001b[0m | \u001b[0m 0.7316  \u001b[0m |\n",
      "| \u001b[0m 72      \u001b[0m | \u001b[0m 0.6985  \u001b[0m | \u001b[0m 0.6873  \u001b[0m | \u001b[0m 0.6634  \u001b[0m | \u001b[0m 0.6375  \u001b[0m | \u001b[0m 0.02744 \u001b[0m | \u001b[0m 17.67   \u001b[0m | \u001b[0m 497.8   \u001b[0m | \u001b[0m 1.499e+0\u001b[0m | \u001b[0m 8.799e+0\u001b[0m | \u001b[0m 4.939   \u001b[0m | \u001b[0m 0.7669  \u001b[0m |\n",
      "| \u001b[0m 73      \u001b[0m | \u001b[0m 0.7026  \u001b[0m | \u001b[0m 0.7046  \u001b[0m | \u001b[0m 0.7326  \u001b[0m | \u001b[0m 0.6266  \u001b[0m | \u001b[0m 0.01285 \u001b[0m | \u001b[0m 7.621   \u001b[0m | \u001b[0m 480.3   \u001b[0m | \u001b[0m 103.3   \u001b[0m | \u001b[0m 989.0   \u001b[0m | \u001b[0m 1.787   \u001b[0m | \u001b[0m 3.077   \u001b[0m |\n",
      "| \u001b[0m 74      \u001b[0m | \u001b[0m 0.6928  \u001b[0m | \u001b[0m 0.5138  \u001b[0m | \u001b[0m 0.706   \u001b[0m | \u001b[0m 0.5295  \u001b[0m | \u001b[0m 0.0489  \u001b[0m | \u001b[0m 15.3    \u001b[0m | \u001b[0m 488.0   \u001b[0m | \u001b[0m 560.4   \u001b[0m | \u001b[0m 4.65e+03\u001b[0m | \u001b[0m 4.786   \u001b[0m | \u001b[0m 4.583   \u001b[0m |\n",
      "| \u001b[0m 75      \u001b[0m | \u001b[0m 0.6529  \u001b[0m | \u001b[0m 0.5524  \u001b[0m | \u001b[0m 0.7822  \u001b[0m | \u001b[0m 0.5814  \u001b[0m | \u001b[0m 0.009893\u001b[0m | \u001b[0m 10.73   \u001b[0m | \u001b[0m 58.76   \u001b[0m | \u001b[0m 106.9   \u001b[0m | \u001b[0m 722.9   \u001b[0m | \u001b[0m 3.664   \u001b[0m | \u001b[0m 0.7733  \u001b[0m |\n",
      "| \u001b[0m 76      \u001b[0m | \u001b[0m 0.6827  \u001b[0m | \u001b[0m 0.5245  \u001b[0m | \u001b[0m 0.5836  \u001b[0m | \u001b[0m 0.6348  \u001b[0m | \u001b[0m 0.01871 \u001b[0m | \u001b[0m 4.336   \u001b[0m | \u001b[0m 14.31   \u001b[0m | \u001b[0m 1.488e+0\u001b[0m | \u001b[0m 1.675e+0\u001b[0m | \u001b[0m 4.494   \u001b[0m | \u001b[0m 0.6443  \u001b[0m |\n",
      "| \u001b[0m 77      \u001b[0m | \u001b[0m 0.6952  \u001b[0m | \u001b[0m 0.6194  \u001b[0m | \u001b[0m 0.5147  \u001b[0m | \u001b[0m 0.7213  \u001b[0m | \u001b[0m 0.02495 \u001b[0m | \u001b[0m 17.95   \u001b[0m | \u001b[0m 9.489   \u001b[0m | \u001b[0m 624.0   \u001b[0m | \u001b[0m 8.917   \u001b[0m | \u001b[0m 0.4153  \u001b[0m | \u001b[0m 2.893   \u001b[0m |\n",
      "| \u001b[0m 78      \u001b[0m | \u001b[0m 0.6303  \u001b[0m | \u001b[0m 0.508   \u001b[0m | \u001b[0m 0.6388  \u001b[0m | \u001b[0m 0.5488  \u001b[0m | \u001b[0m 0.01366 \u001b[0m | \u001b[0m 3.412   \u001b[0m | \u001b[0m 471.3   \u001b[0m | \u001b[0m 118.0   \u001b[0m | \u001b[0m 1.602e+0\u001b[0m | \u001b[0m 4.021   \u001b[0m | \u001b[0m 1.179   \u001b[0m |\n",
      "| \u001b[0m 79      \u001b[0m | \u001b[0m 0.7003  \u001b[0m | \u001b[0m 0.6243  \u001b[0m | \u001b[0m 0.5269  \u001b[0m | \u001b[0m 0.5413  \u001b[0m | \u001b[0m 0.007757\u001b[0m | \u001b[0m 3.097   \u001b[0m | \u001b[0m 471.9   \u001b[0m | \u001b[0m 903.9   \u001b[0m | \u001b[0m 908.3   \u001b[0m | \u001b[0m 4.924   \u001b[0m | \u001b[0m 1.792   \u001b[0m |\n",
      "| \u001b[0m 80      \u001b[0m | \u001b[0m 0.6953  \u001b[0m | \u001b[0m 0.7271  \u001b[0m | \u001b[0m 0.5107  \u001b[0m | \u001b[0m 0.5926  \u001b[0m | \u001b[0m 0.01457 \u001b[0m | \u001b[0m 6.739   \u001b[0m | \u001b[0m 471.6   \u001b[0m | \u001b[0m 557.4   \u001b[0m | \u001b[0m 2.867e+0\u001b[0m | \u001b[0m 3.503   \u001b[0m | \u001b[0m 1.205   \u001b[0m |\n",
      "| \u001b[0m 81      \u001b[0m | \u001b[0m 0.6736  \u001b[0m | \u001b[0m 0.7338  \u001b[0m | \u001b[0m 0.6656  \u001b[0m | \u001b[0m 0.6693  \u001b[0m | \u001b[0m 0.04458 \u001b[0m | \u001b[0m 18.71   \u001b[0m | \u001b[0m 24.63   \u001b[0m | \u001b[0m 687.1   \u001b[0m | \u001b[0m 3.902e+0\u001b[0m | \u001b[0m 3.877   \u001b[0m | \u001b[0m 4.099   \u001b[0m |\n",
      "| \u001b[0m 82      \u001b[0m | \u001b[0m 0.6847  \u001b[0m | \u001b[0m 0.5172  \u001b[0m | \u001b[0m 0.6342  \u001b[0m | \u001b[0m 0.5213  \u001b[0m | \u001b[0m 0.02434 \u001b[0m | \u001b[0m 9.072   \u001b[0m | \u001b[0m 12.56   \u001b[0m | \u001b[0m 119.6   \u001b[0m | \u001b[0m 3.616e+0\u001b[0m | \u001b[0m 4.534   \u001b[0m | \u001b[0m 3.065   \u001b[0m |\n",
      "| \u001b[0m 83      \u001b[0m | \u001b[0m 0.7024  \u001b[0m | \u001b[0m 0.5735  \u001b[0m | \u001b[0m 0.7211  \u001b[0m | \u001b[0m 0.7682  \u001b[0m | \u001b[0m 0.04748 \u001b[0m | \u001b[0m 10.99   \u001b[0m | \u001b[0m 453.7   \u001b[0m | \u001b[0m 1.496e+0\u001b[0m | \u001b[0m 7.064e+0\u001b[0m | \u001b[0m 4.886   \u001b[0m | \u001b[0m 4.207   \u001b[0m |\n",
      "| \u001b[0m 84      \u001b[0m | \u001b[0m 0.67    \u001b[0m | \u001b[0m 0.5196  \u001b[0m | \u001b[0m 0.6202  \u001b[0m | \u001b[0m 0.7062  \u001b[0m | \u001b[0m 0.04858 \u001b[0m | \u001b[0m 16.0    \u001b[0m | \u001b[0m 23.02   \u001b[0m | \u001b[0m 1.492e+0\u001b[0m | \u001b[0m 3.936e+0\u001b[0m | \u001b[0m 4.049   \u001b[0m | \u001b[0m 1.14    \u001b[0m |\n",
      "| \u001b[0m 85      \u001b[0m | \u001b[0m 0.6801  \u001b[0m | \u001b[0m 0.6771  \u001b[0m | \u001b[0m 0.6871  \u001b[0m | \u001b[0m 0.586   \u001b[0m | \u001b[0m 0.01686 \u001b[0m | \u001b[0m 5.13    \u001b[0m | \u001b[0m 490.0   \u001b[0m | \u001b[0m 1.43e+03\u001b[0m | \u001b[0m 1.889e+0\u001b[0m | \u001b[0m 4.699   \u001b[0m | \u001b[0m 2.393   \u001b[0m |\n",
      "| \u001b[0m 86      \u001b[0m | \u001b[0m 0.6914  \u001b[0m | \u001b[0m 0.5733  \u001b[0m | \u001b[0m 0.783   \u001b[0m | \u001b[0m 0.7006  \u001b[0m | \u001b[0m 0.005229\u001b[0m | \u001b[0m 16.32   \u001b[0m | \u001b[0m 14.37   \u001b[0m | \u001b[0m 1.483e+0\u001b[0m | \u001b[0m 6.425e+0\u001b[0m | \u001b[0m 2.648   \u001b[0m | \u001b[0m 3.908   \u001b[0m |\n",
      "| \u001b[0m 87      \u001b[0m | \u001b[0m 0.7165  \u001b[0m | \u001b[0m 0.7648  \u001b[0m | \u001b[0m 0.7752  \u001b[0m | \u001b[0m 0.7158  \u001b[0m | \u001b[0m 0.02076 \u001b[0m | \u001b[0m 16.96   \u001b[0m | \u001b[0m 98.07   \u001b[0m | \u001b[0m 857.9   \u001b[0m | \u001b[0m 3.307e+0\u001b[0m | \u001b[0m 3.398   \u001b[0m | \u001b[0m 2.922   \u001b[0m |\n",
      "| \u001b[0m 88      \u001b[0m | \u001b[0m 0.67    \u001b[0m | \u001b[0m 0.5773  \u001b[0m | \u001b[0m 0.6213  \u001b[0m | \u001b[0m 0.5085  \u001b[0m | \u001b[0m 0.02703 \u001b[0m | \u001b[0m 15.98   \u001b[0m | \u001b[0m 400.5   \u001b[0m | \u001b[0m 1.482e+0\u001b[0m | \u001b[0m 4.937e+0\u001b[0m | \u001b[0m 2.94    \u001b[0m | \u001b[0m 0.388   \u001b[0m |\n",
      "| \u001b[0m 89      \u001b[0m | \u001b[0m 0.6977  \u001b[0m | \u001b[0m 0.5652  \u001b[0m | \u001b[0m 0.756   \u001b[0m | \u001b[0m 0.6911  \u001b[0m | \u001b[0m 0.01131 \u001b[0m | \u001b[0m 5.539   \u001b[0m | \u001b[0m 475.2   \u001b[0m | \u001b[0m 1.112e+0\u001b[0m | \u001b[0m 3.042e+0\u001b[0m | \u001b[0m 4.905   \u001b[0m | \u001b[0m 4.748   \u001b[0m |\n",
      "| \u001b[0m 90      \u001b[0m | \u001b[0m 0.6818  \u001b[0m | \u001b[0m 0.5558  \u001b[0m | \u001b[0m 0.6954  \u001b[0m | \u001b[0m 0.5623  \u001b[0m | \u001b[0m 0.008944\u001b[0m | \u001b[0m 15.72   \u001b[0m | \u001b[0m 479.7   \u001b[0m | \u001b[0m 157.1   \u001b[0m | \u001b[0m 3.647e+0\u001b[0m | \u001b[0m 3.042   \u001b[0m | \u001b[0m 3.732   \u001b[0m |\n",
      "| \u001b[0m 91      \u001b[0m | \u001b[0m 0.6975  \u001b[0m | \u001b[0m 0.7193  \u001b[0m | \u001b[0m 0.7731  \u001b[0m | \u001b[0m 0.5798  \u001b[0m | \u001b[0m 0.04349 \u001b[0m | \u001b[0m 11.96   \u001b[0m | \u001b[0m 461.2   \u001b[0m | \u001b[0m 902.0   \u001b[0m | \u001b[0m 9.997e+0\u001b[0m | \u001b[0m 2.191   \u001b[0m | \u001b[0m 2.655   \u001b[0m |\n",
      "| \u001b[0m 92      \u001b[0m | \u001b[0m 0.7022  \u001b[0m | \u001b[0m 0.5745  \u001b[0m | \u001b[0m 0.5796  \u001b[0m | \u001b[0m 0.7148  \u001b[0m | \u001b[0m 0.04535 \u001b[0m | \u001b[0m 17.35   \u001b[0m | \u001b[0m 142.1   \u001b[0m | \u001b[0m 1.173e+0\u001b[0m | \u001b[0m 35.07   \u001b[0m | \u001b[0m 4.998   \u001b[0m | \u001b[0m 0.2462  \u001b[0m |\n",
