enem4.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "from catboost import CatBoostClassifier\n",
    "from sklearn.impute import SimpleImputer\n",
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn.compose import ColumnTransformer\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.preprocessing import StandardScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_df = pd.read_csv('train.csv')\n",
    "train_target = train_df.pop('IN_TREINEIRO')\n",
    "test_df = pd.read_csv('test.csv')\n",
    "test_columns = test_df.columns.values\n",
    "train_df = train_df[test_columns]\n",
    "train_df.drop('NU_INSCRICAO', axis=1, inplace=True)\n",
    "test_inscricao = test_df.pop('NU_INSCRICAO')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "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>CO_UF_RESIDENCIA</th>\n",
       "      <th>SG_UF_RESIDENCIA</th>\n",
       "      <th>NU_IDADE</th>\n",
       "      <th>TP_SEXO</th>\n",
       "      <th>TP_COR_RACA</th>\n",
       "      <th>TP_NACIONALIDADE</th>\n",
       "      <th>TP_ST_CONCLUSAO</th>\n",
       "      <th>TP_ANO_CONCLUIU</th>\n",
       "      <th>TP_ESCOLA</th>\n",
       "      <th>TP_ENSINO</th>\n",
       "      <th>...</th>\n",
       "      <th>NU_NOTA_COMP5</th>\n",
       "      <th>NU_NOTA_REDACAO</th>\n",
       "      <th>Q001</th>\n",
       "      <th>Q002</th>\n",
       "      <th>Q006</th>\n",
       "      <th>Q024</th>\n",
       "      <th>Q025</th>\n",
       "      <th>Q026</th>\n",
       "      <th>Q027</th>\n",
       "      <th>Q047</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>43</td>\n",
       "      <td>RS</td>\n",
       "      <td>24</td>\n",
       "      <td>M</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>80.0</td>\n",
       "      <td>520.0</td>\n",
       "      <td>D</td>\n",
       "      <td>D</td>\n",
       "      <td>C</td>\n",
       "      <td>A</td>\n",
       "      <td>A</td>\n",
       "      <td>C</td>\n",
       "      <td>H</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>23</td>\n",
       "      <td>CE</td>\n",
       "      <td>17</td>\n",
       "      <td>F</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1.0</td>\n",
       "      <td>...</td>\n",
       "      <td>80.0</td>\n",
       "      <td>580.0</td>\n",
       "      <td>A</td>\n",
       "      <td>A</td>\n",
       "      <td>B</td>\n",
       "      <td>A</td>\n",
       "      <td>A</td>\n",
       "      <td>A</td>\n",
       "      <td>NaN</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>23</td>\n",
       "      <td>CE</td>\n",
       "      <td>21</td>\n",
       "      <td>F</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>D</td>\n",
       "      <td>D</td>\n",
       "      <td>C</td>\n",
       "      <td>A</td>\n",
       "      <td>A</td>\n",
       "      <td>A</td>\n",
       "      <td>NaN</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>33</td>\n",
       "      <td>RJ</td>\n",
       "      <td>25</td>\n",
       "      <td>F</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>9</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>H</td>\n",
       "      <td>E</td>\n",
       "      <td>E</td>\n",
       "      <td>C</td>\n",
       "      <td>B</td>\n",
       "      <td>C</td>\n",
       "      <td>F</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>13</td>\n",
       "      <td>AM</td>\n",
       "      <td>28</td>\n",
       "      <td>M</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>E</td>\n",
       "      <td>D</td>\n",
       "      <td>C</td>\n",
       "      <td>A</td>\n",
       "      <td>A</td>\n",
       "      <td>B</td>\n",
       "      <td>F</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 42 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   CO_UF_RESIDENCIA SG_UF_RESIDENCIA  NU_IDADE TP_SEXO  TP_COR_RACA  \\\n",
       "0                43               RS        24       M            1   \n",
       "1                23               CE        17       F            3   \n",
       "2                23               CE        21       F            3   \n",
       "3                33               RJ        25       F            0   \n",
       "4                13               AM        28       M            2   \n",
       "\n",
       "   TP_NACIONALIDADE  TP_ST_CONCLUSAO  TP_ANO_CONCLUIU  TP_ESCOLA  TP_ENSINO  \\\n",
       "0                 1                1                4          1        NaN   \n",
       "1                 1                2                0          2        1.0   \n",
       "2                 1                3                0          1        NaN   \n",
       "3                 1                1                9          1        NaN   \n",
       "4                 1                1                4          1        NaN   \n",
       "\n",
       "   ...  NU_NOTA_COMP5  NU_NOTA_REDACAO  Q001  Q002  Q006  Q024  Q025  Q026  \\\n",
       "0  ...           80.0            520.0     D     D     C     A     A     C   \n",
       "1  ...           80.0            580.0     A     A     B     A     A     A   \n",
       "2  ...            NaN              NaN     D     D     C     A     A     A   \n",
       "3  ...            NaN              NaN     H     E     E     C     B     C   \n",
       "4  ...            NaN              NaN     E     D     C     A     A     B   \n",
       "\n",
       "   Q027  Q047  \n",
       "0     H     A  \n",
       "1   NaN     A  \n",
       "2   NaN     A  \n",
       "3     F     D  \n",
       "4     F     A  \n",
       "\n",
       "[5 rows x 42 columns]"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 13730 entries, 0 to 13729\n",
      "Data columns (total 42 columns):\n",
      "CO_UF_RESIDENCIA          13730 non-null int64\n",
      "SG_UF_RESIDENCIA          13730 non-null object\n",
      "NU_IDADE                  13730 non-null int64\n",
      "TP_SEXO                   13730 non-null object\n",
      "TP_COR_RACA               13730 non-null int64\n",
      "TP_NACIONALIDADE          13730 non-null int64\n",
      "TP_ST_CONCLUSAO           13730 non-null int64\n",
      "TP_ANO_CONCLUIU           13730 non-null int64\n",
      "TP_ESCOLA                 13730 non-null int64\n",
      "TP_ENSINO                 4282 non-null float64\n",
      "TP_DEPENDENCIA_ADM_ESC    4282 non-null float64\n",
      "IN_BAIXA_VISAO            13730 non-null int64\n",
      "IN_CEGUEIRA               13730 non-null int64\n",
      "IN_SURDEZ                 13730 non-null int64\n",
      "IN_DISLEXIA               13730 non-null int64\n",
      "IN_DISCALCULIA            13730 non-null int64\n",
      "IN_SABATISTA              13730 non-null int64\n",
      "IN_GESTANTE               13730 non-null int64\n",
      "IN_IDOSO                  13730 non-null int64\n",
      "TP_PRESENCA_CN            13730 non-null int64\n",
      "TP_PRESENCA_CH            13730 non-null int64\n",
      "TP_PRESENCA_LC            13730 non-null int64\n",
      "TP_PRESENCA_MT            13730 non-null int64\n",
      "NU_NOTA_CN                10341 non-null float64\n",
      "NU_NOTA_CH                10341 non-null float64\n",
      "NU_NOTA_LC                10133 non-null float64\n",
      "TP_LINGUA                 13730 non-null int64\n",
      "TP_STATUS_REDACAO         10133 non-null float64\n",
      "NU_NOTA_COMP1             10133 non-null float64\n",
      "NU_NOTA_COMP2             10133 non-null float64\n",
      "NU_NOTA_COMP3             10133 non-null float64\n",
      "NU_NOTA_COMP4             10133 non-null float64\n",
      "NU_NOTA_COMP5             10133 non-null float64\n",
      "NU_NOTA_REDACAO           10133 non-null float64\n",
      "Q001                      13730 non-null object\n",
      "Q002                      13730 non-null object\n",
      "Q006                      13730 non-null object\n",
      "Q024                      13730 non-null object\n",
      "Q025                      13730 non-null object\n",
      "Q026                      13730 non-null object\n",
      "Q027                      6357 non-null object\n",
      "Q047                      13730 non-null object\n",
      "dtypes: float64(12), int64(20), object(10)\n",
      "memory usage: 4.4+ MB\n"
     ]
    }
   ],
   "source": [
    "train_df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    13730.000000\n",
       "mean         0.129862\n",
       "std          0.336163\n",
       "min          0.000000\n",
       "25%          0.000000\n",
       "50%          0.000000\n",
       "75%          0.000000\n",
       "max          1.000000\n",
       "Name: IN_TREINEIRO, dtype: float64"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_target.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "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>CO_UF_RESIDENCIA</th>\n",
       "      <th>SG_UF_RESIDENCIA</th>\n",
       "      <th>NU_IDADE</th>\n",
       "      <th>TP_SEXO</th>\n",
       "      <th>TP_COR_RACA</th>\n",
       "      <th>TP_NACIONALIDADE</th>\n",
       "      <th>TP_ST_CONCLUSAO</th>\n",
       "      <th>TP_ANO_CONCLUIU</th>\n",
       "      <th>TP_ESCOLA</th>\n",
       "      <th>TP_ENSINO</th>\n",
       "      <th>...</th>\n",
       "      <th>NU_NOTA_COMP5</th>\n",
       "      <th>NU_NOTA_REDACAO</th>\n",
       "      <th>Q001</th>\n",
       "      <th>Q002</th>\n",
       "      <th>Q006</th>\n",
       "      <th>Q024</th>\n",
       "      <th>Q025</th>\n",
       "      <th>Q026</th>\n",
       "      <th>Q027</th>\n",
       "      <th>Q047</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>43</td>\n",
       "      <td>RS</td>\n",
       "      <td>19</td>\n",
       "      <td>F</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>E</td>\n",
       "      <td>E</td>\n",
       "      <td>H</td>\n",
       "      <td>B</td>\n",
       "      <td>B</td>\n",
       "      <td>C</td>\n",
       "      <td>D</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>15</td>\n",
       "      <td>PA</td>\n",
       "      <td>24</td>\n",
       "      <td>M</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>B</td>\n",
       "      <td>C</td>\n",
       "      <td>B</td>\n",
       "      <td>A</td>\n",
       "      <td>A</td>\n",
       "      <td>C</td>\n",
       "      <td>D</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>29</td>\n",
       "      <td>BA</td>\n",
       "      <td>16</td>\n",
       "      <td>F</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>E</td>\n",
       "      <td>F</td>\n",
       "      <td>G</td>\n",
       "      <td>B</td>\n",
       "      <td>B</td>\n",
       "      <td>A</td>\n",
       "      <td>NaN</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>41</td>\n",
       "      <td>PR</td>\n",
       "      <td>17</td>\n",
       "      <td>F</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1.0</td>\n",
       "      <td>...</td>\n",
       "      <td>40.0</td>\n",
       "      <td>480.0</td>\n",
       "      <td>E</td>\n",
       "      <td>E</td>\n",
       "      <td>E</td>\n",
       "      <td>C</td>\n",
       "      <td>B</td>\n",
       "      <td>B</td>\n",
       "      <td>C</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>53</td>\n",
       "      <td>DF</td>\n",
       "      <td>19</td>\n",
       "      <td>F</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>80.0</td>\n",
       "      <td>720.0</td>\n",
       "      <td>E</td>\n",
       "      <td>E</td>\n",
       "      <td>E</td>\n",
       "      <td>B</td>\n",
       "      <td>B</td>\n",
       "      <td>B</td>\n",
       "      <td>D</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 42 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   CO_UF_RESIDENCIA SG_UF_RESIDENCIA  NU_IDADE TP_SEXO  TP_COR_RACA  \\\n",
       "0                43               RS        19       F            1   \n",
       "1                15               PA        24       M            3   \n",
       "2                29               BA        16       F            2   \n",
       "3                41               PR        17       F            1   \n",
       "4                53               DF        19       F            1   \n",
       "\n",
       "   TP_NACIONALIDADE  TP_ST_CONCLUSAO  TP_ANO_CONCLUIU  TP_ESCOLA  TP_ENSINO  \\\n",
       "0                 1                1                3          1        NaN   \n",
       "1                 2                1                4          1        NaN   \n",
       "2                 1                3                0          1        NaN   \n",
       "3                 1                2                0          2        1.0   \n",
       "4                 1                1                1          1        NaN   \n",
       "\n",
       "   ...  NU_NOTA_COMP5  NU_NOTA_REDACAO  Q001  Q002  Q006  Q024  Q025  Q026  \\\n",
