00_Data_PreProcessing.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
    "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5"
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "sns.set(font_scale=2.2)\n",
    "plt.style.use('seaborn')\n",
    "\n",
    "from sklearn.preprocessing import LabelEncoder, MinMaxScaler, StandardScaler, OneHotEncoder\n",
    "from sklearn.model_selection import StratifiedKFold, train_test_split, ShuffleSplit\n",
    "from sklearn.metrics import f1_score\n",
    "import itertools\n",
    "import lightgbm as lgb\n",
    "import xgboost as xgb\n",
    "from xgboost import XGBClassifier\n",
    "import shap\n",
    "from tqdm import tqdm\n",
    "import warnings \n",
    "warnings.filterwarnings('ignore')\n",
    "import time"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "da0c2dd917069d712ebf552c0a636d3085a8a93e"
   },
   "source": [
    "# 1. Check datasets"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "d62fec9476338b5bd9238d0ef0b1eede68d64172"
   },
   "source": [
    "## 1.1 Read dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "_uuid": "f7c77f102b5e10e76e3e49a6027271cb3e776b2a"
   },
   "outputs": [],
   "source": [
    "df_train = pd.read_csv('train.csv')\n",
    "df_test = pd.read_csv('test.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "_uuid": "51c5dd9205fee608be90dc89faa6a9d7efc84a07"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "df_train shape: (9557, 143)    df_test shape:  (23856, 142)\n"
     ]
    }
   ],
   "source": [
    "print('df_train shape:', df_train.shape, '  ', 'df_test shape: ', df_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "_uuid": "959cb5a1e6fb528e4260e9514b6c017203f5077a"
   },
   "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>Id</th>\n",
       "      <th>v2a1</th>\n",
       "      <th>hacdor</th>\n",
       "      <th>rooms</th>\n",
       "      <th>hacapo</th>\n",
       "      <th>v14a</th>\n",
       "      <th>refrig</th>\n",
       "      <th>v18q</th>\n",
       "      <th>v18q1</th>\n",
       "      <th>r4h1</th>\n",
       "      <th>...</th>\n",
       "      <th>SQBescolari</th>\n",
       "      <th>SQBage</th>\n",
       "      <th>SQBhogar_total</th>\n",
       "      <th>SQBedjefe</th>\n",
       "      <th>SQBhogar_nin</th>\n",
       "      <th>SQBovercrowding</th>\n",
       "      <th>SQBdependency</th>\n",
       "      <th>SQBmeaned</th>\n",
       "      <th>agesq</th>\n",
       "      <th>Target</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ID_279628684</td>\n",
       "      <td>190000.0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>100</td>\n",
       "      <td>1849</td>\n",
       "      <td>1</td>\n",
       "      <td>100</td>\n",
       "      <td>0</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>1849</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ID_f29eb3ddd</td>\n",
       "      <td>135000.0</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>144</td>\n",
       "      <td>4489</td>\n",
       "      <td>1</td>\n",
       "      <td>144</td>\n",
       "      <td>0</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>64.0</td>\n",
       "      <td>144.0</td>\n",
       "      <td>4489</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ID_68de51c94</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>8</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>121</td>\n",
       "      <td>8464</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.250000</td>\n",
       "      <td>64.0</td>\n",
       "      <td>121.0</td>\n",
       "      <td>8464</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ID_d671db89c</td>\n",
       "      <td>180000.0</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>81</td>\n",
       "      <td>289</td>\n",
       "      <td>16</td>\n",
       "      <td>121</td>\n",
       "      <td>4</td>\n",
       "      <td>1.777778</td>\n",
       "      <td>1.0</td>\n",
       "      <td>121.0</td>\n",
       "      <td>289</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>ID_d56d6f5f5</td>\n",
       "      <td>180000.0</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>121</td>\n",
       "      <td>1369</td>\n",
       "      <td>16</td>\n",
       "      <td>121</td>\n",
       "      <td>4</td>\n",
       "      <td>1.777778</td>\n",
       "      <td>1.0</td>\n",
       "      <td>121.0</td>\n",
       "      <td>1369</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 143 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "             Id      v2a1  hacdor  rooms  hacapo  v14a  refrig  v18q  v18q1  \\\n",
       "0  ID_279628684  190000.0       0      3       0     1       1     0    NaN   \n",
       "1  ID_f29eb3ddd  135000.0       0      4       0     1       1     1    1.0   \n",
       "2  ID_68de51c94       NaN       0      8       0     1       1     0    NaN   \n",
       "3  ID_d671db89c  180000.0       0      5       0     1       1     1    1.0   \n",
       "4  ID_d56d6f5f5  180000.0       0      5       0     1       1     1    1.0   \n",
       "\n",
       "   r4h1  ...  SQBescolari  SQBage  SQBhogar_total  SQBedjefe  SQBhogar_nin  \\\n",
       "0     0  ...          100    1849               1        100             0   \n",
       "1     0  ...          144    4489               1        144             0   \n",