      "| \u001b[0m 93      \u001b[0m | \u001b[0m 0.7093  \u001b[0m | \u001b[0m 0.7542  \u001b[0m | \u001b[0m 0.7432  \u001b[0m | \u001b[0m 0.6571  \u001b[0m | \u001b[0m 0.00685 \u001b[0m | \u001b[0m 5.787   \u001b[0m | \u001b[0m 13.84   \u001b[0m | \u001b[0m 526.3   \u001b[0m | \u001b[0m 1.852e+0\u001b[0m | \u001b[0m 4.593   \u001b[0m | \u001b[0m 3.408   \u001b[0m |\n",
      "| \u001b[0m 94      \u001b[0m | \u001b[0m 0.6383  \u001b[0m | \u001b[0m 0.5754  \u001b[0m | \u001b[0m 0.5873  \u001b[0m | \u001b[0m 0.5086  \u001b[0m | \u001b[0m 0.03622 \u001b[0m | \u001b[0m 7.691   \u001b[0m | \u001b[0m 19.34   \u001b[0m | \u001b[0m 120.7   \u001b[0m | \u001b[0m 7.469e+0\u001b[0m | \u001b[0m 2.709   \u001b[0m | \u001b[0m 1.679   \u001b[0m |\n",
      "| \u001b[0m 95      \u001b[0m | \u001b[0m 0.6703  \u001b[0m | \u001b[0m 0.5507  \u001b[0m | \u001b[0m 0.5788  \u001b[0m | \u001b[0m 0.5605  \u001b[0m | \u001b[0m 0.03152 \u001b[0m | \u001b[0m 4.954   \u001b[0m | \u001b[0m 43.65   \u001b[0m | \u001b[0m 100.5   \u001b[0m | \u001b[0m 8.573e+0\u001b[0m | \u001b[0m 3.862   \u001b[0m | \u001b[0m 1.95    \u001b[0m |\n",
      "| \u001b[0m 96      \u001b[0m | \u001b[0m 0.6983  \u001b[0m | \u001b[0m 0.6009  \u001b[0m | \u001b[0m 0.7963  \u001b[0m | \u001b[0m 0.6506  \u001b[0m | \u001b[0m 0.01272 \u001b[0m | \u001b[0m 4.479   \u001b[0m | \u001b[0m 476.9   \u001b[0m | \u001b[0m 127.7   \u001b[0m | \u001b[0m 8.179e+0\u001b[0m | \u001b[0m 4.978   \u001b[0m | \u001b[0m 0.5906  \u001b[0m |\n",
      "| \u001b[0m 97      \u001b[0m | \u001b[0m 0.6954  \u001b[0m | \u001b[0m 0.6036  \u001b[0m | \u001b[0m 0.6433  \u001b[0m | \u001b[0m 0.6722  \u001b[0m | \u001b[0m 0.03426 \u001b[0m | \u001b[0m 18.25   \u001b[0m | \u001b[0m 9.214   \u001b[0m | \u001b[0m 145.6   \u001b[0m | \u001b[0m 6.48e+03\u001b[0m | \u001b[0m 1.871   \u001b[0m | \u001b[0m 2.738   \u001b[0m |\n",
      "| \u001b[0m 98      \u001b[0m | \u001b[0m 0.6921  \u001b[0m | \u001b[0m 0.6254  \u001b[0m | \u001b[0m 0.618   \u001b[0m | \u001b[0m 0.6745  \u001b[0m | \u001b[0m 0.02188 \u001b[0m | \u001b[0m 18.71   \u001b[0m | \u001b[0m 474.3   \u001b[0m | \u001b[0m 873.8   \u001b[0m | \u001b[0m 7.723e+0\u001b[0m | \u001b[0m 4.049   \u001b[0m | \u001b[0m 1.563   \u001b[0m |\n",
      "| \u001b[0m 99      \u001b[0m | \u001b[0m 0.7009  \u001b[0m | \u001b[0m 0.6443  \u001b[0m | \u001b[0m 0.7915  \u001b[0m | \u001b[0m 0.5605  \u001b[0m | \u001b[0m 0.03356 \u001b[0m | \u001b[0m 18.55   \u001b[0m | \u001b[0m 11.13   \u001b[0m | \u001b[0m 1.032e+0\u001b[0m | \u001b[0m 9.614e+0\u001b[0m | \u001b[0m 4.573   \u001b[0m | \u001b[0m 4.408   \u001b[0m |\n",
      "| \u001b[0m 100     \u001b[0m | \u001b[0m 0.7057  \u001b[0m | \u001b[0m 0.7865  \u001b[0m | \u001b[0m 0.6498  \u001b[0m | \u001b[0m 0.6625  \u001b[0m | \u001b[0m 0.02398 \u001b[0m | \u001b[0m 7.903   \u001b[0m | \u001b[0m 13.6    \u001b[0m | \u001b[0m 852.5   \u001b[0m | \u001b[0m 9.972e+0\u001b[0m | \u001b[0m 1.769   \u001b[0m | \u001b[0m 4.744   \u001b[0m |\n",
      "=================================================================================================================================================\n"
     ]
    }
   ],
   "source": [
    "optimizer.maximize(n_iter=75, init_points=25, acq='ucb')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "|   iter    |  target   | colsam... | colsam... | colsam... | learni... | max_depth | min_ch... | n_esti... | random... | reg_alpha | reg_la... |\n",
      "-------------------------------------------------------------------------------------------------------------------------------------------------\n",
      "| \u001b[0m 101     \u001b[0m | \u001b[0m 0.685   \u001b[0m | \u001b[0m 0.5001  \u001b[0m | \u001b[0m 0.793   \u001b[0m | \u001b[0m 0.613   \u001b[0m | \u001b[0m 0.04882 \u001b[0m | \u001b[0m 12.68   \u001b[0m | \u001b[0m 416.0   \u001b[0m | \u001b[0m 904.6   \u001b[0m | \u001b[0m 6.281e+0\u001b[0m | \u001b[0m 1.428   \u001b[0m | \u001b[0m 2.934   \u001b[0m |\n",
      "| \u001b[0m 102     \u001b[0m | \u001b[0m 0.7163  \u001b[0m | \u001b[0m 0.725   \u001b[0m | \u001b[0m 0.7575  \u001b[0m | \u001b[0m 0.7265  \u001b[0m | \u001b[0m 0.03641 \u001b[0m | \u001b[0m 16.83   \u001b[0m | \u001b[0m 167.4   \u001b[0m | \u001b[0m 1.039e+0\u001b[0m | \u001b[0m 4.509e+0\u001b[0m | \u001b[0m 1.911   \u001b[0m | \u001b[0m 2.054   \u001b[0m |\n",
      "| \u001b[0m 103     \u001b[0m | \u001b[0m 0.7095  \u001b[0m | \u001b[0m 0.6204  \u001b[0m | \u001b[0m 0.5952  \u001b[0m | \u001b[0m 0.6866  \u001b[0m | \u001b[0m 0.02436 \u001b[0m | \u001b[0m 18.58   \u001b[0m | \u001b[0m 341.8   \u001b[0m | \u001b[0m 378.0   \u001b[0m | \u001b[0m 4.267e+0\u001b[0m | \u001b[0m 1.717   \u001b[0m | \u001b[0m 3.988   \u001b[0m |\n",
      "| \u001b[0m 104     \u001b[0m | \u001b[0m 0.7114  \u001b[0m | \u001b[0m 0.764   \u001b[0m | \u001b[0m 0.7712  \u001b[0m | \u001b[0m 0.6988  \u001b[0m | \u001b[0m 0.01716 \u001b[0m | \u001b[0m 7.038   \u001b[0m | \u001b[0m 428.8   \u001b[0m | \u001b[0m 838.8   \u001b[0m | \u001b[0m 8.022e+0\u001b[0m | \u001b[0m 2.862   \u001b[0m | \u001b[0m 3.666   \u001b[0m |\n",
      "| \u001b[0m 105     \u001b[0m | \u001b[0m 0.6971  \u001b[0m | \u001b[0m 0.6557  \u001b[0m | \u001b[0m 0.7313  \u001b[0m | \u001b[0m 0.6707  \u001b[0m | \u001b[0m 0.02596 \u001b[0m | \u001b[0m 8.483   \u001b[0m | \u001b[0m 42.49   \u001b[0m | \u001b[0m 629.1   \u001b[0m | \u001b[0m 796.3   \u001b[0m | \u001b[0m 4.914   \u001b[0m | \u001b[0m 0.9081  \u001b[0m |\n",
      "| \u001b[0m 106     \u001b[0m | \u001b[0m 0.7142  \u001b[0m | \u001b[0m 0.6788  \u001b[0m | \u001b[0m 0.7957  \u001b[0m | \u001b[0m 0.743   \u001b[0m | \u001b[0m 0.04772 \u001b[0m | \u001b[0m 3.949   \u001b[0m | \u001b[0m 23.8    \u001b[0m | \u001b[0m 114.8   \u001b[0m | \u001b[0m 8.049e+0\u001b[0m | \u001b[0m 3.453   \u001b[0m | \u001b[0m 4.602   \u001b[0m |\n",
      "| \u001b[0m 107     \u001b[0m | \u001b[0m 0.6811  \u001b[0m | \u001b[0m 0.5194  \u001b[0m | \u001b[0m 0.7122  \u001b[0m | \u001b[0m 0.5898  \u001b[0m | \u001b[0m 0.02977 \u001b[0m | \u001b[0m 14.66   \u001b[0m | \u001b[0m 497.9   \u001b[0m | \u001b[0m 1.236e+0\u001b[0m | \u001b[0m 8.249e+0\u001b[0m | \u001b[0m 1.352   \u001b[0m | \u001b[0m 3.664   \u001b[0m |\n",
      "| \u001b[0m 108     \u001b[0m | \u001b[0m 0.7035  \u001b[0m | \u001b[0m 0.6857  \u001b[0m | \u001b[0m 0.6778  \u001b[0m | \u001b[0m 0.6668  \u001b[0m | \u001b[0m 0.04541 \u001b[0m | \u001b[0m 16.4    \u001b[0m | \u001b[0m 445.6   \u001b[0m | \u001b[0m 847.6   \u001b[0m | \u001b[0m 3.764e+0\u001b[0m | \u001b[0m 4.799   \u001b[0m | \u001b[0m 4.589   \u001b[0m |\n",
      "| \u001b[0m 109     \u001b[0m | \u001b[0m 0.7108  \u001b[0m | \u001b[0m 0.6709  \u001b[0m | \u001b[0m 0.7887  \u001b[0m | \u001b[0m 0.5886  \u001b[0m | \u001b[0m 0.02419 \u001b[0m | \u001b[0m 11.39   \u001b[0m | \u001b[0m 423.9   \u001b[0m | \u001b[0m 1.436e+0\u001b[0m | \u001b[0m 2.829e+0\u001b[0m | \u001b[0m 3.088   \u001b[0m | \u001b[0m 1.851   \u001b[0m |\n",
      "| \u001b[0m 110     \u001b[0m | \u001b[0m 0.6559  \u001b[0m | \u001b[0m 0.5003  \u001b[0m | \u001b[0m 0.7989  \u001b[0m | \u001b[0m 0.6769  \u001b[0m | \u001b[0m 0.04969 \u001b[0m | \u001b[0m 16.96   \u001b[0m | \u001b[0m 331.1   \u001b[0m | \u001b[0m 719.7   \u001b[0m | \u001b[0m 3.325e+0\u001b[0m | \u001b[0m 4.617   \u001b[0m | \u001b[0m 3.601   \u001b[0m |\n",
      "| \u001b[0m 111     \u001b[0m | \u001b[0m 0.6973  \u001b[0m | \u001b[0m 0.7985  \u001b[0m | \u001b[0m 0.6486  \u001b[0m | \u001b[0m 0.6966  \u001b[0m | \u001b[0m 0.02871 \u001b[0m | \u001b[0m 18.51   \u001b[0m | \u001b[0m 147.8   \u001b[0m | \u001b[0m 1.183e+0\u001b[0m | \u001b[0m 39.24   \u001b[0m | \u001b[0m 3.466   \u001b[0m | \u001b[0m 0.3269  \u001b[0m |\n",
      "| \u001b[0m 112     \u001b[0m | \u001b[0m 0.7071  \u001b[0m | \u001b[0m 0.7805  \u001b[0m | \u001b[0m 0.6695  \u001b[0m | \u001b[0m 0.5511  \u001b[0m | \u001b[0m 0.01343 \u001b[0m | \u001b[0m 4.909   \u001b[0m | \u001b[0m 16.54   \u001b[0m | \u001b[0m 925.7   \u001b[0m | \u001b[0m 3.235e+0\u001b[0m | \u001b[0m 3.228   \u001b[0m | \u001b[0m 4.908   \u001b[0m |\n",