       "0  ...            NaN              NaN     E     E     H     B     B     C   \n",
       "1  ...            NaN              NaN     B     C     B     A     A     C   \n",
       "2  ...            NaN              NaN     E     F     G     B     B     A   \n",
       "3  ...           40.0            480.0     E     E     E     C     B     B   \n",
       "4  ...           80.0            720.0     E     E     E     B     B     B   \n",
       "\n",
       "   Q027  Q047  \n",
       "0     D     A  \n",
       "1     D     A  \n",
       "2   NaN     D  \n",
       "3     C     A  \n",
       "4     D     A  \n",
       "\n",
       "[5 rows x 42 columns]"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 4570 entries, 0 to 4569\n",
      "Data columns (total 42 columns):\n",
      "CO_UF_RESIDENCIA          4570 non-null int64\n",
      "SG_UF_RESIDENCIA          4570 non-null object\n",
      "NU_IDADE                  4570 non-null int64\n",
      "TP_SEXO                   4570 non-null object\n",
      "TP_COR_RACA               4570 non-null int64\n",
      "TP_NACIONALIDADE          4570 non-null int64\n",
      "TP_ST_CONCLUSAO           4570 non-null int64\n",
      "TP_ANO_CONCLUIU           4570 non-null int64\n",
      "TP_ESCOLA                 4570 non-null int64\n",
      "TP_ENSINO                 1426 non-null float64\n",
      "TP_DEPENDENCIA_ADM_ESC    1426 non-null float64\n",
      "IN_BAIXA_VISAO            4570 non-null int64\n",
      "IN_CEGUEIRA               4570 non-null int64\n",
      "IN_SURDEZ                 4570 non-null int64\n",
      "IN_DISLEXIA               4570 non-null int64\n",
      "IN_DISCALCULIA            4570 non-null int64\n",
      "IN_SABATISTA              4570 non-null int64\n",
      "IN_GESTANTE               4570 non-null int64\n",
      "IN_IDOSO                  4570 non-null int64\n",
      "TP_PRESENCA_CN            4570 non-null int64\n",
      "TP_PRESENCA_CH            4570 non-null int64\n",
      "TP_PRESENCA_LC            4570 non-null int64\n",
      "TP_PRESENCA_MT            4570 non-null int64\n",
      "NU_NOTA_CN                3458 non-null float64\n",
      "NU_NOTA_CH                3458 non-null float64\n",
      "NU_NOTA_LC                3400 non-null float64\n",
      "TP_LINGUA                 4570 non-null int64\n",
      "TP_STATUS_REDACAO         3400 non-null float64\n",
      "NU_NOTA_COMP1             3400 non-null float64\n",
      "NU_NOTA_COMP2             3400 non-null float64\n",
      "NU_NOTA_COMP3             3400 non-null float64\n",
      "NU_NOTA_COMP4             3400 non-null float64\n",
      "NU_NOTA_COMP5             3400 non-null float64\n",
      "NU_NOTA_REDACAO           3400 non-null float64\n",
      "Q001                      4570 non-null object\n",
      "Q002                      4570 non-null object\n",
      "Q006                      4570 non-null object\n",
      "Q024                      4570 non-null object\n",
      "Q025                      4570 non-null object\n",
      "Q026                      4570 non-null object\n",
      "Q027                      2133 non-null object\n",
      "Q047                      4570 non-null object\n",
      "dtypes: float64(12), int64(20), object(10)\n",
      "memory usage: 1.5+ MB\n"
     ]
    }
   ],
   "source": [
    "test_df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "numeric_features = train_df.select_dtypes(include='number').columns.values\n",
    "cat_features = train_df.select_dtypes(exclude='number').columns.values\n",
    "processors = {'num_imputer': SimpleImputer(strategy='median'),\n",
    "              'scaler': StandardScaler(),\n",
    "              'cat_imputer': SimpleImputer(strategy='constant', fill_value='NA')}\n",
    "def preprocess_fit(df, processors):\n",
    "    num_df = df[numeric_features]\n",
    "    cat_df = df[cat_features]\n",
    "    num_processed = processors['num_imputer'].fit_transform(num_df)\n",
    "    #num_processed = processors['scaler'].fit_transform(num_processed)\n",
    "    num_df = pd.DataFrame(num_processed, columns=num_df.columns)\n",
    "    cat_processed = processors['cat_imputer'].fit_transform(cat_df)\n",
    "    cat_df = pd.DataFrame(cat_processed, columns=cat_df.columns)\n",
    "    return pd.concat((num_df, cat_df), axis=1)\n",
    "    \n",
    "train_df = preprocess_fit(train_df, processors)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "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>CO_UF_RESIDENCIA</th>\n",
       "      <th>NU_IDADE</th>\n",
       "      <th>TP_COR_RACA</th>\n",
       "      <th>TP_NACIONALIDADE</th>\n",
       "      <th>TP_ST_CONCLUSAO</th>\n",
       "      <th>TP_ANO_CONCLUIU</th>\n",
       "      <th>TP_ESCOLA</th>\n",
       "      <th>TP_ENSINO</th>\n",
       "      <th>TP_DEPENDENCIA_ADM_ESC</th>\n",
       "      <th>IN_BAIXA_VISAO</th>\n",
       "      <th>...</th>\n",
       "      <th>SG_UF_RESIDENCIA</th>\n",
       "      <th>TP_SEXO</th>\n",
       "      <th>Q001</th>\n",
       "      <th>Q002</th>\n",
       "      <th>Q006</th>\n",
       "      <th>Q024</th>\n",
       "      <th>Q025</th>\n",
       "      <th>Q026</th>\n",
       "      <th>Q027</th>\n",
       "      <th>Q047</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>43.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>RS</td>\n",
       "      <td>M</td>\n",
       "      <td>D</td>\n",
       "      <td>D</td>\n",
       "      <td>C</td>\n",
       "      <td>A</td>\n",
       "      <td>A</td>\n",
       "      <td>C</td>\n",
       "      <td>H</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>23.0</td>\n",
       "      <td>17.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>CE</td>\n",
       "      <td>F</td>\n",
       "      <td>A</td>\n",
       "      <td>A</td>\n",
       "      <td>B</td>\n",
       "      <td>A</td>\n",
       "      <td>A</td>\n",
       "      <td>A</td>\n",
       "      <td>NA</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>23.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>CE</td>\n",
       "      <td>F</td>\n",
       "      <td>D</td>\n",
       "      <td>D</td>\n",
       "      <td>C</td>\n",
       "      <td>A</td>\n",
       "      <td>A</td>\n",
       "      <td>A</td>\n",
       "      <td>NA</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>33.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>RJ</td>\n",
       "      <td>F</td>\n",
       "      <td>H</td>\n",
       "      <td>E</td>\n",
       "      <td>E</td>\n",
       "      <td>C</td>\n",
       "      <td>B</td>\n",
       "      <td>C</td>\n",
       "      <td>F</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>13.0</td>\n",
       "      <td>28.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>AM</td>\n",
       "      <td>M</td>\n",
       "      <td>E</td>\n",
       "      <td>D</td>\n",
       "      <td>C</td>\n",
       "      <td>A</td>\n",
       "      <td>A</td>\n",
       "      <td>B</td>\n",
       "      <td>F</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 42 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   CO_UF_RESIDENCIA  NU_IDADE  TP_COR_RACA  TP_NACIONALIDADE  TP_ST_CONCLUSAO  \\\n",
       "0              43.0      24.0          1.0               1.0              1.0   \n",
       "1              23.0      17.0          3.0               1.0              2.0   \n",
       "2              23.0      21.0          3.0               1.0              3.0   \n",
       "3              33.0      25.0          0.0               1.0              1.0   \n",
       "4              13.0      28.0          2.0               1.0              1.0   \n",
       "\n",
       "   TP_ANO_CONCLUIU  TP_ESCOLA  TP_ENSINO  TP_DEPENDENCIA_ADM_ESC  \\\n",
       "0              4.0        1.0        1.0                     2.0   \n",
       "1              0.0        2.0        1.0                     2.0   \n",
       "2              0.0        1.0        1.0                     2.0   \n",
       "3              9.0        1.0        1.0                     2.0   \n",
       "4              4.0        1.0        1.0                     2.0   \n",
       "\n",
       "   IN_BAIXA_VISAO  ...  SG_UF_RESIDENCIA  TP_SEXO  Q001  Q002  Q006  Q024  \\\n",
       "0             0.0  ...                RS        M     D     D     C     A   \n",
       "1             0.0  ...                CE        F     A     A     B     A   \n",
       "2             0.0  ...                CE        F     D     D     C     A   \n",
       "3             0.0  ...                RJ        F     H     E     E     C   \n",
       "4             0.0  ...                AM        M     E     D     C     A   \n",
       "\n",
       "   Q025  Q026  Q027  Q047  \n",
       "0     A     C     H     A  \n",
       "1     A     A    NA     A  \n",
       "2     A     A    NA     A  \n",
       "3     B     C     F     D  \n",
       "4     A     B     F     A  \n",
       "\n",
       "[5 rows x 42 columns]"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "def preprocess_transform(df, processors):\n",
    "    num_df = df[numeric_features]\n",
    "    cat_df = df[cat_features]\n",
    "    num_processed = processors['num_imputer'].transform(num_df)\n",
    "    #num_processed = processors['scaler'].transform(num_processed)\n",
    "    num_df = pd.DataFrame(num_processed, columns=num_df.columns)\n",
    "    cat_processed = processors['cat_imputer'].transform(cat_df)\n",
    "    cat_df = pd.DataFrame(cat_processed, columns=cat_df.columns)\n",
    "    return pd.concat((num_df, cat_df), axis=1)\n",
    "    \n",
    "test_df = preprocess_transform(test_df, processors)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "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>CO_UF_RESIDENCIA</th>\n",
       "      <th>NU_IDADE</th>\n",
       "      <th>TP_COR_RACA</th>\n",
       "      <th>TP_NACIONALIDADE</th>\n",
       "      <th>TP_ST_CONCLUSAO</th>\n",
       "      <th>TP_ANO_CONCLUIU</th>\n",
       "      <th>TP_ESCOLA</th>\n",
       "      <th>TP_ENSINO</th>\n",
       "      <th>TP_DEPENDENCIA_ADM_ESC</th>\n",
       "      <th>IN_BAIXA_VISAO</th>\n",
       "      <th>...</th>\n",
       "      <th>SG_UF_RESIDENCIA</th>\n",
       "      <th>TP_SEXO</th>\n",
       "      <th>Q001</th>\n",
       "      <th>Q002</th>\n",
       "      <th>Q006</th>\n",
       "      <th>Q024</th>\n",
       "      <th>Q025</th>\n",
       "      <th>Q026</th>\n",
       "      <th>Q027</th>\n",
       "      <th>Q047</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>43.0</td>\n",
       "      <td>19.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>RS</td>\n",
       "      <td>F</td>\n",
       "      <td>E</td>\n",
       "      <td>E</td>\n",
       "      <td>H</td>\n",
       "      <td>B</td>\n",
       "      <td>B</td>\n",
       "      <td>C</td>\n",
       "      <td>D</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>15.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>PA</td>\n",
       "      <td>M</td>\n",
       "      <td>B</td>\n",
       "      <td>C</td>\n",
       "      <td>B</td>\n",
       "      <td>A</td>\n",
       "      <td>A</td>\n",
       "      <td>C</td>\n",
       "      <td>D</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>29.0</td>\n",
       "      <td>16.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>BA</td>\n",
       "      <td>F</td>\n",
       "      <td>E</td>\n",
       "      <td>F</td>\n",
       "      <td>G</td>\n",
       "      <td>B</td>\n",
       "      <td>B</td>\n",
       "      <td>A</td>\n",
       "      <td>NA</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>41.0</td>\n",
       "      <td>17.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>PR</td>\n",
       "      <td>F</td>\n",
       "      <td>E</td>\n",
       "      <td>E</td>\n",
       "      <td>E</td>\n",
       "      <td>C</td>\n",
       "      <td>B</td>\n",
       "      <td>B</td>\n",
       "      <td>C</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>53.0</td>\n",
       "      <td>19.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>DF</td>\n",
       "      <td>F</td>\n",
       "      <td>E</td>\n",
       "      <td>E</td>\n",
       "      <td>E</td>\n",
       "      <td>B</td>\n",
       "      <td>B</td>\n",
       "      <td>B</td>\n",
       "      <td>D</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 42 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   CO_UF_RESIDENCIA  NU_IDADE  TP_COR_RACA  TP_NACIONALIDADE  TP_ST_CONCLUSAO  \\\n",
       "0              43.0      19.0          1.0               1.0              1.0   \n",
       "1              15.0      24.0          3.0               2.0              1.0   \n",
       "2              29.0      16.0          2.0               1.0              3.0   \n",
       "3              41.0      17.0          1.0               1.0              2.0   \n",