       "2     0  ...          121    8464               1          0             0   \n",
       "3     0  ...           81     289              16        121             4   \n",
       "4     0  ...          121    1369              16        121             4   \n",
       "\n",
       "   SQBovercrowding  SQBdependency  SQBmeaned  agesq  Target  \n",
       "0         1.000000            0.0      100.0   1849       4  \n",
       "1         1.000000           64.0      144.0   4489       4  \n",
       "2         0.250000           64.0      121.0   8464       4  \n",
       "3         1.777778            1.0      121.0    289       4  \n",
       "4         1.777778            1.0      121.0   1369       4  \n",
       "\n",
       "[5 rows x 143 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "_uuid": "14d4a2ed4c13b02a7be164a61443232cdfff3fb4"
   },
   "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>Id</th>\n",
       "      <th>v2a1</th>\n",
       "      <th>hacdor</th>\n",
       "      <th>rooms</th>\n",
       "      <th>hacapo</th>\n",
       "      <th>v14a</th>\n",
       "      <th>refrig</th>\n",
       "      <th>v18q</th>\n",
       "      <th>v18q1</th>\n",
       "      <th>r4h1</th>\n",
       "      <th>...</th>\n",
       "      <th>age</th>\n",
       "      <th>SQBescolari</th>\n",
       "      <th>SQBage</th>\n",
       "      <th>SQBhogar_total</th>\n",
       "      <th>SQBedjefe</th>\n",
       "      <th>SQBhogar_nin</th>\n",
       "      <th>SQBovercrowding</th>\n",
       "      <th>SQBdependency</th>\n",
       "      <th>SQBmeaned</th>\n",
       "      <th>agesq</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ID_2f6873615</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>16</td>\n",
       "      <td>9</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>2.25</td>\n",
       "      <td>0.25</td>\n",
       "      <td>272.25</td>\n",
       "      <td>16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ID_1c78846d2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>41</td>\n",
       "      <td>256</td>\n",
       "      <td>1681</td>\n",
       "      <td>9</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>2.25</td>\n",
       "      <td>0.25</td>\n",
       "      <td>272.25</td>\n",
       "      <td>1681</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ID_e5442cf6a</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>41</td>\n",
       "      <td>289</td>\n",
       "      <td>1681</td>\n",
       "      <td>9</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>2.25</td>\n",
       "      <td>0.25</td>\n",
       "      <td>272.25</td>\n",
       "      <td>1681</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ID_a8db26a79</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>14</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>59</td>\n",
       "      <td>256</td>\n",
       "      <td>3481</td>\n",
       "      <td>1</td>\n",
       "      <td>256</td>\n",
       "      <td>0</td>\n",
       "      <td>1.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>256.00</td>\n",
       "      <td>3481</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>ID_a62966799</td>\n",
       "      <td>175000.0</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>18</td>\n",
       "      <td>121</td>\n",
       "      <td>324</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.25</td>\n",
       "      <td>64.00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>324</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 142 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "             Id      v2a1  hacdor  rooms  hacapo  v14a  refrig  v18q  v18q1  \\\n",
       "0  ID_2f6873615       NaN       0      5       0     1       1     0    NaN   \n",
       "1  ID_1c78846d2       NaN       0      5       0     1       1     0    NaN   \n",
       "2  ID_e5442cf6a       NaN       0      5       0     1       1     0    NaN   \n",
       "3  ID_a8db26a79       NaN       0     14       0     1       1     1    1.0   \n",
       "4  ID_a62966799  175000.0       0      4       0     1       1     1    1.0   \n",
       "\n",
       "   r4h1  ...  age  SQBescolari  SQBage  SQBhogar_total  SQBedjefe  \\\n",
       "0     1  ...    4            0      16               9          0   \n",
       "1     1  ...   41          256    1681               9          0   \n",
       "2     1  ...   41          289    1681               9          0   \n",
       "3     0  ...   59          256    3481               1        256   \n",
       "4     0  ...   18          121     324               1          0   \n",
       "\n",
       "   SQBhogar_nin  SQBovercrowding  SQBdependency  SQBmeaned  agesq  \n",
       "0             1             2.25           0.25     272.25     16  \n",
       "1             1             2.25           0.25     272.25   1681  \n",
       "2             1             2.25           0.25     272.25   1681  \n",
       "3             0             1.00           0.00     256.00   3481  \n",
       "4             1             0.25          64.00        NaN    324  \n",
       "\n",
       "[5 rows x 142 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_test.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "e7068c40ca5ab2493cdc7379f6f7da1856a8d7d1"