      "| \u001b[0m 113     \u001b[0m | \u001b[0m 0.7145  \u001b[0m | \u001b[0m 0.7539  \u001b[0m | \u001b[0m 0.7945  \u001b[0m | \u001b[0m 0.6649  \u001b[0m | \u001b[0m 0.01149 \u001b[0m | \u001b[0m 4.074   \u001b[0m | \u001b[0m 497.5   \u001b[0m | \u001b[0m 689.3   \u001b[0m | \u001b[0m 2.26e+03\u001b[0m | \u001b[0m 4.896   \u001b[0m | \u001b[0m 1.188   \u001b[0m |\n",
      "| \u001b[0m 114     \u001b[0m | \u001b[0m 0.695   \u001b[0m | \u001b[0m 0.5165  \u001b[0m | \u001b[0m 0.7136  \u001b[0m | \u001b[0m 0.5567  \u001b[0m | \u001b[0m 0.03718 \u001b[0m | \u001b[0m 9.984   \u001b[0m | \u001b[0m 14.68   \u001b[0m | \u001b[0m 965.8   \u001b[0m | \u001b[0m 3.443e+0\u001b[0m | \u001b[0m 2.523   \u001b[0m | \u001b[0m 3.993   \u001b[0m |\n",
      "| \u001b[0m 115     \u001b[0m | \u001b[0m 0.7156  \u001b[0m | \u001b[0m 0.7217  \u001b[0m | \u001b[0m 0.7397  \u001b[0m | \u001b[0m 0.7131  \u001b[0m | \u001b[0m 0.01093 \u001b[0m | \u001b[0m 6.417   \u001b[0m | \u001b[0m 330.4   \u001b[0m | \u001b[0m 1.493e+0\u001b[0m | \u001b[0m 3.05e+03\u001b[0m | \u001b[0m 4.873   \u001b[0m | \u001b[0m 1.191   \u001b[0m |\n",
      "| \u001b[0m 116     \u001b[0m | \u001b[0m 0.7151  \u001b[0m | \u001b[0m 0.7999  \u001b[0m | \u001b[0m 0.6384  \u001b[0m | \u001b[0m 0.6995  \u001b[0m | \u001b[0m 0.03545 \u001b[0m | \u001b[0m 4.293   \u001b[0m | \u001b[0m 163.7   \u001b[0m | \u001b[0m 714.6   \u001b[0m | \u001b[0m 2.373e+0\u001b[0m | \u001b[0m 4.971   \u001b[0m | \u001b[0m 0.5167  \u001b[0m |\n",
      "| \u001b[0m 117     \u001b[0m | \u001b[0m 0.6726  \u001b[0m | \u001b[0m 0.6049  \u001b[0m | \u001b[0m 0.6326  \u001b[0m | \u001b[0m 0.5159  \u001b[0m | \u001b[0m 0.03857 \u001b[0m | \u001b[0m 16.91   \u001b[0m | \u001b[0m 492.7   \u001b[0m | \u001b[0m 1.134e+0\u001b[0m | \u001b[0m 3.354e+0\u001b[0m | \u001b[0m 4.095   \u001b[0m | \u001b[0m 1.672   \u001b[0m |\n",
      "| \u001b[0m 118     \u001b[0m | \u001b[0m 0.6995  \u001b[0m | \u001b[0m 0.6984  \u001b[0m | \u001b[0m 0.5551  \u001b[0m | \u001b[0m 0.7988  \u001b[0m | \u001b[0m 0.02001 \u001b[0m | \u001b[0m 18.9    \u001b[0m | \u001b[0m 499.7   \u001b[0m | \u001b[0m 790.6   \u001b[0m | \u001b[0m 4.261e+0\u001b[0m | \u001b[0m 2.554   \u001b[0m | \u001b[0m 1.229   \u001b[0m |\n",
      "| \u001b[0m 119     \u001b[0m | \u001b[0m 0.7054  \u001b[0m | \u001b[0m 0.7418  \u001b[0m | \u001b[0m 0.5556  \u001b[0m | \u001b[0m 0.6642  \u001b[0m | \u001b[0m 0.02258 \u001b[0m | \u001b[0m 5.625   \u001b[0m | \u001b[0m 483.0   \u001b[0m | \u001b[0m 1.495e+0\u001b[0m | \u001b[0m 1.025e+0\u001b[0m | \u001b[0m 1.148   \u001b[0m | \u001b[0m 2.829   \u001b[0m |\n",
      "| \u001b[0m 120     \u001b[0m | \u001b[0m 0.701   \u001b[0m | \u001b[0m 0.7936  \u001b[0m | \u001b[0m 0.7338  \u001b[0m | \u001b[0m 0.5905  \u001b[0m | \u001b[0m 0.01119 \u001b[0m | \u001b[0m 6.007   \u001b[0m | \u001b[0m 35.19   \u001b[0m | \u001b[0m 754.9   \u001b[0m | \u001b[0m 7.945e+0\u001b[0m | \u001b[0m 2.593   \u001b[0m | \u001b[0m 3.445   \u001b[0m |\n",
      "| \u001b[0m 121     \u001b[0m | \u001b[0m 0.7079  \u001b[0m | \u001b[0m 0.5892  \u001b[0m | \u001b[0m 0.5586  \u001b[0m | \u001b[0m 0.7279  \u001b[0m | \u001b[0m 0.04016 \u001b[0m | \u001b[0m 4.147   \u001b[0m | \u001b[0m 481.5   \u001b[0m | \u001b[0m 546.0   \u001b[0m | \u001b[0m 9.281e+0\u001b[0m | \u001b[0m 0.7084  \u001b[0m | \u001b[0m 2.509   \u001b[0m |\n",
      "| \u001b[0m 122     \u001b[0m | \u001b[0m 0.657   \u001b[0m | \u001b[0m 0.5056  \u001b[0m | \u001b[0m 0.7857  \u001b[0m | \u001b[0m 0.6035  \u001b[0m | \u001b[0m 0.03152 \u001b[0m | \u001b[0m 7.749   \u001b[0m | \u001b[0m 270.9   \u001b[0m | \u001b[0m 915.7   \u001b[0m | \u001b[0m 5.246e+0\u001b[0m | \u001b[0m 4.106   \u001b[0m | \u001b[0m 2.997   \u001b[0m |\n",
      "| \u001b[0m 123     \u001b[0m | \u001b[0m 0.7036  \u001b[0m | \u001b[0m 0.7681  \u001b[0m | \u001b[0m 0.5447  \u001b[0m | \u001b[0m 0.7458  \u001b[0m | \u001b[0m 0.04393 \u001b[0m | \u001b[0m 10.26   \u001b[0m | \u001b[0m 154.8   \u001b[0m | \u001b[0m 1.261e+0\u001b[0m | \u001b[0m 5.173e+0\u001b[0m | \u001b[0m 4.022   \u001b[0m | \u001b[0m 4.429   \u001b[0m |\n",
      "| \u001b[0m 124     \u001b[0m | \u001b[0m 0.7111  \u001b[0m | \u001b[0m 0.7013  \u001b[0m | \u001b[0m 0.6555  \u001b[0m | \u001b[0m 0.7934  \u001b[0m | \u001b[0m 0.03603 \u001b[0m | \u001b[0m 6.044   \u001b[0m | \u001b[0m 414.4   \u001b[0m | \u001b[0m 1.152e+0\u001b[0m | \u001b[0m 4.939e+0\u001b[0m | \u001b[0m 3.85    \u001b[0m | \u001b[0m 1.228   \u001b[0m |\n",
      "| \u001b[0m 125     \u001b[0m | \u001b[0m 0.7048  \u001b[0m | \u001b[0m 0.6152  \u001b[0m | \u001b[0m 0.6222  \u001b[0m | \u001b[0m 0.6995  \u001b[0m | \u001b[0m 0.01454 \u001b[0m | \u001b[0m 6.222   \u001b[0m | \u001b[0m 394.4   \u001b[0m | \u001b[0m 103.1   \u001b[0m | \u001b[0m 4.57e+03\u001b[0m | \u001b[0m 4.584   \u001b[0m | \u001b[0m 4.503   \u001b[0m |\n",
      "=================================================================================================================================================\n"
     ]
    }
   ],
   "source": [
    "optimizer.maximize(n_iter=20, init_points=5, acq='ei')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.7196878687075241,\n",
       " {'colsample_bylevel': 0.7023288136056443,\n",
       "  'colsample_bynode': 0.7936743595485696,\n",
       "  'colsample_bytree': 0.789571791734966,\n",
       "  'learning_rate': 0.04091766991506028,\n",
       "  'max_depth': 3.1267246830096624,\n",
       "  'min_child_weight': 18.807595456643067,\n",
       "  'n_estimators': 1492.0803424752537,\n",
       "  'random_state': 87.62020925109914,\n",
       "  'reg_alpha': 4.198821312556088,\n",
       "  'reg_lambda': 0.6668707851598255})"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "idx_max = np.argmax(pd.DataFrame(optimizer.res).target)\n",
    "pd.DataFrame(optimizer.res).target[idx_max],pd.DataFrame(optimizer.res).params[idx_max]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'colsample_bylevel': 0.7023288136056443, 'colsample_bynode': 0.7936743595485696, 'colsample_bytree': 0.789571791734966, 'learning_rate': 0.04091766991506028, 'max_depth': 3.1267246830096624, 'min_child_weight': 18.807595456643067, 'n_estimators': 1492.0803424752537, 'random_state': 87.62020925109914, 'reg_alpha': 4.198821312556088, 'reg_lambda': 0.6668707851598255}\n"
     ]
    }
   ],
   "source": [
    "params = optimizer.max['params']\n",
    "print(params)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "params['n_estimators'] = int(params['n_estimators'])\n",
    "params['max_depth'] = int(params['max_depth'])\n",
    "params['objective'] = 'binary:logistic'\n",
    "params['booster'] = 'gbtree'\n",
    "params['subsample'] = 0.7\n",
    "params['random_state'] = int(params['random_state'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0]\tvalidation_0-auc:0.80863\tvalidation_1-auc:0.62062\n",
      "Multiple eval metrics have been passed: 'validation_1-auc' will be used for early stopping.\n",
      "\n",
      "Will train until validation_1-auc hasn't improved in 100 rounds.\n",
      "[1]\tvalidation_0-auc:0.81448\tvalidation_1-auc:0.55742\n",
      "[2]\tvalidation_0-auc:0.81582\tvalidation_1-auc:0.61768\n",
      "[3]\tvalidation_0-auc:0.81617\tvalidation_1-auc:0.60783\n",
      "[4]\tvalidation_0-auc:0.81859\tvalidation_1-auc:0.57679\n",
      "[5]\tvalidation_0-auc:0.82411\tvalidation_1-auc:0.62534\n",
      "[6]\tvalidation_0-auc:0.82701\tvalidation_1-auc:0.65878\n",
      "[7]\tvalidation_0-auc:0.83105\tvalidation_1-auc:0.67018\n",
      "[8]\tvalidation_0-auc:0.82996\tvalidation_1-auc:0.66969\n",
      "[9]\tvalidation_0-auc:0.83288\tvalidation_1-auc:0.66697\n",
      "[10]\tvalidation_0-auc:0.83391\tvalidation_1-auc:0.68462\n",
      "[11]\tvalidation_0-auc:0.83496\tvalidation_1-auc:0.70215\n",
      "[12]\tvalidation_0-auc:0.83439\tvalidation_1-auc:0.69822\n",
      "[13]\tvalidation_0-auc:0.83219\tvalidation_1-auc:0.68469\n",
      "[14]\tvalidation_0-auc:0.83370\tvalidation_1-auc:0.68473\n",