       "4              53.0      19.0          1.0               1.0              1.0   \n",
       "\n",
       "   TP_ANO_CONCLUIU  TP_ESCOLA  TP_ENSINO  TP_DEPENDENCIA_ADM_ESC  \\\n",
       "0              3.0        1.0        1.0                     2.0   \n",
       "1              4.0        1.0        1.0                     2.0   \n",
       "2              0.0        1.0        1.0                     2.0   \n",
       "3              0.0        2.0        1.0                     2.0   \n",
       "4              1.0        1.0        1.0                     2.0   \n",
       "\n",
       "   IN_BAIXA_VISAO  ...  SG_UF_RESIDENCIA  TP_SEXO  Q001  Q002  Q006  Q024  \\\n",
       "0             0.0  ...                RS        F     E     E     H     B   \n",
       "1             0.0  ...                PA        M     B     C     B     A   \n",
       "2             0.0  ...                BA        F     E     F     G     B   \n",
       "3             0.0  ...                PR        F     E     E     E     C   \n",
       "4             0.0  ...                DF        F     E     E     E     B   \n",
       "\n",
       "   Q025  Q026  Q027  Q047  \n",
       "0     B     C     D     A  \n",
       "1     A     C     D     A  \n",
       "2     B     A    NA     D  \n",
       "3     B     B     C     A  \n",
       "4     B     B     D     A  \n",
       "\n",
       "[5 rows x 42 columns]"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "cat = CatBoostClassifier(cat_features=cat_features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Learning rate set to 0.03153\n",
      "0:\tlearn: 0.5873118\ttotal: 31.7ms\tremaining: 31.6s\n",
      "1:\tlearn: 0.4965702\ttotal: 61.3ms\tremaining: 30.6s\n",
      "2:\tlearn: 0.4237562\ttotal: 90.7ms\tremaining: 30.2s\n",
      "3:\tlearn: 0.3592273\ttotal: 109ms\tremaining: 27.2s\n",
      "4:\tlearn: 0.3054241\ttotal: 136ms\tremaining: 27s\n",
      "5:\tlearn: 0.2604828\ttotal: 156ms\tremaining: 25.8s\n",
      "6:\tlearn: 0.2229919\ttotal: 183ms\tremaining: 25.9s\n",
      "7:\tlearn: 0.1857374\ttotal: 210ms\tremaining: 26s\n",
      "8:\tlearn: 0.1564399\ttotal: 238ms\tremaining: 26.2s\n",
      "9:\tlearn: 0.1338687\ttotal: 265ms\tremaining: 26.2s\n",
      "10:\tlearn: 0.1141476\ttotal: 289ms\tremaining: 25.9s\n",
      "11:\tlearn: 0.0979413\ttotal: 314ms\tremaining: 25.9s\n",
      "12:\tlearn: 0.0848893\ttotal: 340ms\tremaining: 25.8s\n",
      "13:\tlearn: 0.0737815\ttotal: 369ms\tremaining: 26s\n",
      "14:\tlearn: 0.0643629\ttotal: 396ms\tremaining: 26s\n",
      "15:\tlearn: 0.0565457\ttotal: 423ms\tremaining: 26s\n",
      "16:\tlearn: 0.0502358\ttotal: 446ms\tremaining: 25.8s\n",
      "17:\tlearn: 0.0447174\ttotal: 474ms\tremaining: 25.8s\n",
      "18:\tlearn: 0.0398219\ttotal: 504ms\tremaining: 26s\n",
      "19:\tlearn: 0.0361858\ttotal: 524ms\tremaining: 25.7s\n",
      "20:\tlearn: 0.0327446\ttotal: 542ms\tremaining: 25.3s\n",
      "21:\tlearn: 0.0298114\ttotal: 569ms\tremaining: 25.3s\n",
      "22:\tlearn: 0.0273564\ttotal: 586ms\tremaining: 24.9s\n",
      "23:\tlearn: 0.0255628\ttotal: 604ms\tremaining: 24.5s\n",
      "24:\tlearn: 0.0239462\ttotal: 626ms\tremaining: 24.4s\n",
      "25:\tlearn: 0.0221451\ttotal: 654ms\tremaining: 24.5s\n",
      "26:\tlearn: 0.0207803\ttotal: 681ms\tremaining: 24.5s\n",
      "27:\tlearn: 0.0193984\ttotal: 690ms\tremaining: 23.9s\n",
      "28:\tlearn: 0.0183650\ttotal: 707ms\tremaining: 23.7s\n",
      "29:\tlearn: 0.0172670\ttotal: 734ms\tremaining: 23.7s\n",
      "30:\tlearn: 0.0165295\ttotal: 753ms\tremaining: 23.5s\n",
      "31:\tlearn: 0.0157298\ttotal: 782ms\tremaining: 23.6s\n",
      "32:\tlearn: 0.0150068\ttotal: 802ms\tremaining: 23.5s\n",
      "33:\tlearn: 0.0144779\ttotal: 829ms\tremaining: 23.6s\n",
      "34:\tlearn: 0.0143966\ttotal: 840ms\tremaining: 23.1s\n",
      "35:\tlearn: 0.0138812\ttotal: 867ms\tremaining: 23.2s\n",
      "36:\tlearn: 0.0133627\ttotal: 892ms\tremaining: 23.2s\n",
      "37:\tlearn: 0.0129558\ttotal: 918ms\tremaining: 23.2s\n",
      "38:\tlearn: 0.0129008\ttotal: 923ms\tremaining: 22.8s\n",
      "39:\tlearn: 0.0125441\ttotal: 949ms\tremaining: 22.8s\n",
      "40:\tlearn: 0.0121270\ttotal: 970ms\tremaining: 22.7s\n",
      "41:\tlearn: 0.0118065\ttotal: 999ms\tremaining: 22.8s\n",
      "42:\tlearn: 0.0117714\ttotal: 1.04s\tremaining: 23.1s\n",
      "43:\tlearn: 0.0117237\ttotal: 1.07s\tremaining: 23.2s\n",
      "44:\tlearn: 0.0114149\ttotal: 1.1s\tremaining: 23.3s\n",
      "45:\tlearn: 0.0111953\ttotal: 1.13s\tremaining: 23.5s\n",
      "46:\tlearn: 0.0111411\ttotal: 1.16s\tremaining: 23.5s\n",
      "47:\tlearn: 0.0108739\ttotal: 1.19s\tremaining: 23.6s\n",
      "48:\tlearn: 0.0106899\ttotal: 1.22s\tremaining: 23.7s\n",
      "49:\tlearn: 0.0106333\ttotal: 1.25s\tremaining: 23.7s\n",
      "50:\tlearn: 0.0105930\ttotal: 1.27s\tremaining: 23.6s\n",
      "51:\tlearn: 0.0105672\ttotal: 1.3s\tremaining: 23.6s\n",
      "52:\tlearn: 0.0105237\ttotal: 1.33s\tremaining: 23.7s\n",
      "53:\tlearn: 0.0103693\ttotal: 1.36s\tremaining: 23.9s\n",
      "54:\tlearn: 0.0103558\ttotal: 1.37s\tremaining: 23.5s\n",
      "55:\tlearn: 0.0103266\ttotal: 1.39s\tremaining: 23.5s\n",
      "56:\tlearn: 0.0102859\ttotal: 1.42s\tremaining: 23.4s\n",
      "57:\tlearn: 0.0102751\ttotal: 1.43s\tremaining: 23.2s\n",
      "58:\tlearn: 0.0102650\ttotal: 1.45s\tremaining: 23.1s\n",
      "59:\tlearn: 0.0102333\ttotal: 1.47s\tremaining: 23s\n",
      "60:\tlearn: 0.0102245\ttotal: 1.47s\tremaining: 22.7s\n",
      "61:\tlearn: 0.0100345\ttotal: 1.5s\tremaining: 22.7s\n",
      "62:\tlearn: 0.0100062\ttotal: 1.52s\tremaining: 22.7s\n",
      "63:\tlearn: 0.0098885\ttotal: 1.55s\tremaining: 22.7s\n",
      "64:\tlearn: 0.0098134\ttotal: 1.58s\tremaining: 22.7s\n",
      "65:\tlearn: 0.0098079\ttotal: 1.58s\tremaining: 22.4s\n",
      "66:\tlearn: 0.0097148\ttotal: 1.61s\tremaining: 22.4s\n",
      "67:\tlearn: 0.0097102\ttotal: 1.61s\tremaining: 22.1s\n",
      "68:\tlearn: 0.0096886\ttotal: 1.64s\tremaining: 22.1s\n",
      "69:\tlearn: 0.0096770\ttotal: 1.67s\tremaining: 22.2s\n",
      "70:\tlearn: 0.0095099\ttotal: 1.7s\tremaining: 22.2s\n",
      "71:\tlearn: 0.0094936\ttotal: 1.73s\tremaining: 22.3s\n",
      "72:\tlearn: 0.0094818\ttotal: 1.75s\tremaining: 22.2s\n",
      "73:\tlearn: 0.0093827\ttotal: 1.77s\tremaining: 22.2s\n",
      "74:\tlearn: 0.0093552\ttotal: 1.8s\tremaining: 22.2s\n",
      "75:\tlearn: 0.0093436\ttotal: 1.84s\tremaining: 22.3s\n",
      "76:\tlearn: 0.0092722\ttotal: 1.87s\tremaining: 22.4s\n",
      "77:\tlearn: 0.0092518\ttotal: 1.9s\tremaining: 22.4s\n",
      "78:\tlearn: 0.0092403\ttotal: 1.92s\tremaining: 22.4s\n",
      "79:\tlearn: 0.0091632\ttotal: 1.94s\tremaining: 22.3s\n",
      "80:\tlearn: 0.0091215\ttotal: 1.97s\tremaining: 22.4s\n",
      "81:\tlearn: 0.0091074\ttotal: 2s\tremaining: 22.4s\n",
      "82:\tlearn: 0.0091007\ttotal: 2.02s\tremaining: 22.3s\n",
      "83:\tlearn: 0.0090860\ttotal: 2.04s\tremaining: 22.3s\n",
      "84:\tlearn: 0.0090031\ttotal: 2.07s\tremaining: 22.3s\n",
      "85:\tlearn: 0.0089390\ttotal: 2.1s\tremaining: 22.3s\n",
      "86:\tlearn: 0.0089264\ttotal: 2.12s\tremaining: 22.3s\n",
      "87:\tlearn: 0.0089219\ttotal: 2.15s\tremaining: 22.3s\n",
      "88:\tlearn: 0.0088472\ttotal: 2.16s\tremaining: 22.1s\n",
      "89:\tlearn: 0.0088402\ttotal: 2.19s\tremaining: 22.1s\n",
      "90:\tlearn: 0.0088255\ttotal: 2.21s\tremaining: 22.1s\n",
      "91:\tlearn: 0.0088208\ttotal: 2.23s\tremaining: 22s\n",
      "92:\tlearn: 0.0088126\ttotal: 2.26s\tremaining: 22s\n",
      "93:\tlearn: 0.0088052\ttotal: 2.29s\tremaining: 22.1s\n",
      "94:\tlearn: 0.0087826\ttotal: 2.32s\tremaining: 22.1s\n",
      "95:\tlearn: 0.0087675\ttotal: 2.36s\tremaining: 22.2s\n",
      "96:\tlearn: 0.0087607\ttotal: 2.38s\tremaining: 22.1s\n",
      "97:\tlearn: 0.0087564\ttotal: 2.4s\tremaining: 22.1s\n",
      "98:\tlearn: 0.0087391\ttotal: 2.42s\tremaining: 22.1s\n",
      "99:\tlearn: 0.0087060\ttotal: 2.45s\tremaining: 22.1s\n",
      "100:\tlearn: 0.0086915\ttotal: 2.47s\tremaining: 22s\n",
      "101:\tlearn: 0.0086361\ttotal: 2.5s\tremaining: 22s\n",
      "102:\tlearn: 0.0085699\ttotal: 2.52s\tremaining: 22s\n",
      "103:\tlearn: 0.0085655\ttotal: 2.54s\tremaining: 21.9s\n",
      "104:\tlearn: 0.0085591\ttotal: 2.56s\tremaining: 21.8s\n",
      "105:\tlearn: 0.0085509\ttotal: 2.58s\tremaining: 21.7s\n",
      "106:\tlearn: 0.0085449\ttotal: 2.6s\tremaining: 21.7s\n",
      "107:\tlearn: 0.0085183\ttotal: 2.64s\tremaining: 21.8s\n",
      "108:\tlearn: 0.0085028\ttotal: 2.67s\tremaining: 21.8s\n",
      "109:\tlearn: 0.0084279\ttotal: 2.7s\tremaining: 21.8s\n",
      "110:\tlearn: 0.0084233\ttotal: 2.72s\tremaining: 21.8s\n",
      "111:\tlearn: 0.0084110\ttotal: 2.75s\tremaining: 21.8s\n",
      "112:\tlearn: 0.0083588\ttotal: 2.78s\tremaining: 21.8s\n",
      "113:\tlearn: 0.0083528\ttotal: 2.8s\tremaining: 21.8s\n",
      "114:\tlearn: 0.0083397\ttotal: 2.83s\tremaining: 21.8s\n",
      "115:\tlearn: 0.0083270\ttotal: 2.85s\tremaining: 21.8s\n",
      "116:\tlearn: 0.0083102\ttotal: 2.88s\tremaining: 21.8s\n",
      "117:\tlearn: 0.0082704\ttotal: 2.91s\tremaining: 21.8s\n",
      "118:\tlearn: 0.0082695\ttotal: 2.94s\tremaining: 21.8s\n",
      "119:\tlearn: 0.0082495\ttotal: 2.96s\tremaining: 21.7s\n",
      "120:\tlearn: 0.0082202\ttotal: 2.99s\tremaining: 21.7s\n",
      "121:\tlearn: 0.0082031\ttotal: 3.02s\tremaining: 21.7s\n",
      "122:\tlearn: 0.0081976\ttotal: 3.04s\tremaining: 21.7s\n",
      "123:\tlearn: 0.0081935\ttotal: 3.07s\tremaining: 21.7s\n",
      "124:\tlearn: 0.0081601\ttotal: 3.09s\tremaining: 21.7s\n",
      "125:\tlearn: 0.0081477\ttotal: 3.12s\tremaining: 21.7s\n",
      "126:\tlearn: 0.0081354\ttotal: 3.15s\tremaining: 21.6s\n",
      "127:\tlearn: 0.0081226\ttotal: 3.17s\tremaining: 21.6s\n",
      "128:\tlearn: 0.0081153\ttotal: 3.19s\tremaining: 21.6s\n",
      "129:\tlearn: 0.0080879\ttotal: 3.22s\tremaining: 21.6s\n",
      "130:\tlearn: 0.0080772\ttotal: 3.24s\tremaining: 21.5s\n",
      "131:\tlearn: 0.0080388\ttotal: 3.27s\tremaining: 21.5s\n",
      "132:\tlearn: 0.0080304\ttotal: 3.3s\tremaining: 21.5s\n",
      "133:\tlearn: 0.0080290\ttotal: 3.31s\tremaining: 21.4s\n",
      "134:\tlearn: 0.0080054\ttotal: 3.34s\tremaining: 21.4s\n",
      "135:\tlearn: 0.0080043\ttotal: 3.35s\tremaining: 21.3s\n",
      "136:\tlearn: 0.0080023\ttotal: 3.38s\tremaining: 21.3s\n",
      "137:\tlearn: 0.0079970\ttotal: 3.39s\tremaining: 21.2s\n",
      "138:\tlearn: 0.0079787\ttotal: 3.41s\tremaining: 21.1s\n",
      "139:\tlearn: 0.0079459\ttotal: 3.44s\tremaining: 21.1s\n",
      "140:\tlearn: 0.0079438\ttotal: 3.46s\tremaining: 21.1s\n",
      "141:\tlearn: 0.0079401\ttotal: 3.48s\tremaining: 21s\n",
      "142:\tlearn: 0.0079287\ttotal: 3.51s\tremaining: 21s\n",
      "143:\tlearn: 0.0079235\ttotal: 3.53s\tremaining: 21s\n",
      "144:\tlearn: 0.0079208\ttotal: 3.57s\tremaining: 21s\n",
      "145:\tlearn: 0.0078973\ttotal: 3.59s\tremaining: 21s\n",
      "146:\tlearn: 0.0078877\ttotal: 3.62s\tremaining: 21s\n",
      "147:\tlearn: 0.0078680\ttotal: 3.65s\tremaining: 21s\n",
      "148:\tlearn: 0.0078656\ttotal: 3.66s\tremaining: 20.9s\n",
      "149:\tlearn: 0.0078514\ttotal: 3.69s\tremaining: 20.9s\n",
      "150:\tlearn: 0.0078452\ttotal: 3.72s\tremaining: 20.9s\n",
      "151:\tlearn: 0.0078417\ttotal: 3.73s\tremaining: 20.8s\n",
      "152:\tlearn: 0.0078385\ttotal: 3.75s\tremaining: 20.8s\n",
      "153:\tlearn: 0.0078348\ttotal: 3.77s\tremaining: 20.7s\n",
      "154:\tlearn: 0.0078292\ttotal: 3.79s\tremaining: 20.7s\n",
      "155:\tlearn: 0.0078147\ttotal: 3.82s\tremaining: 20.7s\n",
      "156:\tlearn: 0.0078017\ttotal: 3.85s\tremaining: 20.7s\n",
      "157:\tlearn: 0.0077954\ttotal: 3.88s\tremaining: 20.7s\n",
      "158:\tlearn: 0.0077738\ttotal: 3.91s\tremaining: 20.7s\n",
      "159:\tlearn: 0.0077703\ttotal: 3.93s\tremaining: 20.6s\n",
      "160:\tlearn: 0.0077611\ttotal: 3.95s\tremaining: 20.6s\n",
      "161:\tlearn: 0.0077557\ttotal: 3.97s\tremaining: 20.5s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "162:\tlearn: 0.0077221\ttotal: 4s\tremaining: 20.5s\n",
      "163:\tlearn: 0.0077118\ttotal: 4.02s\tremaining: 20.5s\n",
      "164:\tlearn: 0.0076915\ttotal: 4.05s\tremaining: 20.5s\n",
      "165:\tlearn: 0.0076857\ttotal: 4.07s\tremaining: 20.4s\n",
      "166:\tlearn: 0.0076215\ttotal: 4.09s\tremaining: 20.4s\n",
      "167:\tlearn: 0.0076157\ttotal: 4.12s\tremaining: 20.4s\n",
      "168:\tlearn: 0.0076084\ttotal: 4.14s\tremaining: 20.4s\n",
      "169:\tlearn: 0.0076016\ttotal: 4.16s\tremaining: 20.3s\n",
      "170:\tlearn: 0.0075748\ttotal: 4.19s\tremaining: 20.3s\n",
      "171:\tlearn: 0.0075692\ttotal: 4.22s\tremaining: 20.3s\n",
      "172:\tlearn: 0.0075361\ttotal: 4.26s\tremaining: 20.3s\n",
      "173:\tlearn: 0.0075050\ttotal: 4.29s\tremaining: 20.4s\n",