   },
   "source": [
    "## 1.2 Make description df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "_kg_hide-input": true,
    "_uuid": "ed6d0f1837d5d8077db467067525f93c560db587"
   },
   "outputs": [],
   "source": [
    "description = [\n",
    "(\"v2a1\",\" Monthly rent payment\"),\n",
    "(\"hacdor\",\" =1 Overcrowding by bedrooms\"),\n",
    "(\"rooms\",\"  number of all rooms in the house\"),\n",
    "(\"hacapo\",\" =1 Overcrowding by rooms\"),\n",
    "(\"v14a\",\" =1 has toilet in the household\"),\n",
    "(\"refrig\",\" =1 if the household has refrigerator\"),\n",
    "(\"v18q\",\" owns a tablet\"),\n",
    "(\"v18q1\",\" number of tablets household owns\"),\n",
    "(\"r4h1\",\" Males younger than 12 years of age\"),\n",
    "(\"r4h2\",\" Males 12 years of age and older\"),\n",
    "(\"r4h3\",\" Total males in the household\"),\n",
    "(\"r4m1\",\" Females younger than 12 years of age\"),\n",
    "(\"r4m2\",\" Females 12 years of age and older\"),\n",
    "(\"r4m3\",\" Total females in the household\"),\n",
    "(\"r4t1\",\" persons younger than 12 years of age\"),\n",
    "(\"r4t2\",\" persons 12 years of age and older\"),\n",
    "(\"r4t3\",\" Total persons in the household\"),\n",
    "(\"tamhog\",\" size of the household\"),\n",
    "(\"tamviv\",\" number of persons living in the household\"),\n",
    "(\"escolari\",\" years of schooling\"),\n",
    "(\"rez_esc\",\" Years behind in school\"),\n",
    "(\"hhsize\",\" household size\"),\n",
    "(\"paredblolad\",\" =1 if predominant material on the outside wall is block or brick\"),\n",
    "(\"paredzocalo\",\" =1 if predominant material on the outside wall is socket (wood, zinc or absbesto\"),\n",
    "(\"paredpreb\",\" =1 if predominant material on the outside wall is prefabricated or cement\"),\n",
    "(\"pareddes\",\" =1 if predominant material on the outside wall is waste material\"),\n",
    "(\"paredmad\",\" =1 if predominant material on the outside wall is wood\"),\n",
    "(\"paredzinc\",\" =1 if predominant material on the outside wall is zink\"),\n",
    "(\"paredfibras\",\" =1 if predominant material on the outside wall is natural fibers\"),\n",
    "(\"paredother\",\" =1 if predominant material on the outside wall is other\"),\n",
    "(\"pisomoscer\",\" =1 if predominant material on the floor is mosaic ceramic   terrazo\"),\n",
    "(\"pisocemento\",\" =1 if predominant material on the floor is cement\"),\n",
    "(\"pisoother\",\" =1 if predominant material on the floor is other\"),\n",
    "(\"pisonatur\",\" =1 if predominant material on the floor is  natural material\"),\n",
    "(\"pisonotiene\",\" =1 if no floor at the household\"),\n",
    "(\"pisomadera\",\" =1 if predominant material on the floor is wood\"),\n",
    "(\"techozinc\",\" =1 if predominant material on the roof is metal foil or zink\"),\n",
    "(\"techoentrepiso\",\" =1 if predominant material on the roof is fiber cement,   mezzanine \"),\n",
    "(\"techocane\",\" =1 if predominant material on the roof is natural fibers\"),\n",
    "(\"techootro\",\" =1 if predominant material on the roof is other\"),\n",
    "(\"cielorazo\",\" =1 if the house has ceiling\"),\n",
    "(\"abastaguadentro\",\" =1 if water provision inside the dwelling\"),\n",
    "(\"abastaguafuera\",\" =1 if water provision outside the dwelling\"),\n",
    "(\"abastaguano\",\" =1 if no water provision\"),\n",
    "(\"public\",\" =1 electricity from CNFL,  ICE, ESPH/JASEC\"),\n",
    "(\"planpri\",\" =1 electricity from private plant\"),\n",
    "(\"noelec\",\" =1 no electricity in the dwelling\"),\n",
    "(\"coopele\",\" =1 electricity from cooperative\"),\n",
    "(\"sanitario1\",\" =1 no toilet in the dwelling\"),\n",
    "(\"sanitario2\",\" =1 toilet connected to sewer or cesspool\"),\n",
    "(\"sanitario3\",\" =1 toilet connected to  septic tank\"),\n",
    "(\"sanitario5\",\" =1 toilet connected to black hole or letrine\"),\n",
    "(\"sanitario6\",\" =1 toilet connected to other system\"),\n",
    "(\"energcocinar1\",\" =1 no main source of energy used for cooking (no kitchen)\"),\n",
    "(\"energcocinar2\",\" =1 main source of energy used for cooking electricity\"),\n",
    "(\"energcocinar3\",\" =1 main source of energy used for cooking gas\"),\n",
    "(\"energcocinar4\",\" =1 main source of energy used for cooking wood charcoal\"),\n",
    "(\"elimbasu1\",\" =1 if rubbish disposal mainly by tanker truck\"),\n",
    "(\"elimbasu2\",\" =1 if rubbish disposal mainly by botan hollow or buried\"),\n",
    "(\"elimbasu3\",\" =1 if rubbish disposal mainly by burning\"),\n",
    "(\"elimbasu4\",\" =1 if rubbish disposal mainly by throwing in an unoccupied space\"),\n",
    "(\"elimbasu5\",\" =1 if rubbish disposal mainly by throwing in river,   creek or sea\"),\n",
    "(\"elimbasu6\",\" =1 if rubbish disposal mainly other\"),\n",
    "(\"epared1\",\" =1 if walls are bad\"),\n",
    "(\"epared2\",\" =1 if walls are regular\"),\n",
    "(\"epared3\",\" =1 if walls are good\"),\n",
    "(\"etecho1\",\" =1 if roof are bad\"),\n",
    "(\"etecho2\",\" =1 if roof are regular\"),\n",
    "(\"etecho3\",\" =1 if roof are good\"),\n",
    "(\"eviv1\",\" =1 if floor are bad\"),\n",
    "(\"eviv2\",\" =1 if floor are regular\"),\n",
    "(\"eviv3\",\" =1 if floor are good\"),\n",