      "[15]\tvalidation_0-auc:0.83408\tvalidation_1-auc:0.69015\n",
      "[16]\tvalidation_0-auc:0.83495\tvalidation_1-auc:0.70612\n",
      "[17]\tvalidation_0-auc:0.83535\tvalidation_1-auc:0.69714\n",
      "[18]\tvalidation_0-auc:0.83682\tvalidation_1-auc:0.69808\n",
      "[19]\tvalidation_0-auc:0.83716\tvalidation_1-auc:0.69699\n",
      "[20]\tvalidation_0-auc:0.83720\tvalidation_1-auc:0.69540\n",
      "[21]\tvalidation_0-auc:0.83722\tvalidation_1-auc:0.69499\n",
      "[22]\tvalidation_0-auc:0.83740\tvalidation_1-auc:0.69619\n",
      "[23]\tvalidation_0-auc:0.83812\tvalidation_1-auc:0.69788\n",
      "[24]\tvalidation_0-auc:0.83877\tvalidation_1-auc:0.70013\n",
      "[25]\tvalidation_0-auc:0.83889\tvalidation_1-auc:0.69841\n",
      "[26]\tvalidation_0-auc:0.83909\tvalidation_1-auc:0.69773\n",
      "[27]\tvalidation_0-auc:0.83965\tvalidation_1-auc:0.69950\n",
      "[28]\tvalidation_0-auc:0.83966\tvalidation_1-auc:0.69939\n",
      "[29]\tvalidation_0-auc:0.83983\tvalidation_1-auc:0.70091\n",
      "[30]\tvalidation_0-auc:0.84035\tvalidation_1-auc:0.70159\n",
      "[31]\tvalidation_0-auc:0.84057\tvalidation_1-auc:0.70218\n",
      "[32]\tvalidation_0-auc:0.84153\tvalidation_1-auc:0.70320\n",
      "[33]\tvalidation_0-auc:0.84163\tvalidation_1-auc:0.70323\n",
      "[34]\tvalidation_0-auc:0.84190\tvalidation_1-auc:0.70450\n",
      "[35]\tvalidation_0-auc:0.84255\tvalidation_1-auc:0.70646\n",
      "[36]\tvalidation_0-auc:0.84288\tvalidation_1-auc:0.70683\n",
      "[37]\tvalidation_0-auc:0.84294\tvalidation_1-auc:0.70754\n",
      "[38]\tvalidation_0-auc:0.84308\tvalidation_1-auc:0.70745\n",
      "[39]\tvalidation_0-auc:0.84324\tvalidation_1-auc:0.70829\n",
      "[40]\tvalidation_0-auc:0.84421\tvalidation_1-auc:0.70870\n",
      "[41]\tvalidation_0-auc:0.84462\tvalidation_1-auc:0.70942\n",
      "[42]\tvalidation_0-auc:0.84491\tvalidation_1-auc:0.70775\n",
      "[43]\tvalidation_0-auc:0.84517\tvalidation_1-auc:0.70951\n",
      "[44]\tvalidation_0-auc:0.84515\tvalidation_1-auc:0.70996\n",
      "[45]\tvalidation_0-auc:0.84565\tvalidation_1-auc:0.71025\n",
      "[46]\tvalidation_0-auc:0.84591\tvalidation_1-auc:0.70935\n",
      "[47]\tvalidation_0-auc:0.84606\tvalidation_1-auc:0.70892\n",
      "[48]\tvalidation_0-auc:0.84609\tvalidation_1-auc:0.71003\n",
      "[49]\tvalidation_0-auc:0.84630\tvalidation_1-auc:0.71099\n",
      "[50]\tvalidation_0-auc:0.84638\tvalidation_1-auc:0.71080\n",
      "[51]\tvalidation_0-auc:0.84663\tvalidation_1-auc:0.71041\n",
      "[52]\tvalidation_0-auc:0.84766\tvalidation_1-auc:0.71096\n",
      "[53]\tvalidation_0-auc:0.84825\tvalidation_1-auc:0.71145\n",
      "[54]\tvalidation_0-auc:0.84883\tvalidation_1-auc:0.71208\n",
      "[55]\tvalidation_0-auc:0.84908\tvalidation_1-auc:0.71235\n",
      "[56]\tvalidation_0-auc:0.84914\tvalidation_1-auc:0.71247\n",
      "[57]\tvalidation_0-auc:0.84980\tvalidation_1-auc:0.71343\n",
      "[58]\tvalidation_0-auc:0.85003\tvalidation_1-auc:0.71360\n",
      "[59]\tvalidation_0-auc:0.85014\tvalidation_1-auc:0.71382\n",
      "[60]\tvalidation_0-auc:0.85041\tvalidation_1-auc:0.71440\n",
      "[61]\tvalidation_0-auc:0.85088\tvalidation_1-auc:0.71352\n",
      "[62]\tvalidation_0-auc:0.85110\tvalidation_1-auc:0.71400\n",
      "[63]\tvalidation_0-auc:0.85154\tvalidation_1-auc:0.71375\n",
      "[64]\tvalidation_0-auc:0.85149\tvalidation_1-auc:0.71337\n",
      "[65]\tvalidation_0-auc:0.85207\tvalidation_1-auc:0.71325\n",
      "[66]\tvalidation_0-auc:0.85224\tvalidation_1-auc:0.71348\n",
      "[67]\tvalidation_0-auc:0.85246\tvalidation_1-auc:0.71366\n",
      "[68]\tvalidation_0-auc:0.85250\tvalidation_1-auc:0.71292\n",
      "[69]\tvalidation_0-auc:0.85267\tvalidation_1-auc:0.71318\n",
      "[70]\tvalidation_0-auc:0.85293\tvalidation_1-auc:0.71360\n",
      "[71]\tvalidation_0-auc:0.85305\tvalidation_1-auc:0.71387\n",
      "[72]\tvalidation_0-auc:0.85362\tvalidation_1-auc:0.71409\n",
      "[73]\tvalidation_0-auc:0.85398\tvalidation_1-auc:0.71446\n",
      "[74]\tvalidation_0-auc:0.85407\tvalidation_1-auc:0.71444\n",
      "[75]\tvalidation_0-auc:0.85443\tvalidation_1-auc:0.71470\n",
      "[76]\tvalidation_0-auc:0.85477\tvalidation_1-auc:0.71464\n",
      "[77]\tvalidation_0-auc:0.85479\tvalidation_1-auc:0.71513\n",
      "[78]\tvalidation_0-auc:0.85503\tvalidation_1-auc:0.71536\n",
      "[79]\tvalidation_0-auc:0.85537\tvalidation_1-auc:0.71545\n",
      "[80]\tvalidation_0-auc:0.85555\tvalidation_1-auc:0.71564\n",
      "[81]\tvalidation_0-auc:0.85575\tvalidation_1-auc:0.71580\n",
      "[82]\tvalidation_0-auc:0.85575\tvalidation_1-auc:0.71589\n",
      "[83]\tvalidation_0-auc:0.85578\tvalidation_1-auc:0.71584\n",
      "[84]\tvalidation_0-auc:0.85595\tvalidation_1-auc:0.71585\n",
      "[85]\tvalidation_0-auc:0.85610\tvalidation_1-auc:0.71611\n",
      "[86]\tvalidation_0-auc:0.85628\tvalidation_1-auc:0.71595\n",
      "[87]\tvalidation_0-auc:0.85652\tvalidation_1-auc:0.71632\n",
      "[88]\tvalidation_0-auc:0.85653\tvalidation_1-auc:0.71641\n",
      "[89]\tvalidation_0-auc:0.85681\tvalidation_1-auc:0.71640\n",
      "[90]\tvalidation_0-auc:0.85692\tvalidation_1-auc:0.71625\n",
      "[91]\tvalidation_0-auc:0.85718\tvalidation_1-auc:0.71649\n",
      "[92]\tvalidation_0-auc:0.85742\tvalidation_1-auc:0.71677\n",
      "[93]\tvalidation_0-auc:0.85747\tvalidation_1-auc:0.71674\n",
      "[94]\tvalidation_0-auc:0.85765\tvalidation_1-auc:0.71683\n",
      "[95]\tvalidation_0-auc:0.85774\tvalidation_1-auc:0.71690\n",
      "[96]\tvalidation_0-auc:0.85789\tvalidation_1-auc:0.71702\n",
      "[97]\tvalidation_0-auc:0.85821\tvalidation_1-auc:0.71715\n",
      "[98]\tvalidation_0-auc:0.85848\tvalidation_1-auc:0.71683\n",
      "[99]\tvalidation_0-auc:0.85879\tvalidation_1-auc:0.71687\n",
      "[100]\tvalidation_0-auc:0.85889\tvalidation_1-auc:0.71672\n",
      "[101]\tvalidation_0-auc:0.85911\tvalidation_1-auc:0.71675\n",
      "[102]\tvalidation_0-auc:0.85930\tvalidation_1-auc:0.71648\n",
      "[103]\tvalidation_0-auc:0.85945\tvalidation_1-auc:0.71647\n",
      "[104]\tvalidation_0-auc:0.85960\tvalidation_1-auc:0.71656\n",
      "[105]\tvalidation_0-auc:0.85968\tvalidation_1-auc:0.71665\n",
      "[106]\tvalidation_0-auc:0.85991\tvalidation_1-auc:0.71673\n",
      "[107]\tvalidation_0-auc:0.86009\tvalidation_1-auc:0.71673\n",
      "[108]\tvalidation_0-auc:0.86027\tvalidation_1-auc:0.71705\n",
      "[109]\tvalidation_0-auc:0.86045\tvalidation_1-auc:0.71693\n",
      "[110]\tvalidation_0-auc:0.86068\tvalidation_1-auc:0.71693\n",
      "[111]\tvalidation_0-auc:0.86080\tvalidation_1-auc:0.71702\n",
      "[112]\tvalidation_0-auc:0.86083\tvalidation_1-auc:0.71698\n",
      "[113]\tvalidation_0-auc:0.86091\tvalidation_1-auc:0.71716\n",
      "[114]\tvalidation_0-auc:0.86105\tvalidation_1-auc:0.71717\n",
      "[115]\tvalidation_0-auc:0.86128\tvalidation_1-auc:0.71705\n",
      "[116]\tvalidation_0-auc:0.86137\tvalidation_1-auc:0.71716\n",
      "[117]\tvalidation_0-auc:0.86153\tvalidation_1-auc:0.71702\n",
      "[118]\tvalidation_0-auc:0.86168\tvalidation_1-auc:0.71707\n",
      "[119]\tvalidation_0-auc:0.86181\tvalidation_1-auc:0.71713\n",
      "[120]\tvalidation_0-auc:0.86200\tvalidation_1-auc:0.71737\n",
      "[121]\tvalidation_0-auc:0.86216\tvalidation_1-auc:0.71751\n",
      "[122]\tvalidation_0-auc:0.86230\tvalidation_1-auc:0.71747\n",
      "[123]\tvalidation_0-auc:0.86236\tvalidation_1-auc:0.71756\n",
      "[124]\tvalidation_0-auc:0.86249\tvalidation_1-auc:0.71765\n",
      "[125]\tvalidation_0-auc:0.86258\tvalidation_1-auc:0.71764\n",
      "[126]\tvalidation_0-auc:0.86263\tvalidation_1-auc:0.71764\n",
      "[127]\tvalidation_0-auc:0.86272\tvalidation_1-auc:0.71776\n",
      "[128]\tvalidation_0-auc:0.86286\tvalidation_1-auc:0.71781\n",
      "[129]\tvalidation_0-auc:0.86293\tvalidation_1-auc:0.71778\n",
      "[130]\tvalidation_0-auc:0.86301\tvalidation_1-auc:0.71779\n",