      "174:\tlearn: 0.0075033\ttotal: 4.32s\tremaining: 20.3s\n",
      "175:\tlearn: 0.0074988\ttotal: 4.34s\tremaining: 20.3s\n",
      "176:\tlearn: 0.0074977\ttotal: 4.36s\tremaining: 20.3s\n",
      "177:\tlearn: 0.0074810\ttotal: 4.39s\tremaining: 20.3s\n",
      "178:\tlearn: 0.0074720\ttotal: 4.42s\tremaining: 20.3s\n",
      "179:\tlearn: 0.0074657\ttotal: 4.45s\tremaining: 20.3s\n",
      "180:\tlearn: 0.0074639\ttotal: 4.48s\tremaining: 20.3s\n",
      "181:\tlearn: 0.0074585\ttotal: 4.5s\tremaining: 20.2s\n",
      "182:\tlearn: 0.0074391\ttotal: 4.53s\tremaining: 20.2s\n",
      "183:\tlearn: 0.0074161\ttotal: 4.56s\tremaining: 20.2s\n",
      "184:\tlearn: 0.0073822\ttotal: 4.58s\tremaining: 20.2s\n",
      "185:\tlearn: 0.0073724\ttotal: 4.61s\tremaining: 20.2s\n",
      "186:\tlearn: 0.0073632\ttotal: 4.63s\tremaining: 20.1s\n",
      "187:\tlearn: 0.0073232\ttotal: 4.66s\tremaining: 20.1s\n",
      "188:\tlearn: 0.0073196\ttotal: 4.68s\tremaining: 20.1s\n",
      "189:\tlearn: 0.0073175\ttotal: 4.7s\tremaining: 20s\n",
      "190:\tlearn: 0.0073027\ttotal: 4.73s\tremaining: 20s\n",
      "191:\tlearn: 0.0072832\ttotal: 4.76s\tremaining: 20s\n",
      "192:\tlearn: 0.0072631\ttotal: 4.79s\tremaining: 20s\n",
      "193:\tlearn: 0.0072505\ttotal: 4.81s\tremaining: 20s\n",
      "194:\tlearn: 0.0072474\ttotal: 4.83s\tremaining: 19.9s\n",
      "195:\tlearn: 0.0072328\ttotal: 4.86s\tremaining: 19.9s\n",
      "196:\tlearn: 0.0072208\ttotal: 4.89s\tremaining: 19.9s\n",
      "197:\tlearn: 0.0071829\ttotal: 4.92s\tremaining: 19.9s\n",
      "198:\tlearn: 0.0071791\ttotal: 4.93s\tremaining: 19.9s\n",
      "199:\tlearn: 0.0071544\ttotal: 4.96s\tremaining: 19.8s\n",
      "200:\tlearn: 0.0071342\ttotal: 4.99s\tremaining: 19.8s\n",
      "201:\tlearn: 0.0071169\ttotal: 5.02s\tremaining: 19.8s\n",
      "202:\tlearn: 0.0071144\ttotal: 5.06s\tremaining: 19.9s\n",
      "203:\tlearn: 0.0071128\ttotal: 5.08s\tremaining: 19.8s\n",
      "204:\tlearn: 0.0071072\ttotal: 5.11s\tremaining: 19.8s\n",
      "205:\tlearn: 0.0070957\ttotal: 5.13s\tremaining: 19.8s\n",
      "206:\tlearn: 0.0070880\ttotal: 5.16s\tremaining: 19.8s\n",
      "207:\tlearn: 0.0070771\ttotal: 5.19s\tremaining: 19.8s\n",
      "208:\tlearn: 0.0070683\ttotal: 5.22s\tremaining: 19.8s\n",
      "209:\tlearn: 0.0070616\ttotal: 5.25s\tremaining: 19.8s\n",
      "210:\tlearn: 0.0070566\ttotal: 5.28s\tremaining: 19.8s\n",
      "211:\tlearn: 0.0070448\ttotal: 5.31s\tremaining: 19.7s\n",
      "212:\tlearn: 0.0070414\ttotal: 5.33s\tremaining: 19.7s\n",
      "213:\tlearn: 0.0070350\ttotal: 5.36s\tremaining: 19.7s\n",
      "214:\tlearn: 0.0070205\ttotal: 5.39s\tremaining: 19.7s\n",
      "215:\tlearn: 0.0070144\ttotal: 5.42s\tremaining: 19.7s\n",
      "216:\tlearn: 0.0070109\ttotal: 5.43s\tremaining: 19.6s\n",
      "217:\tlearn: 0.0070079\ttotal: 5.45s\tremaining: 19.6s\n",
      "218:\tlearn: 0.0069989\ttotal: 5.48s\tremaining: 19.5s\n",
      "219:\tlearn: 0.0069917\ttotal: 5.51s\tremaining: 19.5s\n",
      "220:\tlearn: 0.0069729\ttotal: 5.54s\tremaining: 19.5s\n",
      "221:\tlearn: 0.0069729\ttotal: 5.54s\tremaining: 19.4s\n",
      "222:\tlearn: 0.0069701\ttotal: 5.58s\tremaining: 19.4s\n",
      "223:\tlearn: 0.0069641\ttotal: 5.61s\tremaining: 19.4s\n",
      "224:\tlearn: 0.0069628\ttotal: 5.63s\tremaining: 19.4s\n",
      "225:\tlearn: 0.0069611\ttotal: 5.67s\tremaining: 19.4s\n",
      "226:\tlearn: 0.0069554\ttotal: 5.69s\tremaining: 19.4s\n",
      "227:\tlearn: 0.0069537\ttotal: 5.71s\tremaining: 19.3s\n",
      "228:\tlearn: 0.0069483\ttotal: 5.74s\tremaining: 19.3s\n",
      "229:\tlearn: 0.0069386\ttotal: 5.76s\tremaining: 19.3s\n",
      "230:\tlearn: 0.0069230\ttotal: 5.8s\tremaining: 19.3s\n",
      "231:\tlearn: 0.0069094\ttotal: 5.83s\tremaining: 19.3s\n",
      "232:\tlearn: 0.0069048\ttotal: 5.86s\tremaining: 19.3s\n",
      "233:\tlearn: 0.0069048\ttotal: 5.86s\tremaining: 19.2s\n",
      "234:\tlearn: 0.0069018\ttotal: 5.89s\tremaining: 19.2s\n",
      "235:\tlearn: 0.0068972\ttotal: 5.91s\tremaining: 19.1s\n",
      "236:\tlearn: 0.0068780\ttotal: 5.95s\tremaining: 19.2s\n",
      "237:\tlearn: 0.0068780\ttotal: 5.96s\tremaining: 19.1s\n",
      "238:\tlearn: 0.0068562\ttotal: 6s\tremaining: 19.1s\n",
      "239:\tlearn: 0.0068440\ttotal: 6.04s\tremaining: 19.1s\n",
      "240:\tlearn: 0.0068325\ttotal: 6.06s\tremaining: 19.1s\n",
      "241:\tlearn: 0.0068181\ttotal: 6.09s\tremaining: 19.1s\n",
      "242:\tlearn: 0.0068149\ttotal: 6.11s\tremaining: 19s\n",
      "243:\tlearn: 0.0068105\ttotal: 6.13s\tremaining: 19s\n",
      "244:\tlearn: 0.0067995\ttotal: 6.17s\tremaining: 19s\n",
      "245:\tlearn: 0.0067871\ttotal: 6.2s\tremaining: 19s\n",
      "246:\tlearn: 0.0067654\ttotal: 6.24s\tremaining: 19s\n",
      "247:\tlearn: 0.0067643\ttotal: 6.26s\tremaining: 19s\n",
      "248:\tlearn: 0.0067546\ttotal: 6.29s\tremaining: 19s\n",
      "249:\tlearn: 0.0067466\ttotal: 6.32s\tremaining: 19s\n",
      "250:\tlearn: 0.0067342\ttotal: 6.34s\tremaining: 18.9s\n",
      "251:\tlearn: 0.0067326\ttotal: 6.37s\tremaining: 18.9s\n",
      "252:\tlearn: 0.0067228\ttotal: 6.39s\tremaining: 18.9s\n",
      "253:\tlearn: 0.0067164\ttotal: 6.42s\tremaining: 18.9s\n",
      "254:\tlearn: 0.0067137\ttotal: 6.45s\tremaining: 18.8s\n",
      "255:\tlearn: 0.0067063\ttotal: 6.47s\tremaining: 18.8s\n",
      "256:\tlearn: 0.0067044\ttotal: 6.5s\tremaining: 18.8s\n",
      "257:\tlearn: 0.0066682\ttotal: 6.57s\tremaining: 18.9s\n",
      "258:\tlearn: 0.0066518\ttotal: 6.61s\tremaining: 18.9s\n",
      "259:\tlearn: 0.0066406\ttotal: 6.65s\tremaining: 18.9s\n",
      "260:\tlearn: 0.0066348\ttotal: 6.69s\tremaining: 18.9s\n",
      "261:\tlearn: 0.0066115\ttotal: 6.72s\tremaining: 18.9s\n",
      "262:\tlearn: 0.0066081\ttotal: 6.73s\tremaining: 18.9s\n",
      "263:\tlearn: 0.0065928\ttotal: 6.76s\tremaining: 18.9s\n",
      "264:\tlearn: 0.0065928\ttotal: 6.77s\tremaining: 18.8s\n",
      "265:\tlearn: 0.0065798\ttotal: 6.8s\tremaining: 18.8s\n",
      "266:\tlearn: 0.0065765\ttotal: 6.82s\tremaining: 18.7s\n",
      "267:\tlearn: 0.0065610\ttotal: 6.84s\tremaining: 18.7s\n",
      "268:\tlearn: 0.0065504\ttotal: 6.87s\tremaining: 18.7s\n",
      "269:\tlearn: 0.0065479\ttotal: 6.9s\tremaining: 18.7s\n",
      "270:\tlearn: 0.0065384\ttotal: 6.93s\tremaining: 18.6s\n",
      "271:\tlearn: 0.0065176\ttotal: 6.96s\tremaining: 18.6s\n",
      "272:\tlearn: 0.0065125\ttotal: 7s\tremaining: 18.6s\n",
      "273:\tlearn: 0.0065108\ttotal: 7.02s\tremaining: 18.6s\n",
      "274:\tlearn: 0.0065063\ttotal: 7.05s\tremaining: 18.6s\n",
      "275:\tlearn: 0.0065047\ttotal: 7.07s\tremaining: 18.6s\n",
      "276:\tlearn: 0.0064796\ttotal: 7.1s\tremaining: 18.5s\n",
      "277:\tlearn: 0.0064779\ttotal: 7.12s\tremaining: 18.5s\n",
      "278:\tlearn: 0.0064740\ttotal: 7.16s\tremaining: 18.5s\n",
      "279:\tlearn: 0.0064731\ttotal: 7.19s\tremaining: 18.5s\n",
      "280:\tlearn: 0.0064543\ttotal: 7.22s\tremaining: 18.5s\n",
      "281:\tlearn: 0.0064525\ttotal: 7.26s\tremaining: 18.5s\n",
      "282:\tlearn: 0.0064407\ttotal: 7.29s\tremaining: 18.5s\n",
      "283:\tlearn: 0.0064140\ttotal: 7.32s\tremaining: 18.5s\n",
      "284:\tlearn: 0.0063998\ttotal: 7.35s\tremaining: 18.4s\n",
      "285:\tlearn: 0.0063951\ttotal: 7.39s\tremaining: 18.4s\n",
      "286:\tlearn: 0.0063901\ttotal: 7.41s\tremaining: 18.4s\n",
      "287:\tlearn: 0.0063816\ttotal: 7.44s\tremaining: 18.4s\n",
      "288:\tlearn: 0.0063584\ttotal: 7.46s\tremaining: 18.4s\n",
      "289:\tlearn: 0.0063564\ttotal: 7.5s\tremaining: 18.4s\n",
      "290:\tlearn: 0.0063534\ttotal: 7.52s\tremaining: 18.3s\n",
      "291:\tlearn: 0.0063512\ttotal: 7.55s\tremaining: 18.3s\n",
      "292:\tlearn: 0.0063491\ttotal: 7.57s\tremaining: 18.3s\n",
      "293:\tlearn: 0.0063369\ttotal: 7.6s\tremaining: 18.2s\n",
      "294:\tlearn: 0.0063273\ttotal: 7.62s\tremaining: 18.2s\n",
      "295:\tlearn: 0.0063221\ttotal: 7.65s\tremaining: 18.2s\n",
      "296:\tlearn: 0.0063175\ttotal: 7.68s\tremaining: 18.2s\n",
      "297:\tlearn: 0.0063116\ttotal: 7.7s\tremaining: 18.1s\n",
      "298:\tlearn: 0.0063115\ttotal: 7.71s\tremaining: 18.1s\n",
      "299:\tlearn: 0.0063030\ttotal: 7.74s\tremaining: 18s\n",
      "300:\tlearn: 0.0062905\ttotal: 7.76s\tremaining: 18s\n",
      "301:\tlearn: 0.0062740\ttotal: 7.79s\tremaining: 18s\n",
      "302:\tlearn: 0.0062730\ttotal: 7.81s\tremaining: 18s\n",
      "303:\tlearn: 0.0062636\ttotal: 7.83s\tremaining: 17.9s\n",
      "304:\tlearn: 0.0062501\ttotal: 7.86s\tremaining: 17.9s\n",
      "305:\tlearn: 0.0062493\ttotal: 7.89s\tremaining: 17.9s\n",
      "306:\tlearn: 0.0062493\ttotal: 7.91s\tremaining: 17.9s\n",
      "307:\tlearn: 0.0062436\ttotal: 7.94s\tremaining: 17.8s\n",
      "308:\tlearn: 0.0062260\ttotal: 7.97s\tremaining: 17.8s\n",
      "309:\tlearn: 0.0062197\ttotal: 8s\tremaining: 17.8s\n",
      "310:\tlearn: 0.0062184\ttotal: 8.03s\tremaining: 17.8s\n",
      "311:\tlearn: 0.0062116\ttotal: 8.05s\tremaining: 17.8s\n",
      "312:\tlearn: 0.0061936\ttotal: 8.1s\tremaining: 17.8s\n",
      "313:\tlearn: 0.0061905\ttotal: 8.13s\tremaining: 17.8s\n",
      "314:\tlearn: 0.0061823\ttotal: 8.15s\tremaining: 17.7s\n",
      "315:\tlearn: 0.0061745\ttotal: 8.18s\tremaining: 17.7s\n",
      "316:\tlearn: 0.0061678\ttotal: 8.21s\tremaining: 17.7s\n",
      "317:\tlearn: 0.0061621\ttotal: 8.24s\tremaining: 17.7s\n",
      "318:\tlearn: 0.0061591\ttotal: 8.27s\tremaining: 17.6s\n",
      "319:\tlearn: 0.0061517\ttotal: 8.29s\tremaining: 17.6s\n",
      "320:\tlearn: 0.0061417\ttotal: 8.31s\tremaining: 17.6s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "321:\tlearn: 0.0061320\ttotal: 8.34s\tremaining: 17.6s\n",
      "322:\tlearn: 0.0061297\ttotal: 8.36s\tremaining: 17.5s\n",
      "323:\tlearn: 0.0061085\ttotal: 8.38s\tremaining: 17.5s\n",
      "324:\tlearn: 0.0061058\ttotal: 8.4s\tremaining: 17.4s\n",
      "325:\tlearn: 0.0061035\ttotal: 8.43s\tremaining: 17.4s\n",
      "326:\tlearn: 0.0061026\ttotal: 8.44s\tremaining: 17.4s\n",
      "327:\tlearn: 0.0060608\ttotal: 8.48s\tremaining: 17.4s\n",
      "328:\tlearn: 0.0060516\ttotal: 8.51s\tremaining: 17.4s\n",
      "329:\tlearn: 0.0060319\ttotal: 8.54s\tremaining: 17.3s\n",
      "330:\tlearn: 0.0060231\ttotal: 8.57s\tremaining: 17.3s\n",
      "331:\tlearn: 0.0060220\ttotal: 8.58s\tremaining: 17.3s\n",
      "332:\tlearn: 0.0059944\ttotal: 8.61s\tremaining: 17.2s\n",
      "333:\tlearn: 0.0059763\ttotal: 8.64s\tremaining: 17.2s\n",
      "334:\tlearn: 0.0059675\ttotal: 8.66s\tremaining: 17.2s\n",
      "335:\tlearn: 0.0059663\ttotal: 8.69s\tremaining: 17.2s\n",
      "336:\tlearn: 0.0059663\ttotal: 8.7s\tremaining: 17.1s\n",
      "337:\tlearn: 0.0059530\ttotal: 8.74s\tremaining: 17.1s\n",
      "338:\tlearn: 0.0059530\ttotal: 8.75s\tremaining: 17.1s\n",
      "339:\tlearn: 0.0059449\ttotal: 8.78s\tremaining: 17s\n",
      "340:\tlearn: 0.0059364\ttotal: 8.8s\tremaining: 17s\n",
      "341:\tlearn: 0.0059237\ttotal: 8.84s\tremaining: 17s\n",
      "342:\tlearn: 0.0059230\ttotal: 8.86s\tremaining: 17s\n",
      "343:\tlearn: 0.0058860\ttotal: 8.89s\tremaining: 16.9s\n",
      "344:\tlearn: 0.0058860\ttotal: 8.9s\tremaining: 16.9s\n",
      "345:\tlearn: 0.0058848\ttotal: 8.93s\tremaining: 16.9s\n",
      "346:\tlearn: 0.0058831\ttotal: 8.95s\tremaining: 16.8s\n",
      "347:\tlearn: 0.0058808\ttotal: 8.98s\tremaining: 16.8s\n",
      "348:\tlearn: 0.0058662\ttotal: 9.01s\tremaining: 16.8s\n",
      "349:\tlearn: 0.0058375\ttotal: 9.04s\tremaining: 16.8s\n",
      "350:\tlearn: 0.0058318\ttotal: 9.06s\tremaining: 16.8s\n",
      "351:\tlearn: 0.0058253\ttotal: 9.11s\tremaining: 16.8s\n",
      "352:\tlearn: 0.0058181\ttotal: 9.13s\tremaining: 16.7s\n",
      "353:\tlearn: 0.0058180\ttotal: 9.15s\tremaining: 16.7s\n",
      "354:\tlearn: 0.0057996\ttotal: 9.21s\tremaining: 16.7s\n",
      "355:\tlearn: 0.0057758\ttotal: 9.24s\tremaining: 16.7s\n",
      "356:\tlearn: 0.0057749\ttotal: 9.27s\tremaining: 16.7s\n",
      "357:\tlearn: 0.0057602\ttotal: 9.3s\tremaining: 16.7s\n",
      "358:\tlearn: 0.0057511\ttotal: 9.33s\tremaining: 16.7s\n",
      "359:\tlearn: 0.0057439\ttotal: 9.36s\tremaining: 16.6s\n",
      "360:\tlearn: 0.0057439\ttotal: 9.38s\tremaining: 16.6s\n",
      "361:\tlearn: 0.0057423\ttotal: 9.4s\tremaining: 16.6s\n",
      "362:\tlearn: 0.0057421\ttotal: 9.43s\tremaining: 16.5s\n",
      "363:\tlearn: 0.0057227\ttotal: 9.46s\tremaining: 16.5s\n",
      "364:\tlearn: 0.0057227\ttotal: 9.47s\tremaining: 16.5s\n",
      "365:\tlearn: 0.0057164\ttotal: 9.5s\tremaining: 16.5s\n",
      "366:\tlearn: 0.0057113\ttotal: 9.53s\tremaining: 16.4s\n",
      "367:\tlearn: 0.0057025\ttotal: 9.56s\tremaining: 16.4s\n",
      "368:\tlearn: 0.0056841\ttotal: 9.59s\tremaining: 16.4s\n",
      "369:\tlearn: 0.0056770\ttotal: 9.62s\tremaining: 16.4s\n",
      "370:\tlearn: 0.0056694\ttotal: 9.65s\tremaining: 16.4s\n",
      "371:\tlearn: 0.0056652\ttotal: 9.68s\tremaining: 16.3s\n",