    "(\"dis\",\" =1 if disable person\"),\n",
    "(\"male\",\" =1 if male\"),\n",
    "(\"female\",\" =1 if female\"),\n",
    "(\"estadocivil1\",\" =1 if less than 10 years old\"),\n",
    "(\"estadocivil2\",\" =1 if free or coupled uunion\"),\n",
    "(\"estadocivil3\",\" =1 if married\"),\n",
    "(\"estadocivil4\",\" =1 if divorced\"),\n",
    "(\"estadocivil5\",\" =1 if separated\"),\n",
    "(\"estadocivil6\",\" =1 if widow/er\"),\n",
    "(\"estadocivil7\",\" =1 if single\"),\n",
    "(\"parentesco1\",\" =1 if household head\"),\n",
    "(\"parentesco2\",\" =1 if spouse/partner\"),\n",
    "(\"parentesco3\",\" =1 if son/doughter\"),\n",
    "(\"parentesco4\",\" =1 if stepson/doughter\"),\n",
    "(\"parentesco5\",\" =1 if son/doughter in law\"),\n",
    "(\"parentesco6\",\" =1 if grandson/doughter\"),\n",
    "(\"parentesco7\",\" =1 if mother/father\"),\n",
    "(\"parentesco8\",\" =1 if father/mother in law\"),\n",
    "(\"parentesco9\",\" =1 if brother/sister\"),\n",
    "(\"parentesco10\",\" =1 if brother/sister in law\"),\n",
    "(\"parentesco11\",\" =1 if other family member\"),\n",
    "(\"parentesco12\",\" =1 if other non family member\"),\n",
    "(\"idhogar\",\" Household level identifier\"),\n",
    "(\"hogar_nin\",\" Number of children 0 to 19 in household\"),\n",
    "(\"hogar_adul\",\" Number of adults in household\"),\n",
    "(\"hogar_mayor\",\" # of individuals 65+ in the household\"),\n",
    "(\"hogar_total\",\" # of total individuals in the household\"),\n",
    "(\"dependency\",\" Dependency rate\"),\n",
    "(\"edjefe\",\" years of education of male head of household\"),\n",
    "(\"edjefa\",\" years of education of female head of household\"),\n",
    "(\"meaneduc\",\"average years of education for adults (18+)\"),\n",
    "(\"instlevel1\",\" =1 no level of education\"),\n",
    "(\"instlevel2\",\" =1 incomplete primary\"),\n",
    "(\"instlevel3\",\" =1 complete primary\"),\n",
    "(\"instlevel4\",\" =1 incomplete academic secondary level\"),\n",
    "(\"instlevel5\",\" =1 complete academic secondary level\"),\n",
    "(\"instlevel6\",\" =1 incomplete technical secondary level\"),\n",
    "(\"instlevel7\",\" =1 complete technical secondary level\"),\n",
    "(\"instlevel8\",\" =1 undergraduate and higher education\"),\n",
    "(\"instlevel9\",\" =1 postgraduate higher education\"),\n",
    "(\"bedrooms\",\" number of bedrooms\"),\n",
    "(\"overcrowding\",\" # persons per room\"),\n",
    "(\"tipovivi1\",\" =1 own and fully paid house\"),\n",
    "(\"tipovivi2\",\" =1 own,   paying in installments\"),\n",
    "(\"tipovivi3\",\" =1 rented\"),\n",
    "(\"tipovivi4\",\" =1 precarious\"),\n",
    "(\"tipovivi5\",\" =1 other(assigned\"),\n",
    "(\"computer\",\" =1 if the household has notebook or desktop computer,   borrowed)\"),\n",
    "(\"television\",\" =1 if the household has TV\"),\n",
    "(\"mobilephone\",\" =1 if mobile phone\"),\n",
    "(\"qmobilephone\",\" # of mobile phones\"),\n",
    "(\"lugar1\",\" =1 region Central\"),\n",
    "(\"lugar2\",\" =1 region Chorotega\"),\n",
    "(\"lugar3\",\" =1 region PacÃƒÂ­fico central\"),\n",
    "(\"lugar4\",\" =1 region Brunca\"),\n",
    "(\"lugar5\",\" =1 region Huetar AtlÃƒÂ¡ntica\"),\n",
    "(\"lugar6\",\" =1 region Huetar Norte\"),\n",
    "(\"area1\",\" =1 zona urbana\"),\n",
    "(\"area2\",\" =2 zona rural\"),\n",
    "(\"age\",\" Age in years\"),\n",
    "(\"SQBescolari\",\" escolari squared\"),\n",
    "(\"SQBage\",\" age squared\"),\n",
    "(\"SQBhogar_total\",\" hogar_total squared\"),\n",
    "(\"SQBedjefe\",\" edjefe squared\"),\n",
    "(\"SQBhogar_nin\",\" hogar_nin squared\"),\n",
    "(\"SQBovercrowding\",\" overcrowding squared\"),\n",
    "(\"SQBdependency\",\" dependency squared\"),\n",
    "(\"SQBmeaned\",\" meaned squared\"),\n",
    "(\"agesq\",\" Age squared\"),]\n",
    "\n",
    "description = pd.DataFrame(description, columns=['varname', 'description'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "_uuid": "f77544a2816dd03612a5cd72008821d896bb9fb7",
    "scrolled": true
   },
   "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>varname</th>\n",
       "      <th>description</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>v2a1</td>\n",
       "      <td>Monthly rent payment</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>hacdor</td>\n",
       "      <td>=1 Overcrowding by bedrooms</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>rooms</td>\n",
       "      <td>number of all rooms in the house</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>hacapo</td>\n",
       "      <td>=1 Overcrowding by rooms</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>v14a</td>\n",
       "      <td>=1 has toilet in the household</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>136</th>\n",
       "      <td>SQBhogar_nin</td>\n",
       "      <td>hogar_nin squared</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>137</th>\n",
       "      <td>SQBovercrowding</td>\n",
       "      <td>overcrowding squared</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>138</th>\n",
       "      <td>SQBdependency</td>\n",
       "      <td>dependency squared</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>139</th>\n",
       "      <td>SQBmeaned</td>\n",