      "[131]\tvalidation_0-auc:0.86306\tvalidation_1-auc:0.71787\n",
      "[132]\tvalidation_0-auc:0.86311\tvalidation_1-auc:0.71794\n",
      "[133]\tvalidation_0-auc:0.86324\tvalidation_1-auc:0.71806\n",
      "[134]\tvalidation_0-auc:0.86332\tvalidation_1-auc:0.71801\n",
      "[135]\tvalidation_0-auc:0.86339\tvalidation_1-auc:0.71799\n",
      "[136]\tvalidation_0-auc:0.86347\tvalidation_1-auc:0.71781\n",
      "[137]\tvalidation_0-auc:0.86365\tvalidation_1-auc:0.71767\n",
      "[138]\tvalidation_0-auc:0.86374\tvalidation_1-auc:0.71770\n",
      "[139]\tvalidation_0-auc:0.86382\tvalidation_1-auc:0.71772\n",
      "[140]\tvalidation_0-auc:0.86389\tvalidation_1-auc:0.71768\n",
      "[141]\tvalidation_0-auc:0.86413\tvalidation_1-auc:0.71759\n",
      "[142]\tvalidation_0-auc:0.86422\tvalidation_1-auc:0.71759\n",
      "[143]\tvalidation_0-auc:0.86424\tvalidation_1-auc:0.71767\n",
      "[144]\tvalidation_0-auc:0.86427\tvalidation_1-auc:0.71773\n",
      "[145]\tvalidation_0-auc:0.86438\tvalidation_1-auc:0.71765\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[146]\tvalidation_0-auc:0.86448\tvalidation_1-auc:0.71775\n",
      "[147]\tvalidation_0-auc:0.86456\tvalidation_1-auc:0.71784\n",
      "[148]\tvalidation_0-auc:0.86465\tvalidation_1-auc:0.71785\n",
      "[149]\tvalidation_0-auc:0.86471\tvalidation_1-auc:0.71772\n",
      "[150]\tvalidation_0-auc:0.86475\tvalidation_1-auc:0.71780\n",
      "[151]\tvalidation_0-auc:0.86483\tvalidation_1-auc:0.71788\n",
      "[152]\tvalidation_0-auc:0.86490\tvalidation_1-auc:0.71773\n",
      "[153]\tvalidation_0-auc:0.86497\tvalidation_1-auc:0.71778\n",
      "[154]\tvalidation_0-auc:0.86506\tvalidation_1-auc:0.71787\n",
      "[155]\tvalidation_0-auc:0.86519\tvalidation_1-auc:0.71782\n",
      "[156]\tvalidation_0-auc:0.86524\tvalidation_1-auc:0.71785\n",
      "[157]\tvalidation_0-auc:0.86529\tvalidation_1-auc:0.71771\n",
      "[158]\tvalidation_0-auc:0.86533\tvalidation_1-auc:0.71779\n",
      "[159]\tvalidation_0-auc:0.86539\tvalidation_1-auc:0.71769\n",
      "[160]\tvalidation_0-auc:0.86554\tvalidation_1-auc:0.71758\n",
      "[161]\tvalidation_0-auc:0.86564\tvalidation_1-auc:0.71764\n",
      "[162]\tvalidation_0-auc:0.86575\tvalidation_1-auc:0.71759\n",
      "[163]\tvalidation_0-auc:0.86584\tvalidation_1-auc:0.71756\n",
      "[164]\tvalidation_0-auc:0.86590\tvalidation_1-auc:0.71756\n",
      "[165]\tvalidation_0-auc:0.86598\tvalidation_1-auc:0.71722\n",
      "[166]\tvalidation_0-auc:0.86604\tvalidation_1-auc:0.71727\n",
      "[167]\tvalidation_0-auc:0.86610\tvalidation_1-auc:0.71739\n",
      "[168]\tvalidation_0-auc:0.86615\tvalidation_1-auc:0.71742\n",
      "[169]\tvalidation_0-auc:0.86623\tvalidation_1-auc:0.71749\n",
      "[170]\tvalidation_0-auc:0.86630\tvalidation_1-auc:0.71758\n",
      "[171]\tvalidation_0-auc:0.86642\tvalidation_1-auc:0.71756\n",
      "[172]\tvalidation_0-auc:0.86646\tvalidation_1-auc:0.71756\n",
      "[173]\tvalidation_0-auc:0.86655\tvalidation_1-auc:0.71765\n",
      "[174]\tvalidation_0-auc:0.86658\tvalidation_1-auc:0.71768\n",
      "[175]\tvalidation_0-auc:0.86668\tvalidation_1-auc:0.71767\n",
      "[176]\tvalidation_0-auc:0.86672\tvalidation_1-auc:0.71774\n",
      "[177]\tvalidation_0-auc:0.86675\tvalidation_1-auc:0.71758\n",
      "[178]\tvalidation_0-auc:0.86681\tvalidation_1-auc:0.71756\n",
      "[179]\tvalidation_0-auc:0.86686\tvalidation_1-auc:0.71766\n",
      "[180]\tvalidation_0-auc:0.86691\tvalidation_1-auc:0.71761\n",
      "[181]\tvalidation_0-auc:0.86700\tvalidation_1-auc:0.71767\n",
      "[182]\tvalidation_0-auc:0.86705\tvalidation_1-auc:0.71767\n",
      "[183]\tvalidation_0-auc:0.86709\tvalidation_1-auc:0.71769\n",
      "[184]\tvalidation_0-auc:0.86711\tvalidation_1-auc:0.71773\n",
      "[185]\tvalidation_0-auc:0.86712\tvalidation_1-auc:0.71773\n",
      "[186]\tvalidation_0-auc:0.86720\tvalidation_1-auc:0.71780\n",
      "[187]\tvalidation_0-auc:0.86724\tvalidation_1-auc:0.71791\n",
      "[188]\tvalidation_0-auc:0.86727\tvalidation_1-auc:0.71797\n",
      "[189]\tvalidation_0-auc:0.86735\tvalidation_1-auc:0.71837\n",
      "[190]\tvalidation_0-auc:0.86741\tvalidation_1-auc:0.71841\n",
      "[191]\tvalidation_0-auc:0.86744\tvalidation_1-auc:0.71834\n",
      "[192]\tvalidation_0-auc:0.86749\tvalidation_1-auc:0.71847\n",
      "[193]\tvalidation_0-auc:0.86755\tvalidation_1-auc:0.71851\n",
      "[194]\tvalidation_0-auc:0.86760\tvalidation_1-auc:0.71855\n",
      "[195]\tvalidation_0-auc:0.86767\tvalidation_1-auc:0.71862\n",
      "[196]\tvalidation_0-auc:0.86772\tvalidation_1-auc:0.71858\n",
      "[197]\tvalidation_0-auc:0.86776\tvalidation_1-auc:0.71866\n",
      "[198]\tvalidation_0-auc:0.86780\tvalidation_1-auc:0.71866\n",
      "[199]\tvalidation_0-auc:0.86787\tvalidation_1-auc:0.71881\n",
      "[200]\tvalidation_0-auc:0.86793\tvalidation_1-auc:0.71883\n",
      "[201]\tvalidation_0-auc:0.86799\tvalidation_1-auc:0.71881\n",
      "[202]\tvalidation_0-auc:0.86804\tvalidation_1-auc:0.71888\n",
      "[203]\tvalidation_0-auc:0.86811\tvalidation_1-auc:0.71894\n",
      "[204]\tvalidation_0-auc:0.86815\tvalidation_1-auc:0.71893\n",
      "[205]\tvalidation_0-auc:0.86817\tvalidation_1-auc:0.71893\n",
      "[206]\tvalidation_0-auc:0.86821\tvalidation_1-auc:0.71899\n",
      "[207]\tvalidation_0-auc:0.86827\tvalidation_1-auc:0.71896\n",
      "[208]\tvalidation_0-auc:0.86830\tvalidation_1-auc:0.71896\n",
      "[209]\tvalidation_0-auc:0.86835\tvalidation_1-auc:0.71899\n",
      "[210]\tvalidation_0-auc:0.86838\tvalidation_1-auc:0.71881\n",
      "[211]\tvalidation_0-auc:0.86844\tvalidation_1-auc:0.71878\n",
      "[212]\tvalidation_0-auc:0.86848\tvalidation_1-auc:0.71857\n",
      "[213]\tvalidation_0-auc:0.86853\tvalidation_1-auc:0.71860\n",
      "[214]\tvalidation_0-auc:0.86859\tvalidation_1-auc:0.71868\n",
      "[215]\tvalidation_0-auc:0.86865\tvalidation_1-auc:0.71869\n",
      "[216]\tvalidation_0-auc:0.86872\tvalidation_1-auc:0.71867\n",
      "[217]\tvalidation_0-auc:0.86876\tvalidation_1-auc:0.71867\n",
      "[218]\tvalidation_0-auc:0.86889\tvalidation_1-auc:0.71857\n",
      "[219]\tvalidation_0-auc:0.86897\tvalidation_1-auc:0.71856\n",
      "[220]\tvalidation_0-auc:0.86901\tvalidation_1-auc:0.71857\n",
      "[221]\tvalidation_0-auc:0.86903\tvalidation_1-auc:0.71857\n",
      "[222]\tvalidation_0-auc:0.86910\tvalidation_1-auc:0.71849\n",
      "[223]\tvalidation_0-auc:0.86912\tvalidation_1-auc:0.71850\n",
      "[224]\tvalidation_0-auc:0.86918\tvalidation_1-auc:0.71847\n",
      "[225]\tvalidation_0-auc:0.86920\tvalidation_1-auc:0.71840\n",
      "[226]\tvalidation_0-auc:0.86931\tvalidation_1-auc:0.71824\n",
      "[227]\tvalidation_0-auc:0.86935\tvalidation_1-auc:0.71829\n",
      "[228]\tvalidation_0-auc:0.86940\tvalidation_1-auc:0.71837\n",
      "[229]\tvalidation_0-auc:0.86947\tvalidation_1-auc:0.71850\n",
      "[230]\tvalidation_0-auc:0.86953\tvalidation_1-auc:0.71852\n",
      "[231]\tvalidation_0-auc:0.86955\tvalidation_1-auc:0.71843\n",
      "[232]\tvalidation_0-auc:0.86961\tvalidation_1-auc:0.71842\n",
      "[233]\tvalidation_0-auc:0.86967\tvalidation_1-auc:0.71843\n",
      "[234]\tvalidation_0-auc:0.86972\tvalidation_1-auc:0.71838\n",
      "[235]\tvalidation_0-auc:0.86977\tvalidation_1-auc:0.71833\n",
      "[236]\tvalidation_0-auc:0.86981\tvalidation_1-auc:0.71826\n",
      "[237]\tvalidation_0-auc:0.86985\tvalidation_1-auc:0.71826\n",
      "[238]\tvalidation_0-auc:0.86986\tvalidation_1-auc:0.71826\n",
      "[239]\tvalidation_0-auc:0.86990\tvalidation_1-auc:0.71837\n",
      "[240]\tvalidation_0-auc:0.86992\tvalidation_1-auc:0.71835\n",
      "[241]\tvalidation_0-auc:0.86994\tvalidation_1-auc:0.71842\n",
      "[242]\tvalidation_0-auc:0.86998\tvalidation_1-auc:0.71850\n",
      "[243]\tvalidation_0-auc:0.87004\tvalidation_1-auc:0.71850\n",
      "[244]\tvalidation_0-auc:0.87012\tvalidation_1-auc:0.71857\n",
      "[245]\tvalidation_0-auc:0.87016\tvalidation_1-auc:0.71857\n",