      "372:\tlearn: 0.0056500\ttotal: 9.71s\tremaining: 16.3s\n",
      "373:\tlearn: 0.0056410\ttotal: 9.74s\tremaining: 16.3s\n",
      "374:\tlearn: 0.0056376\ttotal: 9.77s\tremaining: 16.3s\n",
      "375:\tlearn: 0.0056110\ttotal: 9.79s\tremaining: 16.3s\n",
      "376:\tlearn: 0.0056110\ttotal: 9.82s\tremaining: 16.2s\n",
      "377:\tlearn: 0.0055961\ttotal: 9.84s\tremaining: 16.2s\n",
      "378:\tlearn: 0.0055942\ttotal: 9.88s\tremaining: 16.2s\n",
      "379:\tlearn: 0.0055859\ttotal: 9.9s\tremaining: 16.2s\n",
      "380:\tlearn: 0.0055854\ttotal: 9.93s\tremaining: 16.1s\n",
      "381:\tlearn: 0.0055840\ttotal: 9.95s\tremaining: 16.1s\n",
      "382:\tlearn: 0.0055830\ttotal: 9.97s\tremaining: 16.1s\n",
      "383:\tlearn: 0.0055691\ttotal: 10s\tremaining: 16s\n",
      "384:\tlearn: 0.0055580\ttotal: 10s\tremaining: 16s\n",
      "385:\tlearn: 0.0055565\ttotal: 10.1s\tremaining: 16s\n",
      "386:\tlearn: 0.0055498\ttotal: 10.1s\tremaining: 16s\n",
      "387:\tlearn: 0.0055496\ttotal: 10.1s\tremaining: 15.9s\n",
      "388:\tlearn: 0.0055491\ttotal: 10.1s\tremaining: 15.9s\n",
      "389:\tlearn: 0.0055491\ttotal: 10.2s\tremaining: 15.9s\n",
      "390:\tlearn: 0.0055491\ttotal: 10.2s\tremaining: 15.9s\n",
      "391:\tlearn: 0.0055379\ttotal: 10.2s\tremaining: 15.8s\n",
      "392:\tlearn: 0.0055282\ttotal: 10.3s\tremaining: 15.8s\n",
      "393:\tlearn: 0.0055193\ttotal: 10.3s\tremaining: 15.8s\n",
      "394:\tlearn: 0.0055062\ttotal: 10.3s\tremaining: 15.8s\n",
      "395:\tlearn: 0.0055048\ttotal: 10.3s\tremaining: 15.8s\n",
      "396:\tlearn: 0.0055025\ttotal: 10.4s\tremaining: 15.7s\n",
      "397:\tlearn: 0.0055025\ttotal: 10.4s\tremaining: 15.7s\n",
      "398:\tlearn: 0.0055025\ttotal: 10.4s\tremaining: 15.7s\n",
      "399:\tlearn: 0.0054831\ttotal: 10.4s\tremaining: 15.7s\n",
      "400:\tlearn: 0.0054814\ttotal: 10.5s\tremaining: 15.6s\n",
      "401:\tlearn: 0.0054681\ttotal: 10.5s\tremaining: 15.6s\n",
      "402:\tlearn: 0.0054610\ttotal: 10.5s\tremaining: 15.6s\n",
      "403:\tlearn: 0.0054538\ttotal: 10.6s\tremaining: 15.6s\n",
      "404:\tlearn: 0.0054533\ttotal: 10.6s\tremaining: 15.6s\n",
      "405:\tlearn: 0.0054307\ttotal: 10.6s\tremaining: 15.5s\n",
      "406:\tlearn: 0.0054141\ttotal: 10.6s\tremaining: 15.5s\n",
      "407:\tlearn: 0.0054018\ttotal: 10.7s\tremaining: 15.5s\n",
      "408:\tlearn: 0.0053997\ttotal: 10.7s\tremaining: 15.5s\n",
      "409:\tlearn: 0.0053972\ttotal: 10.7s\tremaining: 15.4s\n",
      "410:\tlearn: 0.0053870\ttotal: 10.8s\tremaining: 15.4s\n",
      "411:\tlearn: 0.0053708\ttotal: 10.8s\tremaining: 15.4s\n",
      "412:\tlearn: 0.0053562\ttotal: 10.8s\tremaining: 15.4s\n",
      "413:\tlearn: 0.0053478\ttotal: 10.8s\tremaining: 15.3s\n",
      "414:\tlearn: 0.0053403\ttotal: 10.9s\tremaining: 15.3s\n",
      "415:\tlearn: 0.0053403\ttotal: 10.9s\tremaining: 15.3s\n",
      "416:\tlearn: 0.0053337\ttotal: 10.9s\tremaining: 15.3s\n",
      "417:\tlearn: 0.0053265\ttotal: 11s\tremaining: 15.3s\n",
      "418:\tlearn: 0.0053222\ttotal: 11s\tremaining: 15.2s\n",
      "419:\tlearn: 0.0053222\ttotal: 11s\tremaining: 15.2s\n",
      "420:\tlearn: 0.0053125\ttotal: 11s\tremaining: 15.2s\n",
      "421:\tlearn: 0.0053083\ttotal: 11.1s\tremaining: 15.2s\n",
      "422:\tlearn: 0.0052963\ttotal: 11.1s\tremaining: 15.1s\n",
      "423:\tlearn: 0.0052959\ttotal: 11.1s\tremaining: 15.1s\n",
      "424:\tlearn: 0.0052945\ttotal: 11.1s\tremaining: 15.1s\n",
      "425:\tlearn: 0.0052746\ttotal: 11.2s\tremaining: 15.1s\n",
      "426:\tlearn: 0.0052656\ttotal: 11.2s\tremaining: 15.1s\n",
      "427:\tlearn: 0.0052652\ttotal: 11.2s\tremaining: 15s\n",
      "428:\tlearn: 0.0052539\ttotal: 11.3s\tremaining: 15s\n",
      "429:\tlearn: 0.0052508\ttotal: 11.3s\tremaining: 15s\n",
      "430:\tlearn: 0.0052487\ttotal: 11.4s\tremaining: 15s\n",
      "431:\tlearn: 0.0052359\ttotal: 11.4s\tremaining: 15s\n",
      "432:\tlearn: 0.0052359\ttotal: 11.4s\tremaining: 15s\n",
      "433:\tlearn: 0.0052248\ttotal: 11.5s\tremaining: 15s\n",
      "434:\tlearn: 0.0051978\ttotal: 11.5s\tremaining: 15s\n",
      "435:\tlearn: 0.0051869\ttotal: 11.6s\tremaining: 15s\n",
      "436:\tlearn: 0.0051797\ttotal: 11.6s\tremaining: 15s\n",
      "437:\tlearn: 0.0051781\ttotal: 11.6s\tremaining: 14.9s\n",
      "438:\tlearn: 0.0051736\ttotal: 11.7s\tremaining: 14.9s\n",
      "439:\tlearn: 0.0051668\ttotal: 11.7s\tremaining: 14.9s\n",
      "440:\tlearn: 0.0051577\ttotal: 11.7s\tremaining: 14.9s\n",
      "441:\tlearn: 0.0051449\ttotal: 11.8s\tremaining: 14.9s\n",
      "442:\tlearn: 0.0051436\ttotal: 11.8s\tremaining: 14.8s\n",
      "443:\tlearn: 0.0051329\ttotal: 11.8s\tremaining: 14.8s\n",
      "444:\tlearn: 0.0051221\ttotal: 11.9s\tremaining: 14.8s\n",
      "445:\tlearn: 0.0051186\ttotal: 11.9s\tremaining: 14.8s\n",
      "446:\tlearn: 0.0051128\ttotal: 11.9s\tremaining: 14.8s\n",
      "447:\tlearn: 0.0051024\ttotal: 12s\tremaining: 14.8s\n",
      "448:\tlearn: 0.0051014\ttotal: 12s\tremaining: 14.7s\n",
      "449:\tlearn: 0.0051011\ttotal: 12s\tremaining: 14.7s\n",
      "450:\tlearn: 0.0051007\ttotal: 12.1s\tremaining: 14.7s\n",
      "451:\tlearn: 0.0050992\ttotal: 12.1s\tremaining: 14.6s\n",
      "452:\tlearn: 0.0050926\ttotal: 12.1s\tremaining: 14.6s\n",
      "453:\tlearn: 0.0050750\ttotal: 12.2s\tremaining: 14.6s\n",
      "454:\tlearn: 0.0050555\ttotal: 12.2s\tremaining: 14.6s\n",
      "455:\tlearn: 0.0050555\ttotal: 12.2s\tremaining: 14.6s\n",
      "456:\tlearn: 0.0050454\ttotal: 12.2s\tremaining: 14.5s\n",
      "457:\tlearn: 0.0050399\ttotal: 12.3s\tremaining: 14.5s\n",
      "458:\tlearn: 0.0050308\ttotal: 12.3s\tremaining: 14.5s\n",
      "459:\tlearn: 0.0050308\ttotal: 12.3s\tremaining: 14.4s\n",
      "460:\tlearn: 0.0050250\ttotal: 12.3s\tremaining: 14.4s\n",
      "461:\tlearn: 0.0050131\ttotal: 12.4s\tremaining: 14.4s\n",
      "462:\tlearn: 0.0050070\ttotal: 12.4s\tremaining: 14.4s\n",
      "463:\tlearn: 0.0050070\ttotal: 12.4s\tremaining: 14.3s\n",
      "464:\tlearn: 0.0049907\ttotal: 12.4s\tremaining: 14.3s\n",
      "465:\tlearn: 0.0049903\ttotal: 12.5s\tremaining: 14.3s\n",
      "466:\tlearn: 0.0049901\ttotal: 12.5s\tremaining: 14.2s\n",
      "467:\tlearn: 0.0049839\ttotal: 12.5s\tremaining: 14.2s\n",
      "468:\tlearn: 0.0049709\ttotal: 12.5s\tremaining: 14.2s\n",
      "469:\tlearn: 0.0049692\ttotal: 12.5s\tremaining: 14.1s\n",
      "470:\tlearn: 0.0049691\ttotal: 12.6s\tremaining: 14.1s\n",
      "471:\tlearn: 0.0049659\ttotal: 12.6s\tremaining: 14.1s\n",
      "472:\tlearn: 0.0049655\ttotal: 12.6s\tremaining: 14s\n",
      "473:\tlearn: 0.0049611\ttotal: 12.6s\tremaining: 14s\n",
      "474:\tlearn: 0.0049492\ttotal: 12.7s\tremaining: 14s\n",
      "475:\tlearn: 0.0049389\ttotal: 12.7s\tremaining: 14s\n",
      "476:\tlearn: 0.0049212\ttotal: 12.7s\tremaining: 13.9s\n",
      "477:\tlearn: 0.0049087\ttotal: 12.7s\tremaining: 13.9s\n",
      "478:\tlearn: 0.0049087\ttotal: 12.8s\tremaining: 13.9s\n",
      "479:\tlearn: 0.0049045\ttotal: 12.8s\tremaining: 13.9s\n",
      "480:\tlearn: 0.0048993\ttotal: 12.8s\tremaining: 13.8s\n",
      "481:\tlearn: 0.0048981\ttotal: 12.8s\tremaining: 13.8s\n",
      "482:\tlearn: 0.0048955\ttotal: 12.9s\tremaining: 13.8s\n",
      "483:\tlearn: 0.0048886\ttotal: 12.9s\tremaining: 13.8s\n",
      "484:\tlearn: 0.0048886\ttotal: 12.9s\tremaining: 13.7s\n",
      "485:\tlearn: 0.0048846\ttotal: 13s\tremaining: 13.7s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "486:\tlearn: 0.0048816\ttotal: 13s\tremaining: 13.7s\n",
      "487:\tlearn: 0.0048719\ttotal: 13s\tremaining: 13.7s\n",
      "488:\tlearn: 0.0048673\ttotal: 13.1s\tremaining: 13.6s\n",
      "489:\tlearn: 0.0048648\ttotal: 13.1s\tremaining: 13.6s\n",
      "490:\tlearn: 0.0048646\ttotal: 13.1s\tremaining: 13.6s\n",
      "491:\tlearn: 0.0048646\ttotal: 13.1s\tremaining: 13.6s\n",
      "492:\tlearn: 0.0048635\ttotal: 13.2s\tremaining: 13.5s\n",
      "493:\tlearn: 0.0048597\ttotal: 13.2s\tremaining: 13.5s\n",
      "494:\tlearn: 0.0048493\ttotal: 13.2s\tremaining: 13.5s\n",
      "495:\tlearn: 0.0048343\ttotal: 13.3s\tremaining: 13.5s\n",
      "496:\tlearn: 0.0048295\ttotal: 13.3s\tremaining: 13.5s\n",
      "497:\tlearn: 0.0048237\ttotal: 13.3s\tremaining: 13.4s\n",
      "498:\tlearn: 0.0048127\ttotal: 13.4s\tremaining: 13.4s\n",
      "499:\tlearn: 0.0048044\ttotal: 13.4s\tremaining: 13.4s\n",
      "500:\tlearn: 0.0047940\ttotal: 13.4s\tremaining: 13.4s\n",
      "501:\tlearn: 0.0047915\ttotal: 13.5s\tremaining: 13.4s\n",
      "502:\tlearn: 0.0047866\ttotal: 13.5s\tremaining: 13.4s\n",
      "503:\tlearn: 0.0047862\ttotal: 13.5s\tremaining: 13.3s\n",
      "504:\tlearn: 0.0047859\ttotal: 13.6s\tremaining: 13.3s\n",
      "505:\tlearn: 0.0047838\ttotal: 13.6s\tremaining: 13.3s\n",
      "506:\tlearn: 0.0047510\ttotal: 13.6s\tremaining: 13.3s\n",
      "507:\tlearn: 0.0047367\ttotal: 13.7s\tremaining: 13.2s\n",
      "508:\tlearn: 0.0047366\ttotal: 13.7s\tremaining: 13.2s\n",
      "509:\tlearn: 0.0047337\ttotal: 13.7s\tremaining: 13.2s\n",
      "510:\tlearn: 0.0047241\ttotal: 13.8s\tremaining: 13.2s\n",
      "511:\tlearn: 0.0047237\ttotal: 13.8s\tremaining: 13.1s\n",
      "512:\tlearn: 0.0047145\ttotal: 13.8s\tremaining: 13.1s\n",
      "513:\tlearn: 0.0047121\ttotal: 13.9s\tremaining: 13.1s\n",
      "514:\tlearn: 0.0047011\ttotal: 13.9s\tremaining: 13.1s\n",
      "515:\tlearn: 0.0046987\ttotal: 13.9s\tremaining: 13.1s\n",
      "516:\tlearn: 0.0046965\ttotal: 14s\tremaining: 13.1s\n",
      "517:\tlearn: 0.0046963\ttotal: 14s\tremaining: 13s\n",
      "518:\tlearn: 0.0046963\ttotal: 14s\tremaining: 13s\n",
      "519:\tlearn: 0.0046943\ttotal: 14.1s\tremaining: 13s\n",
      "520:\tlearn: 0.0046918\ttotal: 14.1s\tremaining: 12.9s\n",
      "521:\tlearn: 0.0046878\ttotal: 14.1s\tremaining: 12.9s\n",
      "522:\tlearn: 0.0046878\ttotal: 14.1s\tremaining: 12.9s\n",
      "523:\tlearn: 0.0046811\ttotal: 14.2s\tremaining: 12.9s\n",
      "524:\tlearn: 0.0046795\ttotal: 14.2s\tremaining: 12.9s\n",
      "525:\tlearn: 0.0046792\ttotal: 14.2s\tremaining: 12.8s\n",
      "526:\tlearn: 0.0046730\ttotal: 14.3s\tremaining: 12.8s\n",
      "527:\tlearn: 0.0046601\ttotal: 14.3s\tremaining: 12.8s\n",
      "528:\tlearn: 0.0046588\ttotal: 14.3s\tremaining: 12.8s\n",
      "529:\tlearn: 0.0046585\ttotal: 14.3s\tremaining: 12.7s\n",
      "530:\tlearn: 0.0046495\ttotal: 14.4s\tremaining: 12.7s\n",
      "531:\tlearn: 0.0046310\ttotal: 14.4s\tremaining: 12.7s\n",
      "532:\tlearn: 0.0046219\ttotal: 14.4s\tremaining: 12.7s\n",
      "533:\tlearn: 0.0046219\ttotal: 14.5s\tremaining: 12.6s\n",
      "534:\tlearn: 0.0046145\ttotal: 14.5s\tremaining: 12.6s\n",
      "535:\tlearn: 0.0046145\ttotal: 14.5s\tremaining: 12.6s\n",
      "536:\tlearn: 0.0046091\ttotal: 14.6s\tremaining: 12.6s\n",
      "537:\tlearn: 0.0046006\ttotal: 14.6s\tremaining: 12.5s\n",
      "538:\tlearn: 0.0045961\ttotal: 14.6s\tremaining: 12.5s\n",
      "539:\tlearn: 0.0045833\ttotal: 14.7s\tremaining: 12.5s\n",
      "540:\tlearn: 0.0045797\ttotal: 14.7s\tremaining: 12.5s\n",
      "541:\tlearn: 0.0045773\ttotal: 14.7s\tremaining: 12.5s\n",
      "542:\tlearn: 0.0045707\ttotal: 14.8s\tremaining: 12.4s\n",
      "543:\tlearn: 0.0045704\ttotal: 14.8s\tremaining: 12.4s\n",
      "544:\tlearn: 0.0045583\ttotal: 14.8s\tremaining: 12.4s\n",
      "545:\tlearn: 0.0045464\ttotal: 14.9s\tremaining: 12.4s\n",
      "546:\tlearn: 0.0045383\ttotal: 14.9s\tremaining: 12.4s\n",
      "547:\tlearn: 0.0045341\ttotal: 15s\tremaining: 12.3s\n",
      "548:\tlearn: 0.0045246\ttotal: 15s\tremaining: 12.3s\n",
      "549:\tlearn: 0.0045246\ttotal: 15s\tremaining: 12.3s\n",
      "550:\tlearn: 0.0045231\ttotal: 15s\tremaining: 12.3s\n",
      "551:\tlearn: 0.0045207\ttotal: 15.1s\tremaining: 12.2s\n",
      "552:\tlearn: 0.0045164\ttotal: 15.1s\tremaining: 12.2s\n",
      "553:\tlearn: 0.0045049\ttotal: 15.1s\tremaining: 12.2s\n",
      "554:\tlearn: 0.0044876\ttotal: 15.2s\tremaining: 12.2s\n",
      "555:\tlearn: 0.0044786\ttotal: 15.2s\tremaining: 12.1s\n",
      "556:\tlearn: 0.0044561\ttotal: 15.2s\tremaining: 12.1s\n",
      "557:\tlearn: 0.0044518\ttotal: 15.3s\tremaining: 12.1s\n",
      "558:\tlearn: 0.0044477\ttotal: 15.3s\tremaining: 12.1s\n",
      "559:\tlearn: 0.0044412\ttotal: 15.3s\tremaining: 12.1s\n",
      "560:\tlearn: 0.0044392\ttotal: 15.4s\tremaining: 12s\n",
      "561:\tlearn: 0.0044390\ttotal: 15.4s\tremaining: 12s\n",
      "562:\tlearn: 0.0044313\ttotal: 15.4s\tremaining: 12s\n",
      "563:\tlearn: 0.0044271\ttotal: 15.5s\tremaining: 12s\n",
      "564:\tlearn: 0.0044271\ttotal: 15.5s\tremaining: 11.9s\n",
      "565:\tlearn: 0.0044232\ttotal: 15.5s\tremaining: 11.9s\n",
      "566:\tlearn: 0.0044073\ttotal: 15.6s\tremaining: 11.9s\n",
      "567:\tlearn: 0.0044071\ttotal: 15.6s\tremaining: 11.9s\n",
      "568:\tlearn: 0.0044043\ttotal: 15.6s\tremaining: 11.8s\n",
      "569:\tlearn: 0.0043860\ttotal: 15.7s\tremaining: 11.8s\n",