       "      <td>meaned squared</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>140</th>\n",
       "      <td>agesq</td>\n",
       "      <td>Age squared</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>141 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "             varname                         description\n",
       "0               v2a1                Monthly rent payment\n",
       "1             hacdor         =1 Overcrowding by bedrooms\n",
       "2              rooms    number of all rooms in the house\n",
       "3             hacapo            =1 Overcrowding by rooms\n",
       "4               v14a      =1 has toilet in the household\n",
       "..               ...                                 ...\n",
       "136     SQBhogar_nin                   hogar_nin squared\n",
       "137  SQBovercrowding                overcrowding squared\n",
       "138    SQBdependency                  dependency squared\n",
       "139        SQBmeaned                      meaned squared\n",
       "140            agesq                         Age squared\n",
       "\n",
       "[141 rows x 2 columns]"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "description"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "eb86cfc0f9a39d2a18d8d49d559c6ef2dcbbeb7f"
   },
   "source": [
    "## 1.3 Check null data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "_uuid": "36bf2f52050d81eff74c993b9294cb8618d9415a"
   },
   "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>Total</th>\n",
       "      <th>Percent</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>rez_esc</th>\n",
       "      <td>7928</td>\n",
       "      <td>82.954902</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>v18q1</th>\n",
       "      <td>7342</td>\n",
       "      <td>76.823271</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>v2a1</th>\n",
       "      <td>6860</td>\n",
       "      <td>71.779847</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>meaneduc</th>\n",
       "      <td>5</td>\n",
       "      <td>0.052318</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SQBmeaned</th>\n",
       "      <td>5</td>\n",
       "      <td>0.052318</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>techozinc</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>techoentrepiso</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>techocane</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>techootro</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cielorazo</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>abastaguadentro</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sanitario3</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>abastaguafuera</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>abastaguano</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>public</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>planpri</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>noelec</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>coopele</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sanitario1</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sanitario2</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 Total    Percent\n",
       "rez_esc           7928  82.954902\n",
       "v18q1             7342  76.823271\n",
       "v2a1              6860  71.779847\n",
       "meaneduc             5   0.052318\n",
       "SQBmeaned            5   0.052318\n",
       "techozinc            0   0.000000\n",
       "techoentrepiso       0   0.000000\n",
       "techocane            0   0.000000\n",
       "techootro            0   0.000000\n",
       "cielorazo            0   0.000000\n",
       "abastaguadentro      0   0.000000\n",
       "sanitario3           0   0.000000\n",
       "abastaguafuera       0   0.000000\n",
       "abastaguano          0   0.000000\n",
       "public               0   0.000000\n",
       "planpri              0   0.000000\n",
       "noelec               0   0.000000\n",
       "coopele              0   0.000000\n",
       "sanitario1           0   0.000000\n",
       "sanitario2           0   0.000000"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "total = df_train.isnull().sum().sort_values(ascending=False)\n",
    "percent = 100 * (df_train.isnull().sum() / df_train.isnull().count()).sort_values(ascending=False)\n",
    "missing_df = pd.concat([total, percent], axis=1, keys=['Total', 'Percent'])\n",
    "\n",
    "missing_df.head(20)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "509bbb24b15a3dc8fa6a69bdfd14de313b64ae27"
   },
   "source": [
    "## 1.4 Filll missing values"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "fa651eecfb831a687227a6702f3e2a8858ca2361"
   },
   "source": [
    "- Below cell is from this kernel. https://www.kaggle.com/skooch/lgbm-w-random-split-2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "_uuid": "0c79c0249a4815ae829bbb1af0f59ac393576683"
   },
   "outputs": [],
   "source": [
    "# if education is \"yes\" and person is head of household, fill with escolari\n",
    "df_train.loc[(df_train['edjefa'] == \"yes\") & (df_train['parentesco1'] == 1), \"edjefa\"] = df_train.loc[(df_train['edjefa'] == \"yes\") & (df_train['parentesco1'] == 1), \"escolari\"]\n",