      "[246]\tvalidation_0-auc:0.87017\tvalidation_1-auc:0.71862\n",
      "[247]\tvalidation_0-auc:0.87021\tvalidation_1-auc:0.71862\n",
      "[248]\tvalidation_0-auc:0.87025\tvalidation_1-auc:0.71858\n",
      "[249]\tvalidation_0-auc:0.87029\tvalidation_1-auc:0.71858\n",
      "[250]\tvalidation_0-auc:0.87029\tvalidation_1-auc:0.71861\n",
      "[251]\tvalidation_0-auc:0.87034\tvalidation_1-auc:0.71858\n",
      "[252]\tvalidation_0-auc:0.87037\tvalidation_1-auc:0.71866\n",
      "[253]\tvalidation_0-auc:0.87043\tvalidation_1-auc:0.71855\n",
      "[254]\tvalidation_0-auc:0.87045\tvalidation_1-auc:0.71857\n",
      "[255]\tvalidation_0-auc:0.87051\tvalidation_1-auc:0.71863\n",
      "[256]\tvalidation_0-auc:0.87053\tvalidation_1-auc:0.71867\n",
      "[257]\tvalidation_0-auc:0.87055\tvalidation_1-auc:0.71861\n",
      "[258]\tvalidation_0-auc:0.87060\tvalidation_1-auc:0.71868\n",
      "[259]\tvalidation_0-auc:0.87063\tvalidation_1-auc:0.71877\n",
      "[260]\tvalidation_0-auc:0.87068\tvalidation_1-auc:0.71879\n",
      "[261]\tvalidation_0-auc:0.87075\tvalidation_1-auc:0.71880\n",
      "[262]\tvalidation_0-auc:0.87078\tvalidation_1-auc:0.71872\n",
      "[263]\tvalidation_0-auc:0.87081\tvalidation_1-auc:0.71877\n",
      "[264]\tvalidation_0-auc:0.87083\tvalidation_1-auc:0.71883\n",
      "[265]\tvalidation_0-auc:0.87087\tvalidation_1-auc:0.71885\n",
      "[266]\tvalidation_0-auc:0.87089\tvalidation_1-auc:0.71887\n",
      "[267]\tvalidation_0-auc:0.87091\tvalidation_1-auc:0.71869\n",
      "[268]\tvalidation_0-auc:0.87096\tvalidation_1-auc:0.71864\n",
      "[269]\tvalidation_0-auc:0.87101\tvalidation_1-auc:0.71865\n",
      "[270]\tvalidation_0-auc:0.87105\tvalidation_1-auc:0.71857\n",
      "[271]\tvalidation_0-auc:0.87110\tvalidation_1-auc:0.71857\n",
      "[272]\tvalidation_0-auc:0.87114\tvalidation_1-auc:0.71861\n",
      "[273]\tvalidation_0-auc:0.87117\tvalidation_1-auc:0.71877\n",
      "[274]\tvalidation_0-auc:0.87120\tvalidation_1-auc:0.71893\n",
      "[275]\tvalidation_0-auc:0.87124\tvalidation_1-auc:0.71897\n",
      "[276]\tvalidation_0-auc:0.87127\tvalidation_1-auc:0.71888\n",
      "[277]\tvalidation_0-auc:0.87130\tvalidation_1-auc:0.71886\n",
      "[278]\tvalidation_0-auc:0.87133\tvalidation_1-auc:0.71888\n",
      "[279]\tvalidation_0-auc:0.87136\tvalidation_1-auc:0.71890\n",
      "[280]\tvalidation_0-auc:0.87140\tvalidation_1-auc:0.71896\n",
      "[281]\tvalidation_0-auc:0.87142\tvalidation_1-auc:0.71901\n",
      "[282]\tvalidation_0-auc:0.87145\tvalidation_1-auc:0.71895\n",
      "[283]\tvalidation_0-auc:0.87148\tvalidation_1-auc:0.71908\n",
      "[284]\tvalidation_0-auc:0.87150\tvalidation_1-auc:0.71910\n",
      "[285]\tvalidation_0-auc:0.87154\tvalidation_1-auc:0.71906\n",
      "[286]\tvalidation_0-auc:0.87158\tvalidation_1-auc:0.71908\n",
      "[287]\tvalidation_0-auc:0.87161\tvalidation_1-auc:0.71910\n",
      "[288]\tvalidation_0-auc:0.87163\tvalidation_1-auc:0.71914\n",
      "[289]\tvalidation_0-auc:0.87167\tvalidation_1-auc:0.71913\n",
      "[290]\tvalidation_0-auc:0.87169\tvalidation_1-auc:0.71909\n",
      "[291]\tvalidation_0-auc:0.87174\tvalidation_1-auc:0.71901\n",
      "[292]\tvalidation_0-auc:0.87177\tvalidation_1-auc:0.71908\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[293]\tvalidation_0-auc:0.87183\tvalidation_1-auc:0.71919\n",
      "[294]\tvalidation_0-auc:0.87187\tvalidation_1-auc:0.71918\n",
      "[295]\tvalidation_0-auc:0.87189\tvalidation_1-auc:0.71915\n",
      "[296]\tvalidation_0-auc:0.87191\tvalidation_1-auc:0.71907\n",
      "[297]\tvalidation_0-auc:0.87197\tvalidation_1-auc:0.71922\n",
      "[298]\tvalidation_0-auc:0.87198\tvalidation_1-auc:0.71922\n",
      "[299]\tvalidation_0-auc:0.87203\tvalidation_1-auc:0.71937\n",
      "[300]\tvalidation_0-auc:0.87204\tvalidation_1-auc:0.71940\n",
      "[301]\tvalidation_0-auc:0.87207\tvalidation_1-auc:0.71938\n",
      "[302]\tvalidation_0-auc:0.87211\tvalidation_1-auc:0.71942\n",
      "[303]\tvalidation_0-auc:0.87216\tvalidation_1-auc:0.71939\n",
      "[304]\tvalidation_0-auc:0.87219\tvalidation_1-auc:0.71932\n",
      "[305]\tvalidation_0-auc:0.87225\tvalidation_1-auc:0.71944\n",
      "[306]\tvalidation_0-auc:0.87228\tvalidation_1-auc:0.71943\n",
      "[307]\tvalidation_0-auc:0.87230\tvalidation_1-auc:0.71933\n",
      "[308]\tvalidation_0-auc:0.87232\tvalidation_1-auc:0.71942\n",
      "[309]\tvalidation_0-auc:0.87235\tvalidation_1-auc:0.71937\n",
      "[310]\tvalidation_0-auc:0.87238\tvalidation_1-auc:0.71928\n",
      "[311]\tvalidation_0-auc:0.87242\tvalidation_1-auc:0.71923\n",
      "[312]\tvalidation_0-auc:0.87244\tvalidation_1-auc:0.71914\n",
      "[313]\tvalidation_0-auc:0.87249\tvalidation_1-auc:0.71911\n",
      "[314]\tvalidation_0-auc:0.87255\tvalidation_1-auc:0.71913\n",
      "[315]\tvalidation_0-auc:0.87257\tvalidation_1-auc:0.71915\n",
      "[316]\tvalidation_0-auc:0.87260\tvalidation_1-auc:0.71923\n",
      "[317]\tvalidation_0-auc:0.87264\tvalidation_1-auc:0.71924\n",
      "[318]\tvalidation_0-auc:0.87266\tvalidation_1-auc:0.71908\n",
      "[319]\tvalidation_0-auc:0.87270\tvalidation_1-auc:0.71910\n",
      "[320]\tvalidation_0-auc:0.87272\tvalidation_1-auc:0.71914\n",
      "[321]\tvalidation_0-auc:0.87274\tvalidation_1-auc:0.71917\n",
      "[322]\tvalidation_0-auc:0.87274\tvalidation_1-auc:0.71916\n",
      "[323]\tvalidation_0-auc:0.87274\tvalidation_1-auc:0.71903\n",
      "[324]\tvalidation_0-auc:0.87277\tvalidation_1-auc:0.71897\n",
      "[325]\tvalidation_0-auc:0.87277\tvalidation_1-auc:0.71905\n",
      "[326]\tvalidation_0-auc:0.87279\tvalidation_1-auc:0.71902\n",
      "[327]\tvalidation_0-auc:0.87282\tvalidation_1-auc:0.71898\n",
      "[328]\tvalidation_0-auc:0.87284\tvalidation_1-auc:0.71902\n",
      "[329]\tvalidation_0-auc:0.87287\tvalidation_1-auc:0.71964\n",
      "[330]\tvalidation_0-auc:0.87290\tvalidation_1-auc:0.71965\n",
      "[331]\tvalidation_0-auc:0.87292\tvalidation_1-auc:0.71969\n",
      "[332]\tvalidation_0-auc:0.87293\tvalidation_1-auc:0.71967\n",
      "[333]\tvalidation_0-auc:0.87296\tvalidation_1-auc:0.71963\n",
      "[334]\tvalidation_0-auc:0.87300\tvalidation_1-auc:0.71957\n",
      "[335]\tvalidation_0-auc:0.87302\tvalidation_1-auc:0.71960\n",
      "[336]\tvalidation_0-auc:0.87305\tvalidation_1-auc:0.71957\n",
      "[337]\tvalidation_0-auc:0.87306\tvalidation_1-auc:0.71954\n",
      "[338]\tvalidation_0-auc:0.87309\tvalidation_1-auc:0.71952\n",
      "[339]\tvalidation_0-auc:0.87311\tvalidation_1-auc:0.71961\n",
      "[340]\tvalidation_0-auc:0.87314\tvalidation_1-auc:0.71947\n",
      "[341]\tvalidation_0-auc:0.87317\tvalidation_1-auc:0.71938\n",
      "[342]\tvalidation_0-auc:0.87319\tvalidation_1-auc:0.71942\n",
      "[343]\tvalidation_0-auc:0.87322\tvalidation_1-auc:0.71946\n",
      "[344]\tvalidation_0-auc:0.87326\tvalidation_1-auc:0.71944\n",
      "[345]\tvalidation_0-auc:0.87329\tvalidation_1-auc:0.71947\n",
      "[346]\tvalidation_0-auc:0.87331\tvalidation_1-auc:0.71935\n",
      "[347]\tvalidation_0-auc:0.87331\tvalidation_1-auc:0.71928\n",
      "[348]\tvalidation_0-auc:0.87334\tvalidation_1-auc:0.71930\n",
      "[349]\tvalidation_0-auc:0.87335\tvalidation_1-auc:0.71927\n",
      "[350]\tvalidation_0-auc:0.87336\tvalidation_1-auc:0.71938\n",
      "[351]\tvalidation_0-auc:0.87339\tvalidation_1-auc:0.71931\n",
      "[352]\tvalidation_0-auc:0.87340\tvalidation_1-auc:0.71927\n",
      "[353]\tvalidation_0-auc:0.87342\tvalidation_1-auc:0.71937\n",
      "[354]\tvalidation_0-auc:0.87344\tvalidation_1-auc:0.71925\n",
      "[355]\tvalidation_0-auc:0.87345\tvalidation_1-auc:0.71928\n",
      "[356]\tvalidation_0-auc:0.87348\tvalidation_1-auc:0.71921\n",
      "[357]\tvalidation_0-auc:0.87354\tvalidation_1-auc:0.71928\n",
      "[358]\tvalidation_0-auc:0.87357\tvalidation_1-auc:0.71928\n",
      "[359]\tvalidation_0-auc:0.87362\tvalidation_1-auc:0.71930\n",
      "[360]\tvalidation_0-auc:0.87363\tvalidation_1-auc:0.71917\n",