      "570:\tlearn: 0.0043854\ttotal: 15.7s\tremaining: 11.8s\n",
      "571:\tlearn: 0.0043848\ttotal: 15.7s\tremaining: 11.8s\n",
      "572:\tlearn: 0.0043845\ttotal: 15.8s\tremaining: 11.7s\n",
      "573:\tlearn: 0.0043758\ttotal: 15.8s\tremaining: 11.7s\n",
      "574:\tlearn: 0.0043706\ttotal: 15.8s\tremaining: 11.7s\n",
      "575:\tlearn: 0.0043652\ttotal: 15.9s\tremaining: 11.7s\n",
      "576:\tlearn: 0.0043585\ttotal: 15.9s\tremaining: 11.7s\n",
      "577:\tlearn: 0.0043482\ttotal: 15.9s\tremaining: 11.6s\n",
      "578:\tlearn: 0.0043331\ttotal: 16s\tremaining: 11.6s\n",
      "579:\tlearn: 0.0043326\ttotal: 16s\tremaining: 11.6s\n",
      "580:\tlearn: 0.0043293\ttotal: 16s\tremaining: 11.6s\n",
      "581:\tlearn: 0.0043290\ttotal: 16.1s\tremaining: 11.5s\n",
      "582:\tlearn: 0.0043210\ttotal: 16.1s\tremaining: 11.5s\n",
      "583:\tlearn: 0.0043189\ttotal: 16.1s\tremaining: 11.5s\n",
      "584:\tlearn: 0.0043175\ttotal: 16.2s\tremaining: 11.5s\n",
      "585:\tlearn: 0.0043124\ttotal: 16.2s\tremaining: 11.5s\n",
      "586:\tlearn: 0.0043103\ttotal: 16.2s\tremaining: 11.4s\n",
      "587:\tlearn: 0.0043026\ttotal: 16.3s\tremaining: 11.4s\n",
      "588:\tlearn: 0.0042979\ttotal: 16.3s\tremaining: 11.4s\n",
      "589:\tlearn: 0.0042903\ttotal: 16.4s\tremaining: 11.4s\n",
      "590:\tlearn: 0.0042889\ttotal: 16.4s\tremaining: 11.3s\n",
      "591:\tlearn: 0.0042878\ttotal: 16.4s\tremaining: 11.3s\n",
      "592:\tlearn: 0.0042700\ttotal: 16.5s\tremaining: 11.3s\n",
      "593:\tlearn: 0.0042696\ttotal: 16.5s\tremaining: 11.3s\n",
      "594:\tlearn: 0.0042696\ttotal: 16.5s\tremaining: 11.2s\n",
      "595:\tlearn: 0.0042694\ttotal: 16.5s\tremaining: 11.2s\n",
      "596:\tlearn: 0.0042591\ttotal: 16.6s\tremaining: 11.2s\n",
      "597:\tlearn: 0.0042509\ttotal: 16.6s\tremaining: 11.2s\n",
      "598:\tlearn: 0.0042508\ttotal: 16.6s\tremaining: 11.1s\n",
      "599:\tlearn: 0.0042444\ttotal: 16.6s\tremaining: 11.1s\n",
      "600:\tlearn: 0.0042406\ttotal: 16.7s\tremaining: 11.1s\n",
      "601:\tlearn: 0.0042393\ttotal: 16.7s\tremaining: 11s\n",
      "602:\tlearn: 0.0042381\ttotal: 16.7s\tremaining: 11s\n",
      "603:\tlearn: 0.0042215\ttotal: 16.8s\tremaining: 11s\n",
      "604:\tlearn: 0.0042215\ttotal: 16.8s\tremaining: 11s\n",
      "605:\tlearn: 0.0042206\ttotal: 16.8s\tremaining: 10.9s\n",
      "606:\tlearn: 0.0042206\ttotal: 16.8s\tremaining: 10.9s\n",
      "607:\tlearn: 0.0042203\ttotal: 16.9s\tremaining: 10.9s\n",
      "608:\tlearn: 0.0042148\ttotal: 16.9s\tremaining: 10.8s\n",
      "609:\tlearn: 0.0042135\ttotal: 16.9s\tremaining: 10.8s\n",
      "610:\tlearn: 0.0042135\ttotal: 17s\tremaining: 10.8s\n",
      "611:\tlearn: 0.0041962\ttotal: 17s\tremaining: 10.8s\n",
      "612:\tlearn: 0.0041950\ttotal: 17s\tremaining: 10.8s\n",
      "613:\tlearn: 0.0041950\ttotal: 17.1s\tremaining: 10.7s\n",
      "614:\tlearn: 0.0041936\ttotal: 17.1s\tremaining: 10.7s\n",
      "615:\tlearn: 0.0041894\ttotal: 17.1s\tremaining: 10.7s\n",
      "616:\tlearn: 0.0041890\ttotal: 17.2s\tremaining: 10.7s\n",
      "617:\tlearn: 0.0041850\ttotal: 17.2s\tremaining: 10.6s\n",
      "618:\tlearn: 0.0041836\ttotal: 17.2s\tremaining: 10.6s\n",
      "619:\tlearn: 0.0041793\ttotal: 17.3s\tremaining: 10.6s\n",
      "620:\tlearn: 0.0041715\ttotal: 17.3s\tremaining: 10.6s\n",
      "621:\tlearn: 0.0041665\ttotal: 17.3s\tremaining: 10.5s\n",
      "622:\tlearn: 0.0041665\ttotal: 17.4s\tremaining: 10.5s\n",
      "623:\tlearn: 0.0041626\ttotal: 17.4s\tremaining: 10.5s\n",
      "624:\tlearn: 0.0041604\ttotal: 17.4s\tremaining: 10.5s\n",
      "625:\tlearn: 0.0041555\ttotal: 17.5s\tremaining: 10.4s\n",
      "626:\tlearn: 0.0041418\ttotal: 17.5s\tremaining: 10.4s\n",
      "627:\tlearn: 0.0041416\ttotal: 17.5s\tremaining: 10.4s\n",
      "628:\tlearn: 0.0041416\ttotal: 17.6s\tremaining: 10.4s\n",
      "629:\tlearn: 0.0041405\ttotal: 17.6s\tremaining: 10.3s\n",
      "630:\tlearn: 0.0041242\ttotal: 17.6s\tremaining: 10.3s\n",
      "631:\tlearn: 0.0041242\ttotal: 17.6s\tremaining: 10.3s\n",
      "632:\tlearn: 0.0041215\ttotal: 17.7s\tremaining: 10.3s\n",
      "633:\tlearn: 0.0041210\ttotal: 17.7s\tremaining: 10.2s\n",
      "634:\tlearn: 0.0041154\ttotal: 17.7s\tremaining: 10.2s\n",
      "635:\tlearn: 0.0041113\ttotal: 17.8s\tremaining: 10.2s\n",
      "636:\tlearn: 0.0041069\ttotal: 17.8s\tremaining: 10.2s\n",
      "637:\tlearn: 0.0040910\ttotal: 17.9s\tremaining: 10.1s\n",
      "638:\tlearn: 0.0040857\ttotal: 17.9s\tremaining: 10.1s\n",
      "639:\tlearn: 0.0040820\ttotal: 17.9s\tremaining: 10.1s\n",
      "640:\tlearn: 0.0040747\ttotal: 17.9s\tremaining: 10s\n",
      "641:\tlearn: 0.0040678\ttotal: 18s\tremaining: 10s\n",
      "642:\tlearn: 0.0040529\ttotal: 18s\tremaining: 9.99s\n",
      "643:\tlearn: 0.0040527\ttotal: 18s\tremaining: 9.97s\n",
      "644:\tlearn: 0.0040492\ttotal: 18.1s\tremaining: 9.95s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "645:\tlearn: 0.0040491\ttotal: 18.1s\tremaining: 9.92s\n",
      "646:\tlearn: 0.0040375\ttotal: 18.1s\tremaining: 9.9s\n",
      "647:\tlearn: 0.0040290\ttotal: 18.2s\tremaining: 9.88s\n",
      "648:\tlearn: 0.0040234\ttotal: 18.2s\tremaining: 9.85s\n",
      "649:\tlearn: 0.0040178\ttotal: 18.2s\tremaining: 9.82s\n",
      "650:\tlearn: 0.0040125\ttotal: 18.3s\tremaining: 9.8s\n",
      "651:\tlearn: 0.0040125\ttotal: 18.3s\tremaining: 9.77s\n",
      "652:\tlearn: 0.0040083\ttotal: 18.3s\tremaining: 9.74s\n",
      "653:\tlearn: 0.0040075\ttotal: 18.4s\tremaining: 9.72s\n",
      "654:\tlearn: 0.0040047\ttotal: 18.4s\tremaining: 9.69s\n",
      "655:\tlearn: 0.0040009\ttotal: 18.4s\tremaining: 9.66s\n",
      "656:\tlearn: 0.0039988\ttotal: 18.5s\tremaining: 9.63s\n",
      "657:\tlearn: 0.0039988\ttotal: 18.5s\tremaining: 9.61s\n",
      "658:\tlearn: 0.0039984\ttotal: 18.5s\tremaining: 9.58s\n",
      "659:\tlearn: 0.0039905\ttotal: 18.6s\tremaining: 9.56s\n",
      "660:\tlearn: 0.0039904\ttotal: 18.6s\tremaining: 9.53s\n",
      "661:\tlearn: 0.0039823\ttotal: 18.6s\tremaining: 9.51s\n",
      "662:\tlearn: 0.0039716\ttotal: 18.7s\tremaining: 9.48s\n",
      "663:\tlearn: 0.0039706\ttotal: 18.7s\tremaining: 9.46s\n",
      "664:\tlearn: 0.0039706\ttotal: 18.7s\tremaining: 9.42s\n",
      "665:\tlearn: 0.0039644\ttotal: 18.8s\tremaining: 9.41s\n",
      "666:\tlearn: 0.0039543\ttotal: 18.8s\tremaining: 9.38s\n",
      "667:\tlearn: 0.0039540\ttotal: 18.8s\tremaining: 9.35s\n",
      "668:\tlearn: 0.0039446\ttotal: 18.9s\tremaining: 9.33s\n",
      "669:\tlearn: 0.0039446\ttotal: 18.9s\tremaining: 9.3s\n",
      "670:\tlearn: 0.0039387\ttotal: 18.9s\tremaining: 9.28s\n",
      "671:\tlearn: 0.0039246\ttotal: 19s\tremaining: 9.25s\n",
      "672:\tlearn: 0.0039207\ttotal: 19s\tremaining: 9.23s\n",
      "673:\tlearn: 0.0039207\ttotal: 19s\tremaining: 9.21s\n",
      "674:\tlearn: 0.0039207\ttotal: 19.1s\tremaining: 9.17s\n",
      "675:\tlearn: 0.0039149\ttotal: 19.1s\tremaining: 9.15s\n",
      "676:\tlearn: 0.0039149\ttotal: 19.1s\tremaining: 9.13s\n",
      "677:\tlearn: 0.0038999\ttotal: 19.2s\tremaining: 9.11s\n",
      "678:\tlearn: 0.0038997\ttotal: 19.2s\tremaining: 9.08s\n",
      "679:\tlearn: 0.0038956\ttotal: 19.3s\tremaining: 9.06s\n",
      "680:\tlearn: 0.0038920\ttotal: 19.3s\tremaining: 9.04s\n",
      "681:\tlearn: 0.0038917\ttotal: 19.3s\tremaining: 9.01s\n",
      "682:\tlearn: 0.0038775\ttotal: 19.4s\tremaining: 8.99s\n",
      "683:\tlearn: 0.0038721\ttotal: 19.4s\tremaining: 8.97s\n",
      "684:\tlearn: 0.0038685\ttotal: 19.4s\tremaining: 8.94s\n",
      "685:\tlearn: 0.0038637\ttotal: 19.5s\tremaining: 8.92s\n",
      "686:\tlearn: 0.0038605\ttotal: 19.5s\tremaining: 8.89s\n",
      "687:\tlearn: 0.0038605\ttotal: 19.6s\tremaining: 8.87s\n",
      "688:\tlearn: 0.0038475\ttotal: 19.6s\tremaining: 8.83s\n",
      "689:\tlearn: 0.0038427\ttotal: 19.6s\tremaining: 8.81s\n",
      "690:\tlearn: 0.0038296\ttotal: 19.6s\tremaining: 8.78s\n",
      "691:\tlearn: 0.0038296\ttotal: 19.7s\tremaining: 8.75s\n",
      "692:\tlearn: 0.0038261\ttotal: 19.7s\tremaining: 8.73s\n",
      "693:\tlearn: 0.0038236\ttotal: 19.7s\tremaining: 8.7s\n",
      "694:\tlearn: 0.0038144\ttotal: 19.8s\tremaining: 8.68s\n",
      "695:\tlearn: 0.0038144\ttotal: 19.8s\tremaining: 8.65s\n",
      "696:\tlearn: 0.0038063\ttotal: 19.8s\tremaining: 8.62s\n",
      "697:\tlearn: 0.0038063\ttotal: 19.9s\tremaining: 8.59s\n",
      "698:\tlearn: 0.0038060\ttotal: 19.9s\tremaining: 8.56s\n",
      "699:\tlearn: 0.0038017\ttotal: 19.9s\tremaining: 8.53s\n",
      "700:\tlearn: 0.0038017\ttotal: 19.9s\tremaining: 8.5s\n",
      "701:\tlearn: 0.0038017\ttotal: 20s\tremaining: 8.47s\n",
      "702:\tlearn: 0.0037954\ttotal: 20s\tremaining: 8.45s\n",
      "703:\tlearn: 0.0037953\ttotal: 20s\tremaining: 8.42s\n",
      "704:\tlearn: 0.0037951\ttotal: 20.1s\tremaining: 8.39s\n",
      "705:\tlearn: 0.0037951\ttotal: 20.1s\tremaining: 8.36s\n",
      "706:\tlearn: 0.0037895\ttotal: 20.1s\tremaining: 8.34s\n",
      "707:\tlearn: 0.0037895\ttotal: 20.1s\tremaining: 8.3s\n",
      "708:\tlearn: 0.0037892\ttotal: 20.2s\tremaining: 8.28s\n",
      "709:\tlearn: 0.0037892\ttotal: 20.2s\tremaining: 8.25s\n",
      "710:\tlearn: 0.0037816\ttotal: 20.2s\tremaining: 8.23s\n",
      "711:\tlearn: 0.0037753\ttotal: 20.3s\tremaining: 8.2s\n",
      "712:\tlearn: 0.0037725\ttotal: 20.3s\tremaining: 8.18s\n",
      "713:\tlearn: 0.0037694\ttotal: 20.4s\tremaining: 8.15s\n",
      "714:\tlearn: 0.0037652\ttotal: 20.4s\tremaining: 8.13s\n",
      "715:\tlearn: 0.0037554\ttotal: 20.4s\tremaining: 8.1s\n",
      "716:\tlearn: 0.0037554\ttotal: 20.5s\tremaining: 8.07s\n",
      "717:\tlearn: 0.0037521\ttotal: 20.5s\tremaining: 8.05s\n",
      "718:\tlearn: 0.0037449\ttotal: 20.5s\tremaining: 8.02s\n",
      "719:\tlearn: 0.0037393\ttotal: 20.6s\tremaining: 8s\n",
      "720:\tlearn: 0.0037380\ttotal: 20.6s\tremaining: 7.97s\n",
      "721:\tlearn: 0.0037376\ttotal: 20.6s\tremaining: 7.95s\n",
      "722:\tlearn: 0.0037309\ttotal: 20.7s\tremaining: 7.92s\n",
      "723:\tlearn: 0.0037241\ttotal: 20.7s\tremaining: 7.89s\n",
      "724:\tlearn: 0.0037150\ttotal: 20.7s\tremaining: 7.86s\n",
      "725:\tlearn: 0.0037052\ttotal: 20.8s\tremaining: 7.83s\n",
      "726:\tlearn: 0.0036992\ttotal: 20.8s\tremaining: 7.81s\n",
      "727:\tlearn: 0.0036860\ttotal: 20.8s\tremaining: 7.78s\n",
      "728:\tlearn: 0.0036788\ttotal: 20.9s\tremaining: 7.76s\n",
      "729:\tlearn: 0.0036725\ttotal: 20.9s\tremaining: 7.73s\n",
      "730:\tlearn: 0.0036719\ttotal: 20.9s\tremaining: 7.7s\n",
      "731:\tlearn: 0.0036719\ttotal: 21s\tremaining: 7.68s\n",
      "732:\tlearn: 0.0036718\ttotal: 21s\tremaining: 7.66s\n",
      "733:\tlearn: 0.0036685\ttotal: 21.1s\tremaining: 7.63s\n",
      "734:\tlearn: 0.0036685\ttotal: 21.1s\tremaining: 7.6s\n",
      "735:\tlearn: 0.0036655\ttotal: 21.1s\tremaining: 7.57s\n",
      "736:\tlearn: 0.0036602\ttotal: 21.1s\tremaining: 7.54s\n",
      "737:\tlearn: 0.0036602\ttotal: 21.2s\tremaining: 7.51s\n",
      "738:\tlearn: 0.0036548\ttotal: 21.2s\tremaining: 7.48s\n",
      "739:\tlearn: 0.0036548\ttotal: 21.2s\tremaining: 7.46s\n",
      "740:\tlearn: 0.0036548\ttotal: 21.3s\tremaining: 7.43s\n",
      "741:\tlearn: 0.0036505\ttotal: 21.3s\tremaining: 7.4s\n",
      "742:\tlearn: 0.0036467\ttotal: 21.3s\tremaining: 7.37s\n",
      "743:\tlearn: 0.0036342\ttotal: 21.4s\tremaining: 7.35s\n",
      "744:\tlearn: 0.0036320\ttotal: 21.4s\tremaining: 7.32s\n",
      "745:\tlearn: 0.0036283\ttotal: 21.4s\tremaining: 7.29s\n",
      "746:\tlearn: 0.0036255\ttotal: 21.4s\tremaining: 7.26s\n",
      "747:\tlearn: 0.0036227\ttotal: 21.5s\tremaining: 7.24s\n",
      "748:\tlearn: 0.0036227\ttotal: 21.5s\tremaining: 7.21s\n",
      "749:\tlearn: 0.0036190\ttotal: 21.5s\tremaining: 7.18s\n",
      "750:\tlearn: 0.0036167\ttotal: 21.6s\tremaining: 7.15s\n",
      "751:\tlearn: 0.0036134\ttotal: 21.6s\tremaining: 7.13s\n",
      "752:\tlearn: 0.0036078\ttotal: 21.6s\tremaining: 7.1s\n",
      "753:\tlearn: 0.0036025\ttotal: 21.7s\tremaining: 7.07s\n",
      "754:\tlearn: 0.0035990\ttotal: 21.7s\tremaining: 7.04s\n",
      "755:\tlearn: 0.0035956\ttotal: 21.7s\tremaining: 7.01s\n",
      "756:\tlearn: 0.0035936\ttotal: 21.8s\tremaining: 6.99s\n",
      "757:\tlearn: 0.0035907\ttotal: 21.8s\tremaining: 6.96s\n",
      "758:\tlearn: 0.0035882\ttotal: 21.8s\tremaining: 6.93s\n",
      "759:\tlearn: 0.0035882\ttotal: 21.9s\tremaining: 6.9s\n",
      "760:\tlearn: 0.0035849\ttotal: 21.9s\tremaining: 6.88s\n",
      "761:\tlearn: 0.0035826\ttotal: 21.9s\tremaining: 6.85s\n",
      "762:\tlearn: 0.0035826\ttotal: 22s\tremaining: 6.82s\n",
      "763:\tlearn: 0.0035708\ttotal: 22s\tremaining: 6.79s\n",
      "764:\tlearn: 0.0035694\ttotal: 22s\tremaining: 6.77s\n",
      "765:\tlearn: 0.0035692\ttotal: 22.1s\tremaining: 6.74s\n",
      "766:\tlearn: 0.0035687\ttotal: 22.1s\tremaining: 6.71s\n",
      "767:\tlearn: 0.0035619\ttotal: 22.1s\tremaining: 6.68s\n",