    "df_train.loc[(df_train['edjefe'] == \"yes\") & (df_train['parentesco1'] == 1), \"edjefe\"] = df_train.loc[(df_train['edjefe'] == \"yes\") & (df_train['parentesco1'] == 1), \"escolari\"]\n",
    "\n",
    "df_test.loc[(df_test['edjefa'] == \"yes\") & (df_test['parentesco1'] == 1), \"edjefa\"] = df_test.loc[(df_test['edjefa'] == \"yes\") & (df_test['parentesco1'] == 1), \"escolari\"]\n",
    "df_test.loc[(df_test['edjefe'] == \"yes\") & (df_test['parentesco1'] == 1), \"edjefe\"] = df_test.loc[(df_test['edjefe'] == \"yes\") & (df_test['parentesco1'] == 1), \"escolari\"]\n",
    "\n",
    "# this field is supposed to be interaction between gender and escolari, but it isn't clear what \"yes\" means, let's fill it with 4\n",
    "df_train.loc[df_train['edjefa'] == \"yes\", \"edjefa\"] = 4\n",
    "df_train.loc[df_train['edjefe'] == \"yes\", \"edjefe\"] = 4\n",
    "\n",
    "df_test.loc[df_test['edjefa'] == \"yes\", \"edjefa\"] = 4\n",
    "df_test.loc[df_test['edjefe'] == \"yes\", \"edjefe\"] = 4\n",
    "\n",
    "# create feature with max education of either head of household\n",
    "df_train['edjef'] = np.max(df_train[['edjefa','edjefe']], axis=1)\n",
    "df_test['edjef'] = np.max(df_test[['edjefa','edjefe']], axis=1)\n",
    "\n",
    "# fix some inconsistencies in the data - some rows indicate both that the household does and does not have a toilet, \n",
    "# if there is no water we'll assume they do not\n",
    "df_train.loc[(df_train.v14a ==  1) & (df_train.sanitario1 ==  1) & (df_train.abastaguano == 0), \"v14a\"] = 0\n",
    "df_train.loc[(df_train.v14a ==  1) & (df_train.sanitario1 ==  1) & (df_train.abastaguano == 0), \"sanitario1\"] = 0\n",
    "\n",
    "df_test.loc[(df_test.v14a ==  1) & (df_test.sanitario1 ==  1) & (df_test.abastaguano == 0), \"v14a\"] = 0\n",
    "df_test.loc[(df_test.v14a ==  1) & (df_test.sanitario1 ==  1) & (df_test.abastaguano == 0), \"sanitario1\"] = 0"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "200a76acc1cd22e7726fa9c97d5340ae353d0f07"
   },
   "source": [
    "### rez_esz, SQBmeaned\n",
    "- rez_esc : Years behind in school -> filled with 0\n",
    "- SQBmeaned : square of the mean years of education of adults (>=18) in the household\n",
    "agesq, Age squared -> same with rez_esc -> filled with 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "_uuid": "127dba2fefd3af6fd04630d9f1946377c3669fca"
   },
   "outputs": [],
   "source": [
    "df_train['rez_esc'].fillna(0, inplace=True)\n",
    "df_test['rez_esc'].fillna(0, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "_uuid": "a5aaa0a807b0bdfbd721742fa490d50077eb254b"
   },
   "outputs": [],
   "source": [
    "df_train['SQBmeaned'].fillna(0, inplace=True)\n",
    "df_test['SQBmeaned'].fillna(0, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "a4a4b7edde6a907f8689ebc9c9084196dad7aee6"
   },
   "source": [
    "### meaneduc\n",
    "- meaneduc: average years of education for adults (18+) -> filled with 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "_uuid": "a16e61ef3f2cca22fc16b52a8bae02f06b86ecb6"
   },
   "outputs": [],
   "source": [
    "df_train['meaneduc'].fillna(0, inplace=True)\n",
    "df_test['meaneduc'].fillna(0, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "7aa115c1ad3131ee08d25f2f70a29feede9172b4"
   },
   "source": [
    "### v18q1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "600b5f04fd584c3e8b6924292548d40e8f41155c"
   },
   "source": [
    "- v18q1: number of tablets household owns -> if v18q(Do you own a tablet?)  == 1, there are some values. If not, only NaN values in v18q1. See below 3 cells."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "_uuid": "dc4dcb27b06e6edcf58d0e164b67c056239939e0"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    7342\n",
       "1    2215\n",
       "Name: v18q, dtype: int64"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train['v18q'].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "31dd039e17ec249a681c0f9a7259d55d8d21dae0"
   },
   "source": [
    "- v18q1: number of tablets household owns -> if v18q  == 1, there are some values. If not, only NaN values there. See below two cells."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "_uuid": "8369998ea4a764a7e1ec93ec020e80a07721f386"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.0    1586\n",
       "2.0     444\n",
       "3.0     129\n",
       "4.0      37\n",
       "5.0      13\n",
       "6.0       6\n",
       "Name: v18q1, dtype: int64"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train.loc[df_train['v18q'] == 1, 'v18q1'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "_uuid": "1af5623dc1c128f4f9f9902255c42ffd2e48da8f"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Series([], Name: v18q1, dtype: int64)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train.loc[df_train['v18q'] == 0, 'v18q1'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "_uuid": "97be2ff581a3e1a28a47c7636476bc04ad05e30f"
   },
   "outputs": [],
   "source": [
    "df_train['v18q1'].fillna(0, inplace=True)\n",
    "df_test['v18q1'].fillna(0, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "092e8fd58e055c60fd7ff93bb36d33f976cf878e"
   },
   "source": [
    "- v2a1: number of tablets household owns -> if tipovivi3(rented?)  == 1, there are some values. If not, there are also some values.\n",