      "[361]\tvalidation_0-auc:0.87365\tvalidation_1-auc:0.71926\n",
      "[362]\tvalidation_0-auc:0.87367\tvalidation_1-auc:0.71923\n",
      "[363]\tvalidation_0-auc:0.87369\tvalidation_1-auc:0.71902\n",
      "[364]\tvalidation_0-auc:0.87372\tvalidation_1-auc:0.71901\n",
      "[365]\tvalidation_0-auc:0.87376\tvalidation_1-auc:0.71901\n",
      "[366]\tvalidation_0-auc:0.87381\tvalidation_1-auc:0.71910\n",
      "[367]\tvalidation_0-auc:0.87383\tvalidation_1-auc:0.71914\n",
      "[368]\tvalidation_0-auc:0.87386\tvalidation_1-auc:0.71910\n",
      "[369]\tvalidation_0-auc:0.87389\tvalidation_1-auc:0.71899\n",
      "[370]\tvalidation_0-auc:0.87393\tvalidation_1-auc:0.71890\n",
      "[371]\tvalidation_0-auc:0.87395\tvalidation_1-auc:0.71890\n",
      "[372]\tvalidation_0-auc:0.87396\tvalidation_1-auc:0.71895\n",
      "[373]\tvalidation_0-auc:0.87399\tvalidation_1-auc:0.71894\n",
      "[374]\tvalidation_0-auc:0.87400\tvalidation_1-auc:0.71901\n",
      "[375]\tvalidation_0-auc:0.87406\tvalidation_1-auc:0.71913\n",
      "[376]\tvalidation_0-auc:0.87408\tvalidation_1-auc:0.71910\n",
      "[377]\tvalidation_0-auc:0.87408\tvalidation_1-auc:0.71907\n",
      "[378]\tvalidation_0-auc:0.87410\tvalidation_1-auc:0.71912\n",
      "[379]\tvalidation_0-auc:0.87411\tvalidation_1-auc:0.71909\n",
      "[380]\tvalidation_0-auc:0.87414\tvalidation_1-auc:0.71916\n",
      "[381]\tvalidation_0-auc:0.87416\tvalidation_1-auc:0.71919\n",
      "[382]\tvalidation_0-auc:0.87422\tvalidation_1-auc:0.71916\n",
      "[383]\tvalidation_0-auc:0.87423\tvalidation_1-auc:0.71917\n",
      "[384]\tvalidation_0-auc:0.87426\tvalidation_1-auc:0.71918\n",
      "[385]\tvalidation_0-auc:0.87428\tvalidation_1-auc:0.71919\n",
      "[386]\tvalidation_0-auc:0.87428\tvalidation_1-auc:0.71916\n",
      "[387]\tvalidation_0-auc:0.87430\tvalidation_1-auc:0.71921\n",
      "[388]\tvalidation_0-auc:0.87431\tvalidation_1-auc:0.71923\n",
      "[389]\tvalidation_0-auc:0.87434\tvalidation_1-auc:0.71928\n",
      "[390]\tvalidation_0-auc:0.87436\tvalidation_1-auc:0.71925\n",
      "[391]\tvalidation_0-auc:0.87438\tvalidation_1-auc:0.71920\n",
      "[392]\tvalidation_0-auc:0.87440\tvalidation_1-auc:0.71917\n",
      "[393]\tvalidation_0-auc:0.87443\tvalidation_1-auc:0.71918\n",
      "[394]\tvalidation_0-auc:0.87446\tvalidation_1-auc:0.71921\n",
      "[395]\tvalidation_0-auc:0.87448\tvalidation_1-auc:0.71925\n",
      "[396]\tvalidation_0-auc:0.87450\tvalidation_1-auc:0.71927\n",
      "[397]\tvalidation_0-auc:0.87454\tvalidation_1-auc:0.71919\n",
      "[398]\tvalidation_0-auc:0.87456\tvalidation_1-auc:0.71926\n",
      "[399]\tvalidation_0-auc:0.87457\tvalidation_1-auc:0.71935\n",
      "[400]\tvalidation_0-auc:0.87462\tvalidation_1-auc:0.71931\n",
      "[401]\tvalidation_0-auc:0.87464\tvalidation_1-auc:0.71936\n",
      "[402]\tvalidation_0-auc:0.87465\tvalidation_1-auc:0.71935\n",
      "[403]\tvalidation_0-auc:0.87467\tvalidation_1-auc:0.71933\n",
      "[404]\tvalidation_0-auc:0.87469\tvalidation_1-auc:0.71935\n",
      "[405]\tvalidation_0-auc:0.87473\tvalidation_1-auc:0.71924\n",
      "[406]\tvalidation_0-auc:0.87480\tvalidation_1-auc:0.71920\n",
      "[407]\tvalidation_0-auc:0.87483\tvalidation_1-auc:0.71915\n",
      "[408]\tvalidation_0-auc:0.87484\tvalidation_1-auc:0.71910\n",
      "[409]\tvalidation_0-auc:0.87489\tvalidation_1-auc:0.71898\n",
      "[410]\tvalidation_0-auc:0.87491\tvalidation_1-auc:0.71889\n",
      "[411]\tvalidation_0-auc:0.87493\tvalidation_1-auc:0.71893\n",
      "[412]\tvalidation_0-auc:0.87494\tvalidation_1-auc:0.71890\n",
      "[413]\tvalidation_0-auc:0.87496\tvalidation_1-auc:0.71895\n",
      "[414]\tvalidation_0-auc:0.87498\tvalidation_1-auc:0.71891\n",
      "[415]\tvalidation_0-auc:0.87501\tvalidation_1-auc:0.71903\n",
      "[416]\tvalidation_0-auc:0.87503\tvalidation_1-auc:0.71902\n",
      "[417]\tvalidation_0-auc:0.87505\tvalidation_1-auc:0.71899\n",
      "[418]\tvalidation_0-auc:0.87506\tvalidation_1-auc:0.71904\n",
      "[419]\tvalidation_0-auc:0.87508\tvalidation_1-auc:0.71901\n",
      "[420]\tvalidation_0-auc:0.87509\tvalidation_1-auc:0.71905\n",
      "[421]\tvalidation_0-auc:0.87513\tvalidation_1-auc:0.71930\n",
      "[422]\tvalidation_0-auc:0.87514\tvalidation_1-auc:0.71934\n",
      "[423]\tvalidation_0-auc:0.87516\tvalidation_1-auc:0.71933\n",
      "[424]\tvalidation_0-auc:0.87516\tvalidation_1-auc:0.71938\n",
      "[425]\tvalidation_0-auc:0.87518\tvalidation_1-auc:0.71944\n",
      "[426]\tvalidation_0-auc:0.87520\tvalidation_1-auc:0.71950\n",
      "[427]\tvalidation_0-auc:0.87523\tvalidation_1-auc:0.71952\n",
      "[428]\tvalidation_0-auc:0.87525\tvalidation_1-auc:0.71952\n",
      "[429]\tvalidation_0-auc:0.87528\tvalidation_1-auc:0.71943\n",
      "[430]\tvalidation_0-auc:0.87532\tvalidation_1-auc:0.71942\n",
      "[431]\tvalidation_0-auc:0.87533\tvalidation_1-auc:0.71945\n",
      "Stopping. Best iteration:\n",
      "[331]\tvalidation_0-auc:0.87292\tvalidation_1-auc:0.71969\n",
      "\n"
     ]
    }
   ],
   "source": [
    "clf = xgb.XGBClassifier(**params, missing=missing_impute)\n",
    "    \n",
    "clf.fit(df_train[feature_cols], df_train[label_col], early_stopping_rounds=100, eval_set=[(df_train[feature_cols], df_train[label_col]), (df_eval[feature_cols], df_eval[label_col])], eval_metric='auc', verbose=True)\n",
    "\n",
    "eval_score = roc_auc_score(df_eval[label_col], clf.predict_proba(df_eval[feature_cols])[:,1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Eval AUC: 0.7196878687075241\n"
     ]
    }
   ],
   "source": [
    "print('Eval AUC: {}'.format(eval_score))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "best_iter = clf.best_iteration\n",
    "params['n_estimators'] = best_iter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_train = pd.concat((df_train, df_eval))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC: 0.7534414911857782, ACC: 0.8277850102287521\n"
     ]
    }
   ],
   "source": [
    "clf = xgb.XGBClassifier(**params, missing=missing_impute)\n",
    "\n",
    "clf.fit(df_train[feature_cols], df_train[label_col], eval_metric='auc', verbose=True)\n",
    "\n",
    "eval_score_auc = roc_auc_score(df_train[label_col], clf.predict(df_train[feature_cols]))\n",
    "eval_score_acc = accuracy_score(df_train[label_col], clf.predict(df_train[feature_cols]))\n",
    "\n",
    "print('AUC: {}, ACC: {}'.format(eval_score_auc, eval_score_acc))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "# preds = clf.predict(df_test[feature_cols])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "preds = clf.predict_proba(df_test[feature_cols])\n",
    "preds = preds[:, 1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Patient_ID\tHealth_Camp_ID\tOutcome"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "submission = pd.DataFrame({'Patient_ID':df_test['Patient_ID'], 'Health_Camp_ID':df_test['Health_Camp_ID'], 'Outcome':preds})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "submission.to_csv('xgb_post_competition_oot_eval.csv', index=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.rcParams['figure.figsize'] = (12, 6)\n",
    "xgb.plot_importance(clf)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# blend"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ITERATION: 1, SEED: 1, AUC: 0.7531462424762566, ACC: 0.827599032917984\n",
      "ITERATION: 2, SEED: 3, AUC: 0.754217529511367, ACC: 0.8277850102287521\n",
      "ITERATION: 3, SEED: 10, AUC: 0.7534323043893612, ACC: 0.8275724647307314\n",
      "ITERATION: 4, SEED: 15, AUC: 0.7523944294668319, ACC: 0.82714737373469\n",
      "ITERATION: 5, SEED: 20, AUC: 0.7531127591964716, ACC: 0.8275724647307314\n",
      "ITERATION: 6, SEED: 33, AUC: 0.7524796450766292, ACC: 0.8271606578283164\n",
      "ITERATION: 7, SEED: 333, AUC: 0.7531340408591697, ACC: 0.8274927601689737\n",
      "ITERATION: 8, SEED: 1997, AUC: 0.7534597580278065, ACC: 0.8278115784160047\n",