      "768:\tlearn: 0.0035589\ttotal: 22.2s\tremaining: 6.65s\n",
      "769:\tlearn: 0.0035557\ttotal: 22.2s\tremaining: 6.63s\n",
      "770:\tlearn: 0.0035447\ttotal: 22.2s\tremaining: 6.6s\n",
      "771:\tlearn: 0.0035394\ttotal: 22.3s\tremaining: 6.57s\n",
      "772:\tlearn: 0.0035376\ttotal: 22.3s\tremaining: 6.54s\n",
      "773:\tlearn: 0.0035345\ttotal: 22.3s\tremaining: 6.51s\n",
      "774:\tlearn: 0.0035240\ttotal: 22.3s\tremaining: 6.49s\n",
      "775:\tlearn: 0.0035222\ttotal: 22.4s\tremaining: 6.46s\n",
      "776:\tlearn: 0.0035222\ttotal: 22.4s\tremaining: 6.43s\n",
      "777:\tlearn: 0.0035063\ttotal: 22.4s\tremaining: 6.4s\n",
      "778:\tlearn: 0.0035026\ttotal: 22.5s\tremaining: 6.38s\n",
      "779:\tlearn: 0.0034994\ttotal: 22.5s\tremaining: 6.35s\n",
      "780:\tlearn: 0.0034938\ttotal: 22.5s\tremaining: 6.32s\n",
      "781:\tlearn: 0.0034908\ttotal: 22.6s\tremaining: 6.29s\n",
      "782:\tlearn: 0.0034902\ttotal: 22.6s\tremaining: 6.26s\n",
      "783:\tlearn: 0.0034896\ttotal: 22.6s\tremaining: 6.24s\n",
      "784:\tlearn: 0.0034868\ttotal: 22.6s\tremaining: 6.2s\n",
      "785:\tlearn: 0.0034853\ttotal: 22.7s\tremaining: 6.17s\n",
      "786:\tlearn: 0.0034805\ttotal: 22.7s\tremaining: 6.15s\n",
      "787:\tlearn: 0.0034794\ttotal: 22.8s\tremaining: 6.12s\n",
      "788:\tlearn: 0.0034794\ttotal: 22.8s\tremaining: 6.09s\n",
      "789:\tlearn: 0.0034767\ttotal: 22.8s\tremaining: 6.07s\n",
      "790:\tlearn: 0.0034751\ttotal: 22.9s\tremaining: 6.04s\n",
      "791:\tlearn: 0.0034751\ttotal: 22.9s\tremaining: 6.01s\n",
      "792:\tlearn: 0.0034746\ttotal: 22.9s\tremaining: 5.98s\n",
      "793:\tlearn: 0.0034737\ttotal: 22.9s\tremaining: 5.95s\n",
      "794:\tlearn: 0.0034669\ttotal: 23s\tremaining: 5.92s\n",
      "795:\tlearn: 0.0034655\ttotal: 23s\tremaining: 5.9s\n",
      "796:\tlearn: 0.0034604\ttotal: 23.1s\tremaining: 5.87s\n",
      "797:\tlearn: 0.0034590\ttotal: 23.1s\tremaining: 5.84s\n",
      "798:\tlearn: 0.0034544\ttotal: 23.1s\tremaining: 5.82s\n",
      "799:\tlearn: 0.0034539\ttotal: 23.2s\tremaining: 5.79s\n",
      "800:\tlearn: 0.0034503\ttotal: 23.2s\tremaining: 5.76s\n",
      "801:\tlearn: 0.0034500\ttotal: 23.2s\tremaining: 5.73s\n",
      "802:\tlearn: 0.0034451\ttotal: 23.3s\tremaining: 5.71s\n",
      "803:\tlearn: 0.0034417\ttotal: 23.3s\tremaining: 5.68s\n",
      "804:\tlearn: 0.0034413\ttotal: 23.3s\tremaining: 5.65s\n",
      "805:\tlearn: 0.0034410\ttotal: 23.4s\tremaining: 5.62s\n",
      "806:\tlearn: 0.0034393\ttotal: 23.4s\tremaining: 5.59s\n",
      "807:\tlearn: 0.0034290\ttotal: 23.4s\tremaining: 5.57s\n",
      "808:\tlearn: 0.0034208\ttotal: 23.5s\tremaining: 5.54s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "809:\tlearn: 0.0034162\ttotal: 23.5s\tremaining: 5.51s\n",
      "810:\tlearn: 0.0034126\ttotal: 23.5s\tremaining: 5.48s\n",
      "811:\tlearn: 0.0034126\ttotal: 23.6s\tremaining: 5.45s\n",
      "812:\tlearn: 0.0034126\ttotal: 23.6s\tremaining: 5.42s\n",
      "813:\tlearn: 0.0034082\ttotal: 23.6s\tremaining: 5.4s\n",
      "814:\tlearn: 0.0034028\ttotal: 23.7s\tremaining: 5.37s\n",
      "815:\tlearn: 0.0034010\ttotal: 23.7s\tremaining: 5.34s\n",
      "816:\tlearn: 0.0033982\ttotal: 23.7s\tremaining: 5.31s\n",
      "817:\tlearn: 0.0033957\ttotal: 23.8s\tremaining: 5.28s\n",
      "818:\tlearn: 0.0033957\ttotal: 23.8s\tremaining: 5.25s\n",
      "819:\tlearn: 0.0033957\ttotal: 23.8s\tremaining: 5.23s\n",
      "820:\tlearn: 0.0033957\ttotal: 23.8s\tremaining: 5.2s\n",
      "821:\tlearn: 0.0033952\ttotal: 23.9s\tremaining: 5.17s\n",
      "822:\tlearn: 0.0033918\ttotal: 23.9s\tremaining: 5.14s\n",
      "823:\tlearn: 0.0033903\ttotal: 23.9s\tremaining: 5.11s\n",
      "824:\tlearn: 0.0033902\ttotal: 23.9s\tremaining: 5.08s\n",
      "825:\tlearn: 0.0033844\ttotal: 24s\tremaining: 5.05s\n",
      "826:\tlearn: 0.0033784\ttotal: 24s\tremaining: 5.03s\n",
      "827:\tlearn: 0.0033767\ttotal: 24.1s\tremaining: 5s\n",
      "828:\tlearn: 0.0033742\ttotal: 24.1s\tremaining: 4.97s\n",
      "829:\tlearn: 0.0033742\ttotal: 24.1s\tremaining: 4.94s\n",
      "830:\tlearn: 0.0033738\ttotal: 24.2s\tremaining: 4.91s\n",
      "831:\tlearn: 0.0033697\ttotal: 24.2s\tremaining: 4.88s\n",
      "832:\tlearn: 0.0033590\ttotal: 24.2s\tremaining: 4.86s\n",
      "833:\tlearn: 0.0033575\ttotal: 24.3s\tremaining: 4.83s\n",
      "834:\tlearn: 0.0033548\ttotal: 24.3s\tremaining: 4.8s\n",
      "835:\tlearn: 0.0033543\ttotal: 24.3s\tremaining: 4.77s\n",
      "836:\tlearn: 0.0033543\ttotal: 24.3s\tremaining: 4.74s\n",
      "837:\tlearn: 0.0033447\ttotal: 24.4s\tremaining: 4.71s\n",
      "838:\tlearn: 0.0033429\ttotal: 24.4s\tremaining: 4.68s\n",
      "839:\tlearn: 0.0033403\ttotal: 24.4s\tremaining: 4.66s\n",
      "840:\tlearn: 0.0033403\ttotal: 24.5s\tremaining: 4.63s\n",
      "841:\tlearn: 0.0033380\ttotal: 24.5s\tremaining: 4.6s\n",
      "842:\tlearn: 0.0033333\ttotal: 24.6s\tremaining: 4.57s\n",
      "843:\tlearn: 0.0033309\ttotal: 24.6s\tremaining: 4.54s\n",
      "844:\tlearn: 0.0033305\ttotal: 24.6s\tremaining: 4.52s\n",
      "845:\tlearn: 0.0033301\ttotal: 24.6s\tremaining: 4.49s\n",
      "846:\tlearn: 0.0033299\ttotal: 24.7s\tremaining: 4.46s\n",
      "847:\tlearn: 0.0033294\ttotal: 24.7s\tremaining: 4.43s\n",
      "848:\tlearn: 0.0033290\ttotal: 24.7s\tremaining: 4.4s\n",
      "849:\tlearn: 0.0033290\ttotal: 24.7s\tremaining: 4.37s\n",
      "850:\tlearn: 0.0033287\ttotal: 24.8s\tremaining: 4.33s\n",
      "851:\tlearn: 0.0033148\ttotal: 24.8s\tremaining: 4.31s\n",
      "852:\tlearn: 0.0033145\ttotal: 24.8s\tremaining: 4.28s\n",
      "853:\tlearn: 0.0033001\ttotal: 24.9s\tremaining: 4.25s\n",
      "854:\tlearn: 0.0032979\ttotal: 24.9s\tremaining: 4.22s\n",
      "855:\tlearn: 0.0032977\ttotal: 24.9s\tremaining: 4.19s\n",
      "856:\tlearn: 0.0032974\ttotal: 24.9s\tremaining: 4.16s\n",
      "857:\tlearn: 0.0032969\ttotal: 24.9s\tremaining: 4.13s\n",
      "858:\tlearn: 0.0032897\ttotal: 25s\tremaining: 4.1s\n",
      "859:\tlearn: 0.0032845\ttotal: 25s\tremaining: 4.07s\n",
      "860:\tlearn: 0.0032797\ttotal: 25.1s\tremaining: 4.04s\n",
      "861:\tlearn: 0.0032756\ttotal: 25.1s\tremaining: 4.01s\n",
      "862:\tlearn: 0.0032756\ttotal: 25.1s\tremaining: 3.99s\n",
      "863:\tlearn: 0.0032685\ttotal: 25.2s\tremaining: 3.96s\n",
      "864:\tlearn: 0.0032654\ttotal: 25.2s\tremaining: 3.93s\n",
      "865:\tlearn: 0.0032630\ttotal: 25.2s\tremaining: 3.9s\n",
      "866:\tlearn: 0.0032542\ttotal: 25.3s\tremaining: 3.87s\n",
      "867:\tlearn: 0.0032488\ttotal: 25.3s\tremaining: 3.85s\n",
      "868:\tlearn: 0.0032440\ttotal: 25.3s\tremaining: 3.82s\n",
      "869:\tlearn: 0.0032356\ttotal: 25.4s\tremaining: 3.79s\n",
      "870:\tlearn: 0.0032356\ttotal: 25.4s\tremaining: 3.76s\n",
      "871:\tlearn: 0.0032335\ttotal: 25.4s\tremaining: 3.73s\n",
      "872:\tlearn: 0.0032335\ttotal: 25.4s\tremaining: 3.7s\n",
      "873:\tlearn: 0.0032335\ttotal: 25.5s\tremaining: 3.67s\n",
      "874:\tlearn: 0.0032320\ttotal: 25.5s\tremaining: 3.64s\n",
      "875:\tlearn: 0.0032300\ttotal: 25.5s\tremaining: 3.62s\n",
      "876:\tlearn: 0.0032297\ttotal: 25.6s\tremaining: 3.58s\n",
      "877:\tlearn: 0.0032297\ttotal: 25.6s\tremaining: 3.55s\n",
      "878:\tlearn: 0.0032295\ttotal: 25.6s\tremaining: 3.52s\n",
      "879:\tlearn: 0.0032291\ttotal: 25.6s\tremaining: 3.5s\n",
      "880:\tlearn: 0.0032288\ttotal: 25.7s\tremaining: 3.47s\n",
      "881:\tlearn: 0.0032288\ttotal: 25.7s\tremaining: 3.44s\n",
      "882:\tlearn: 0.0032267\ttotal: 25.7s\tremaining: 3.41s\n",
      "883:\tlearn: 0.0032254\ttotal: 25.8s\tremaining: 3.38s\n",
      "884:\tlearn: 0.0032234\ttotal: 25.8s\tremaining: 3.35s\n",
      "885:\tlearn: 0.0032191\ttotal: 25.8s\tremaining: 3.32s\n",
      "886:\tlearn: 0.0032112\ttotal: 25.9s\tremaining: 3.29s\n",
      "887:\tlearn: 0.0032073\ttotal: 25.9s\tremaining: 3.27s\n",
      "888:\tlearn: 0.0032048\ttotal: 25.9s\tremaining: 3.24s\n",
      "889:\tlearn: 0.0032041\ttotal: 26s\tremaining: 3.21s\n",
      "890:\tlearn: 0.0032038\ttotal: 26s\tremaining: 3.18s\n",
      "891:\tlearn: 0.0032019\ttotal: 26s\tremaining: 3.15s\n",
      "892:\tlearn: 0.0032004\ttotal: 26.1s\tremaining: 3.12s\n",
      "893:\tlearn: 0.0031998\ttotal: 26.1s\tremaining: 3.1s\n",
      "894:\tlearn: 0.0031962\ttotal: 26.1s\tremaining: 3.07s\n",
      "895:\tlearn: 0.0031962\ttotal: 26.2s\tremaining: 3.04s\n",
      "896:\tlearn: 0.0031906\ttotal: 26.2s\tremaining: 3.01s\n",
      "897:\tlearn: 0.0031883\ttotal: 26.2s\tremaining: 2.98s\n",
      "898:\tlearn: 0.0031883\ttotal: 26.3s\tremaining: 2.95s\n",
      "899:\tlearn: 0.0031880\ttotal: 26.3s\tremaining: 2.92s\n",
      "900:\tlearn: 0.0031822\ttotal: 26.3s\tremaining: 2.89s\n",
      "901:\tlearn: 0.0031819\ttotal: 26.4s\tremaining: 2.87s\n",
      "902:\tlearn: 0.0031816\ttotal: 26.4s\tremaining: 2.83s\n",
      "903:\tlearn: 0.0031796\ttotal: 26.4s\tremaining: 2.81s\n",
      "904:\tlearn: 0.0031793\ttotal: 26.5s\tremaining: 2.78s\n",
      "905:\tlearn: 0.0031781\ttotal: 26.5s\tremaining: 2.75s\n",
      "906:\tlearn: 0.0031777\ttotal: 26.5s\tremaining: 2.72s\n",
      "907:\tlearn: 0.0031777\ttotal: 26.5s\tremaining: 2.69s\n",
      "908:\tlearn: 0.0031774\ttotal: 26.6s\tremaining: 2.66s\n",
      "909:\tlearn: 0.0031771\ttotal: 26.6s\tremaining: 2.63s\n",
      "910:\tlearn: 0.0031744\ttotal: 26.6s\tremaining: 2.6s\n",
      "911:\tlearn: 0.0031722\ttotal: 26.6s\tremaining: 2.57s\n",
      "912:\tlearn: 0.0031673\ttotal: 26.7s\tremaining: 2.54s\n",
      "913:\tlearn: 0.0031673\ttotal: 26.7s\tremaining: 2.51s\n",
      "914:\tlearn: 0.0031598\ttotal: 26.7s\tremaining: 2.48s\n",
      "915:\tlearn: 0.0031580\ttotal: 26.8s\tremaining: 2.45s\n",
      "916:\tlearn: 0.0031580\ttotal: 26.8s\tremaining: 2.42s\n",
      "917:\tlearn: 0.0031560\ttotal: 26.8s\tremaining: 2.4s\n",
      "918:\tlearn: 0.0031511\ttotal: 26.8s\tremaining: 2.37s\n",
      "919:\tlearn: 0.0031479\ttotal: 26.9s\tremaining: 2.34s\n",
      "920:\tlearn: 0.0031465\ttotal: 26.9s\tremaining: 2.31s\n",
      "921:\tlearn: 0.0031451\ttotal: 27s\tremaining: 2.28s\n",
      "922:\tlearn: 0.0031451\ttotal: 27s\tremaining: 2.25s\n",
      "923:\tlearn: 0.0031448\ttotal: 27s\tremaining: 2.22s\n",
      "924:\tlearn: 0.0031446\ttotal: 27s\tremaining: 2.19s\n",
      "925:\tlearn: 0.0031446\ttotal: 27.1s\tremaining: 2.16s\n",
      "926:\tlearn: 0.0031402\ttotal: 27.1s\tremaining: 2.13s\n",
      "927:\tlearn: 0.0031390\ttotal: 27.1s\tremaining: 2.1s\n",
      "928:\tlearn: 0.0031388\ttotal: 27.1s\tremaining: 2.07s\n",
      "929:\tlearn: 0.0031388\ttotal: 27.2s\tremaining: 2.04s\n",
      "930:\tlearn: 0.0031377\ttotal: 27.2s\tremaining: 2.02s\n",
      "931:\tlearn: 0.0031360\ttotal: 27.2s\tremaining: 1.99s\n",
      "932:\tlearn: 0.0031352\ttotal: 27.3s\tremaining: 1.96s\n",
      "933:\tlearn: 0.0031289\ttotal: 27.3s\tremaining: 1.93s\n",
      "934:\tlearn: 0.0031287\ttotal: 27.3s\tremaining: 1.9s\n",
      "935:\tlearn: 0.0031232\ttotal: 27.4s\tremaining: 1.87s\n",
      "936:\tlearn: 0.0031213\ttotal: 27.4s\tremaining: 1.84s\n",
      "937:\tlearn: 0.0031178\ttotal: 27.4s\tremaining: 1.81s\n",
      "938:\tlearn: 0.0031177\ttotal: 27.4s\tremaining: 1.78s\n",
      "939:\tlearn: 0.0031175\ttotal: 27.5s\tremaining: 1.75s\n",
      "940:\tlearn: 0.0031173\ttotal: 27.5s\tremaining: 1.72s\n",
      "941:\tlearn: 0.0031035\ttotal: 27.5s\tremaining: 1.69s\n",
      "942:\tlearn: 0.0030929\ttotal: 27.6s\tremaining: 1.67s\n",
      "943:\tlearn: 0.0030929\ttotal: 27.6s\tremaining: 1.64s\n",
      "944:\tlearn: 0.0030891\ttotal: 27.6s\tremaining: 1.61s\n",
      "945:\tlearn: 0.0030879\ttotal: 27.7s\tremaining: 1.58s\n",
      "946:\tlearn: 0.0030802\ttotal: 27.7s\tremaining: 1.55s\n",
      "947:\tlearn: 0.0030799\ttotal: 27.7s\tremaining: 1.52s\n",
      "948:\tlearn: 0.0030799\ttotal: 27.7s\tremaining: 1.49s\n",
      "949:\tlearn: 0.0030691\ttotal: 27.8s\tremaining: 1.46s\n",
      "950:\tlearn: 0.0030688\ttotal: 27.8s\tremaining: 1.43s\n",
      "951:\tlearn: 0.0030686\ttotal: 27.8s\tremaining: 1.4s\n",
      "952:\tlearn: 0.0030678\ttotal: 27.8s\tremaining: 1.37s\n",
      "953:\tlearn: 0.0030660\ttotal: 27.9s\tremaining: 1.34s\n",
      "954:\tlearn: 0.0030658\ttotal: 27.9s\tremaining: 1.31s\n",
      "955:\tlearn: 0.0030656\ttotal: 27.9s\tremaining: 1.28s\n",
      "956:\tlearn: 0.0030654\ttotal: 27.9s\tremaining: 1.25s\n",
      "957:\tlearn: 0.0030652\ttotal: 28s\tremaining: 1.23s\n",
      "958:\tlearn: 0.0030640\ttotal: 28s\tremaining: 1.2s\n",
      "959:\tlearn: 0.0030640\ttotal: 28s\tremaining: 1.17s\n",
      "960:\tlearn: 0.0030549\ttotal: 28.1s\tremaining: 1.14s\n",
      "961:\tlearn: 0.0030541\ttotal: 28.1s\tremaining: 1.11s\n",
      "962:\tlearn: 0.0030534\ttotal: 28.1s\tremaining: 1.08s\n",
      "963:\tlearn: 0.0030532\ttotal: 28.1s\tremaining: 1.05s\n",
      "964:\tlearn: 0.0030517\ttotal: 28.2s\tremaining: 1.02s\n",