    "- NaN value could be replaced by 0."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "_uuid": "32ccf926a9ae978a69066c4e92eabb594ac0a934",
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    7821\n",
       "1    1736\n",
       "Name: tipovivi3, dtype: int64"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train['tipovivi3'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "_uuid": "4ac878b73107bc2a3fc899e3a6f08814d7e09461",
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x396 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.kdeplot(df_train.loc[df_train['tipovivi3'] == 1, 'v2a1'], label='Monthly rent payment of household(rented=1)')\n",
    "sns.kdeplot(df_train.loc[df_train['tipovivi3'] == 0, 'v2a1'], label='Monthly rent payment of household(rented=0)')\n",
    "plt.xscale('log')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "_uuid": "b33114c2780547e5b2a20c099e1d3aba2dcafead"
   },
   "outputs": [],
   "source": [
    "df_train['v2a1'].fillna(0, inplace=True)\n",
    "df_test['v2a1'].fillna(0, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "_uuid": "bac22efd261f35c249e261a613ac8ed840248675"
   },
   "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>Total</th>\n",
       "      <th>Percent</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>edjef</th>\n",
       "      <td>9557</td>\n",
       "      <td>100.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Target</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sanitario3</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sanitario2</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sanitario1</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>coopele</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>noelec</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>planpri</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>public</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>abastaguano</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>abastaguafuera</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>abastaguadentro</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cielorazo</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>techootro</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>techocane</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>techoentrepiso</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>techozinc</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sanitario5</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sanitario6</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>energcocinar1</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 Total  Percent\n",
       "edjef             9557    100.0\n",
       "Target               0      0.0\n",
       "sanitario3           0      0.0\n",
       "sanitario2           0      0.0\n",
       "sanitario1           0      0.0\n",
       "coopele              0      0.0\n",
       "noelec               0      0.0\n",
       "planpri              0      0.0\n",
       "public               0      0.0\n",
       "abastaguano          0      0.0\n",
       "abastaguafuera       0      0.0\n",
       "abastaguadentro      0      0.0\n",
       "cielorazo            0      0.0\n",
       "techootro            0      0.0\n",
       "techocane            0      0.0\n",
       "techoentrepiso       0      0.0\n",
       "techozinc            0      0.0\n",
       "sanitario5           0      0.0\n",
       "sanitario6           0      0.0\n",
       "energcocinar1        0      0.0"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "total = df_train.isnull().sum().sort_values(ascending=False)\n",
    "percent = 100 * (df_train.isnull().sum() / df_train.isnull().count()).sort_values(ascending=False)\n",
    "missing_df = pd.concat([total, percent], axis=1, keys=['Total', 'Percent'])\n",
    "\n",
    "missing_df.head(20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "_uuid": "bb1719d91bf48298ee014e014eb62278f5faf66d"
   },
   "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>Total</th>\n",
       "      <th>Percent</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>edjef</th>\n",
       "      <td>23856</td>\n",
       "      <td>100.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>abastaguano</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sanitario2</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sanitario1</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>coopele</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>noelec</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>planpri</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>public</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>abastaguafuera</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sanitario5</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>abastaguadentro</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cielorazo</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>techootro</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>techocane</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>techoentrepiso</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>techozinc</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sanitario3</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sanitario6</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>pisonotiene</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>elimbasu5</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 Total  Percent\n",