      "ITERATION: 9, SEED: 2020, AUC: 0.7540257632536711, ACC: 0.8276388851988629\n",
      "ITERATION: 10, SEED: 2021, AUC: 0.7543149467382512, ACC: 0.8279045670713887\n",
      "ITERATION: 11, SEED: 87, AUC: 0.7534414911857782, ACC: 0.8277850102287521\n"
     ]
    }
   ],
   "source": [
    "predictions = []\n",
    "best_seed = params['random_state']\n",
    "SEEDS = [1,3,10,15,20,33,333,1997,2020,2021, best_seed]\n",
    "del params['random_state']\n",
    "for idx, seed_val in enumerate(SEEDS):\n",
    "    clf = xgb.XGBClassifier(**params, missing=missing_impute, random_state=seed_val)\n",
    "    clf.fit(df_train[feature_cols], df_train[label_col], eval_metric='auc', verbose=True)\n",
    "    eval_score_auc = roc_auc_score(df_train[label_col], clf.predict(df_train[feature_cols]))\n",
    "    eval_score_acc = accuracy_score(df_train[label_col], clf.predict(df_train[feature_cols]))\n",
    "    print('ITERATION: {}, SEED: {}, AUC: {}, ACC: {}'.format(idx + 1, seed_val, eval_score_auc, eval_score_acc))\n",
    "    predictions.append(clf.predict_proba(df_test[feature_cols])[:, 1])\n",
    "\n",
    "preds = np.mean(predictions, axis=0).reshape(-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "submission = pd.DataFrame({'Patient_ID':df_test['Patient_ID'], 'Health_Camp_ID':df_test['Health_Camp_ID'], 'Outcome':preds})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "submission.to_csv('xgb_blended.csv', index=None)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# underfit blend"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ITERATION: 1, SEED: 1, AUC: 0.7506778622783944, ACC: 0.8265761577087596\n",
      "ITERATION: 2, SEED: 3, AUC: 0.7523243947111894, ACC: 0.8270012487048009\n",
      "ITERATION: 3, SEED: 10, AUC: 0.750963888607897, ACC: 0.826416748585244\n",
      "ITERATION: 4, SEED: 15, AUC: 0.7505926110849951, ACC: 0.8264300326788703\n",
      "ITERATION: 5, SEED: 20, AUC: 0.750528677137896, ACC: 0.8263370440234863\n",
      "ITERATION: 6, SEED: 33, AUC: 0.7500234664958539, ACC: 0.8262440553681022\n",
      "ITERATION: 7, SEED: 333, AUC: 0.7512439208796619, ACC: 0.8266027258960121\n",
      "ITERATION: 8, SEED: 1997, AUC: 0.7505196682594882, ACC: 0.8267887032067802\n",
      "ITERATION: 9, SEED: 2020, AUC: 0.7525131105646136, ACC: 0.8271208055474375\n",
      "ITERATION: 10, SEED: 2021, AUC: 0.7519440193508751, ACC: 0.8271340896410638\n",
      "ITERATION: 11, SEED: 87, AUC: 0.7505744332019715, ACC: 0.826735566832275\n"
     ]
    }
   ],
   "source": [
    "predictions = []\n",
    "SEEDS = [1,3,10,15,20,33,333,1997,2020,2021, best_seed]\n",
    "params['n_estimators'] = int(best_iter - 0.15*best_iter)\n",
    "for idx, seed_val in enumerate(SEEDS):\n",
    "    clf = xgb.XGBClassifier(**params, missing=missing_impute, random_state=seed_val)\n",
    "    clf.fit(df_train[feature_cols], df_train[label_col], eval_metric='auc', verbose=True)\n",
    "    eval_score_auc = roc_auc_score(df_train[label_col], clf.predict(df_train[feature_cols]))\n",
    "    eval_score_acc = accuracy_score(df_train[label_col], clf.predict(df_train[feature_cols]))\n",
    "    print('ITERATION: {}, SEED: {}, AUC: {}, ACC: {}'.format(idx + 1, seed_val, eval_score_auc, eval_score_acc))\n",
    "    predictions.append(clf.predict_proba(df_test[feature_cols])[:, 1])\n",
    "\n",
    "preds = np.mean(predictions, axis=0).reshape(-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "submission = pd.DataFrame({'Patient_ID':df_test['Patient_ID'], 'Health_Camp_ID':df_test['Health_Camp_ID'], 'Outcome':preds})\n",
    "submission.to_csv('xgb_underfits_blended.csv', index=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# overfit blend"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ITERATION: 1, SEED: 1, AUC: 0.7551398449075902, ACC: 0.8285289194718244\n",
      "ITERATION: 2, SEED: 3, AUC: 0.7555445523348743, ACC: 0.8283429421610563\n",
      "ITERATION: 3, SEED: 10, AUC: 0.7555780534064602, ACC: 0.8284359308164404\n",
      "ITERATION: 4, SEED: 15, AUC: 0.7548810942985236, ACC: 0.8282632375992987\n",
      "ITERATION: 5, SEED: 20, AUC: 0.7549115449658376, ACC: 0.8283296580674301\n",
      "ITERATION: 6, SEED: 33, AUC: 0.7553163769357283, ACC: 0.8286086240335822\n",
      "ITERATION: 7, SEED: 333, AUC: 0.75478670968411, ACC: 0.8281038284757831\n",
      "ITERATION: 8, SEED: 1997, AUC: 0.7552525497394348, ACC: 0.8289141581869869\n",
      "ITERATION: 9, SEED: 2020, AUC: 0.7562811489065423, ACC: 0.8287946013443502\n",
      "ITERATION: 10, SEED: 2021, AUC: 0.7555659229565772, ACC: 0.8285953399399559\n",
      "ITERATION: 11, SEED: 87, AUC: 0.7550759109604909, ACC: 0.8284359308164404\n"
     ]
    }
   ],
   "source": [
    "predictions = []\n",
    "SEEDS = [1,3,10,15,20,33,333,1997,2020,2021, best_seed]\n",
    "params['n_estimators'] = int(best_iter + 0.15*best_iter)\n",
    "for idx, seed_val in enumerate(SEEDS):\n",
    "    clf = xgb.XGBClassifier(**params, missing=missing_impute, random_state=seed_val)\n",
    "    clf.fit(df_train[feature_cols], df_train[label_col], eval_metric='auc', verbose=True)\n",
    "    eval_score_auc = roc_auc_score(df_train[label_col], clf.predict(df_train[feature_cols]))\n",
    "    eval_score_acc = accuracy_score(df_train[label_col], clf.predict(df_train[feature_cols]))\n",
    "    print('ITERATION: {}, SEED: {}, AUC: {}, ACC: {}'.format(idx + 1, seed_val, eval_score_auc, eval_score_acc))\n",
    "    predictions.append(clf.predict_proba(df_test[feature_cols])[:, 1])\n",
    "\n",
    "preds = np.mean(predictions, axis=0).reshape(-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "submission = pd.DataFrame({'Patient_ID':df_test['Patient_ID'], 'Health_Camp_ID':df_test['Health_Camp_ID'], 'Outcome':preds})\n",
    "submission.to_csv('xgb_overfits_blended.csv', index=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ITERATION: 1, SEED: 1, AUC: 0.7561624856005617, ACC: 0.8288875899997343\n",
      "ITERATION: 2, SEED: 3, AUC: 0.7570633414160772, ACC: 0.8290469991232499\n",
      "ITERATION: 3, SEED: 10, AUC: 0.756685945292831, ACC: 0.8289407263742394\n",
      "ITERATION: 4, SEED: 15, AUC: 0.7563755335209558, ACC: 0.8289540104678658\n",
      "ITERATION: 5, SEED: 20, AUC: 0.7560498163523189, ACC: 0.8286351922208348\n",
      "ITERATION: 6, SEED: 33, AUC: 0.7562842348944159, ACC: 0.8289540104678658\n",
      "ITERATION: 7, SEED: 333, AUC: 0.7560741662110897, ACC: 0.828648476314461\n",
      "ITERATION: 8, SEED: 1997, AUC: 0.7564090879679445, ACC: 0.8292462605276442\n",
      "ITERATION: 9, SEED: 2020, AUC: 0.7569416277058251, ACC: 0.8291134195913813\n",
      "ITERATION: 10, SEED: 2021, AUC: 0.7562080993302298, ACC: 0.8287547490634714\n",
      "ITERATION: 11, SEED: 87, AUC: 0.7560590743159398, ACC: 0.8291134195913813\n"
     ]
    }
   ],
   "source": [
    "predictions = []\n",
    "SEEDS = [1,3,10,15,20,33,333,1997,2020,2021, best_seed]\n",
    "params['n_estimators'] = int(best_iter + 0.25*best_iter)\n",
    "for idx, seed_val in enumerate(SEEDS):\n",
    "    clf = xgb.XGBClassifier(**params, missing=missing_impute, random_state=seed_val)\n",
    "    clf.fit(df_train[feature_cols], df_train[label_col], eval_metric='auc', verbose=True)\n",
    "    eval_score_auc = roc_auc_score(df_train[label_col], clf.predict(df_train[feature_cols]))\n",
    "    eval_score_acc = accuracy_score(df_train[label_col], clf.predict(df_train[feature_cols]))\n",
    "    print('ITERATION: {}, SEED: {}, AUC: {}, ACC: {}'.format(idx + 1, seed_val, eval_score_auc, eval_score_acc))\n",
    "    predictions.append(clf.predict_proba(df_test[feature_cols])[:, 1])\n",
    "\n",
    "preds = np.mean(predictions, axis=0).reshape(-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [],
   "source": [
    "submission = pd.DataFrame({'Patient_ID':df_test['Patient_ID'], 'Health_Camp_ID':df_test['Health_Camp_ID'], 'Outcome':preds})\n",
    "submission.to_csv('xgb_overfits2_blended.csv', index=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}