      "965:\tlearn: 0.0030511\ttotal: 28.2s\tremaining: 993ms\n",
      "966:\tlearn: 0.0030511\ttotal: 28.2s\tremaining: 964ms\n",
      "967:\tlearn: 0.0030511\ttotal: 28.3s\tremaining: 934ms\n",
      "968:\tlearn: 0.0030500\ttotal: 28.3s\tremaining: 905ms\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "969:\tlearn: 0.0030498\ttotal: 28.3s\tremaining: 876ms\n",
      "970:\tlearn: 0.0030488\ttotal: 28.3s\tremaining: 847ms\n",
      "971:\tlearn: 0.0030488\ttotal: 28.4s\tremaining: 817ms\n",
      "972:\tlearn: 0.0030440\ttotal: 28.4s\tremaining: 788ms\n",
      "973:\tlearn: 0.0030440\ttotal: 28.4s\tremaining: 759ms\n",
      "974:\tlearn: 0.0030419\ttotal: 28.5s\tremaining: 730ms\n",
      "975:\tlearn: 0.0030417\ttotal: 28.5s\tremaining: 701ms\n",
      "976:\tlearn: 0.0030415\ttotal: 28.5s\tremaining: 672ms\n",
      "977:\tlearn: 0.0030404\ttotal: 28.5s\tremaining: 642ms\n",
      "978:\tlearn: 0.0030404\ttotal: 28.6s\tremaining: 613ms\n",
      "979:\tlearn: 0.0030402\ttotal: 28.6s\tremaining: 584ms\n",
      "980:\tlearn: 0.0030402\ttotal: 28.6s\tremaining: 555ms\n",
      "981:\tlearn: 0.0030400\ttotal: 28.7s\tremaining: 525ms\n",
      "982:\tlearn: 0.0030400\ttotal: 28.7s\tremaining: 496ms\n",
      "983:\tlearn: 0.0030380\ttotal: 28.7s\tremaining: 467ms\n",
      "984:\tlearn: 0.0030353\ttotal: 28.7s\tremaining: 438ms\n",
      "985:\tlearn: 0.0030351\ttotal: 28.8s\tremaining: 408ms\n",
      "986:\tlearn: 0.0030351\ttotal: 28.8s\tremaining: 379ms\n",
      "987:\tlearn: 0.0030327\ttotal: 28.8s\tremaining: 350ms\n",
      "988:\tlearn: 0.0030327\ttotal: 28.9s\tremaining: 321ms\n",
      "989:\tlearn: 0.0030327\ttotal: 28.9s\tremaining: 292ms\n",
      "990:\tlearn: 0.0030325\ttotal: 28.9s\tremaining: 263ms\n",
      "991:\tlearn: 0.0030325\ttotal: 28.9s\tremaining: 233ms\n",
      "992:\tlearn: 0.0030324\ttotal: 29s\tremaining: 204ms\n",
      "993:\tlearn: 0.0030282\ttotal: 29s\tremaining: 175ms\n",
      "994:\tlearn: 0.0030236\ttotal: 29s\tremaining: 146ms\n",
      "995:\tlearn: 0.0030234\ttotal: 29.1s\tremaining: 117ms\n",
      "996:\tlearn: 0.0030234\ttotal: 29.1s\tremaining: 87.5ms\n",
      "997:\tlearn: 0.0030208\ttotal: 29.1s\tremaining: 58.4ms\n",
      "998:\tlearn: 0.0030199\ttotal: 29.2s\tremaining: 29.2ms\n",
      "999:\tlearn: 0.0030199\ttotal: 29.2s\tremaining: 0us\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<catboost.core.CatBoostClassifier at 0x1e4db4c5b88>"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cat.fit(train_df, train_target)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "answers = cat.predict(test_df)\n",
    "answers = pd.Series(answers, name='IN_TREINEIRO')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "result_df = pd.concat((test_inscricao, answers), axis=1)\n",
    "result_df.set_index('NU_INSCRICAO', inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "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>IN_TREINEIRO</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NU_INSCRICAO</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>ba0cc30ba34e7a46764c09dfc38ed83d15828897</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>177f281c68fa032aedbd842a745da68490926cd2</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6cf0d8b97597d7625cdedc7bdb6c0f052286c334</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5c356d810fa57671402502cd0933e5601a2ebf1e</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>df47c07bd881c2db3f38c6048bf77c132ad0ceb3</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                          IN_TREINEIRO\n",
       "NU_INSCRICAO                                          \n",
       "ba0cc30ba34e7a46764c09dfc38ed83d15828897             0\n",
       "177f281c68fa032aedbd842a745da68490926cd2             0\n",
       "6cf0d8b97597d7625cdedc7bdb6c0f052286c334             1\n",
       "5c356d810fa57671402502cd0933e5601a2ebf1e             0\n",
       "df47c07bd881c2db3f38c6048bf77c132ad0ceb3             0"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "result_df.to_csv('answer.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "import eli5\n",
    "from eli5.sklearn import PermutationImportance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "    <style>\n",
       "    table.eli5-weights tr:hover {\n",
       "        filter: brightness(85%);\n",
       "    }\n",
       "</style>\n",
       "\n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "        <table class=\"eli5-weights eli5-feature-importances\" style=\"border-collapse: collapse; border: none; margin-top: 0em; table-layout: auto;\">\n",
       "    <thead>\n",
       "    <tr style=\"border: none;\">\n",
       "        <th style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">Weight</th>\n",
       "        <th style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">Feature</th>\n",
       "    </tr>\n",
       "    </thead>\n",
       "    <tbody>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 80.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.4910\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                TP_ST_CONCLUSAO\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 90.78%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.1625\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                NU_IDADE\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 92.45%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.1220\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                TP_ESCOLA\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 97.74%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0218\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                Q001\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 97.88%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0198\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                CO_UF_RESIDENCIA\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 98.19%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0159\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                SG_UF_RESIDENCIA\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 98.19%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0159\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                NU_NOTA_CN\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 98.36%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0138\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                NU_NOTA_CH\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 98.40%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0133\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                TP_ANO_CONCLUIU\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 98.41%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0132\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                Q002\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 98.46%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0126\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                Q006\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 98.58%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0112\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                Q047\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 98.64%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0105\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                NU_NOTA_COMP5\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 98.69%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0100\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                Q026\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 98.86%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0082\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                NU_NOTA_LC\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 98.99%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0069\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                NU_NOTA_REDACAO\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 99.12%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0056\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                Q027\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 99.14%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0054\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                Q024\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 99.23%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0047\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                IN_GESTANTE\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 99.23%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0047\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                NU_NOTA_COMP4\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "    \n",
       "        \n",
       "            <tr style=\"background-color: hsl(120, 100.00%, 99.23%); border: none;\">\n",
       "                <td colspan=\"2\" style=\"padding: 0 0.5em 0 0.5em; text-align: center; border: none; white-space: nowrap;\">\n",
       "                    <i>&hellip; 22 more &hellip;</i>\n",
       "                </td>\n",
       "            </tr>\n",
       "        \n",
       "    \n",
       "    </tbody>\n",
       "</table>\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "\n"
      ],
      "text/plain": [
       "Explanation(estimator='<catboost.core.CatBoostClassifier object at 0x000001E4DB4C5B88>', description='CatBoost feature importances; \\nvalues are numbers 0 <= x <= 1; all values sum to 1.', error=None, method='feature importances', is_regression=False, targets=None, feature_importances=FeatureImportances(importances=[FeatureWeight(feature='TP_ST_CONCLUSAO', weight=0.4909985, std=None, value=None), FeatureWeight(feature='NU_IDADE', weight=0.16246714, std=None, value=None), FeatureWeight(feature='TP_ESCOLA', weight=0.121975906, std=None, value=None), FeatureWeight(feature='Q001', weight=0.02177252, std=None, value=None), FeatureWeight(feature='CO_UF_RESIDENCIA', weight=0.01983183, std=None, value=None), FeatureWeight(feature='SG_UF_RESIDENCIA', weight=0.015892457, std=None, value=None), FeatureWeight(feature='NU_NOTA_CN', weight=0.015866447, std=None, value=None), FeatureWeight(feature='NU_NOTA_CH', weight=0.013834754, std=None, value=None), FeatureWeight(feature='TP_ANO_CONCLUIU', weight=0.0132526895, std=None, value=None), FeatureWeight(feature='Q002', weight=0.013212435, std=None, value=None), FeatureWeight(feature='Q006', weight=0.012550319, std=None, value=None), FeatureWeight(feature='Q047', weight=0.011239753, std=None, value=None), FeatureWeight(feature='NU_NOTA_COMP5', weight=0.010506468, std=None, value=None), FeatureWeight(feature='Q026', weight=0.010044923, std=None, value=None), FeatureWeight(feature='NU_NOTA_LC', weight=0.008190334, std=None, value=None), FeatureWeight(feature='NU_NOTA_REDACAO', weight=0.0068564676, std=None, value=None), FeatureWeight(feature='Q027', weight=0.005636319, std=None, value=None), FeatureWeight(feature='Q024', weight=0.0054389513, std=None, value=None), FeatureWeight(feature='IN_GESTANTE', weight=0.0047191717, std=None, value=None), FeatureWeight(feature='NU_NOTA_COMP4', weight=0.004716401, std=None, value=None)], remaining=22), decision_tree=None, highlight_spaces=None, transition_features=None, image=None)"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "eli5.explain_weights(cat)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x684 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from matplotlib import pyplot as plt\n",
    "from pdpbox import pdp, get_dataset, info_plots\n",
    "\n",
    "pdp_conclusao = pdp.pdp_isolate(model=cat, \n",
    "                                dataset=train_df, \n",
    "                                model_features=train_df.columns, \n",
    "                                feature='NU_IDADE')\n",
    "\n",
    "# plot it\n",
    "pdp.pdp_plot(pdp_conclusao, 'Idade')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x684 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "pdp_conclusao = pdp.pdp_isolate(model=cat, \n",
    "                                dataset=train_df, \n",
    "                                model_features=train_df.columns, \n",
    "                                feature='TP_ESCOLA')\n",
    "\n",
    "# plot it\n",
    "pdp.pdp_plot(pdp_conclusao, 'Escola')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x684 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "pdp_conclusao = pdp.pdp_isolate(model=cat, \n",
    "                                dataset=train_df, \n",
    "                                model_features=train_df.columns, \n",
    "                                feature='TP_ST_CONCLUSAO')\n",
    "\n",
    "# plot it\n",
    "pdp.pdp_plot(pdp_conclusao, 'Conclusao')\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}