       "edjef            23856    100.0\n",
       "abastaguano          0      0.0\n",
       "sanitario2           0      0.0\n",
       "sanitario1           0      0.0\n",
       "coopele              0      0.0\n",
       "noelec               0      0.0\n",
       "planpri              0      0.0\n",
       "public               0      0.0\n",
       "abastaguafuera       0      0.0\n",
       "sanitario5           0      0.0\n",
       "abastaguadentro      0      0.0\n",
       "cielorazo            0      0.0\n",
       "techootro            0      0.0\n",
       "techocane            0      0.0\n",
       "techoentrepiso       0      0.0\n",
       "techozinc            0      0.0\n",
       "sanitario3           0      0.0\n",
       "sanitario6           0      0.0\n",
       "pisonotiene          0      0.0\n",
       "elimbasu5            0      0.0"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "total = df_test.isnull().sum().sort_values(ascending=False)\n",
    "percent = 100 * (df_test.isnull().sum() / df_test.isnull().count()).sort_values(ascending=False)\n",
    "missing_df = pd.concat([total, percent], axis=1, keys=['Total', 'Percent'])\n",
    "\n",
    "missing_df.head(20)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "2a3a99f2719bbf382bb672b43b0c92f588ac4861"
   },
   "source": [
    "### For now, there are no NaN values."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "34047b73af01f9f447ae8949c53f5d002fca68eb"
   },
   "source": [
    "# 2. Feature engineering"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "1c99f77f2e7bac40b2a4e2304c63fbcce3093576"
   },
   "source": [
    "## 2.1 Object features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "_uuid": "51f880f4f9046dc528a207935bc35a325162917a"
   },
   "outputs": [],
   "source": [
    "features_object = [col for col in df_train.columns if df_train[col].dtype == 'object']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "_uuid": "1ba93d17efb955c9076160b5c8e526cd8ee81942"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['Id', 'idhogar', 'dependency', 'edjefe', 'edjefa']"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features_object"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "b2e20d63e4d89d273e2cbe987e9aed52ade4b2b1"
   },
   "source": [
    "### dependecy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "_uuid": "8a067efcf3850a0e2673f0346430a1fda9fff92b",
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# some dependencies are Na, fill those with the square root of the square\n",
    "df_train['dependency'] = np.sqrt(df_train['SQBdependency'])\n",
    "df_test['dependency'] = np.sqrt(df_test['SQBdependency'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "_uuid": "15db8892dbac81bda52cba8b65a4b0eb6413cdb1"
   },
   "outputs": [],
   "source": [
    "# df_train['dependency'] = df_train['dependency'].replace({np.inf: 0})\n",
    "# df_test['dependency'] = df_test['dependency'].replace({np.inf: 0})\n",
    "\n",
    "# def replace_dependency(x):\n",
    "#     if x == 'yes':\n",
    "#         return 10\n",
    "#     elif x == 'no':\n",
    "#         return 0\n",
    "#     else:\n",
    "#         return x\n",
    "\n",
    "# df_train['dependency'] = df_train['dependency'].apply(replace_dependency).astype(float)\n",
    "# df_test['dependency'] = df_test['dependency'].apply(replace_dependency).astype(float)\n",
    "\n",
    "# - As you can see, setting yes -> 10 and no -> 0 is good choice.\n",
    "# - At first, fill inf value with 0."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "c300c4453bbec936b18ec3e86e6524cf520184d1"
   },
   "source": [
    "### edjefe\n",
    "- edjefe, years of education of male head of household, based on the interaction of escolari (years of education), head of household and gender, yes=1 and no=0\n",
    "- replace yes -> 1 and no -> 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "_uuid": "446ce42d405cb1869062383117ebfa463749969c"
   },
   "outputs": [],
   "source": [
    "def replace_edjefe(x):\n",
    "    if x == 'yes':\n",
    "        return 1\n",
    "    elif x == 'no':\n",
    "        return 0\n",
    "    else:\n",
    "        return x\n",
    "\n",
    "df_train['edjefe'] = df_train['edjefe'].apply(replace_edjefe).astype(float)\n",
    "df_test['edjefe'] = df_test['edjefe'].apply(replace_edjefe).astype(float)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "9470bff2cd416af6190e57632126c7c12e376050"
   },
   "source": [
    "### edjefa"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "450cc5abdfe849b70c3500628efd67f04fece758"
   },
   "source": [
    "- edjefa, years of education of female head of household, based on the interaction of escolari (years of education), head of household and gender, yes=1 and no=0\n",
    "- replace yes -> 1 and no -> 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "_uuid": "f7b42773d15495ffb1f3dd035b00074439cb8f66"
   },
   "outputs": [],
   "source": [
    "def replace_edjefa(x):\n",
    "    if x == 'yes':\n",
    "        return 1\n",
    "    elif x == 'no':\n",
    "        return 0\n",
    "    else:\n",
    "        return x\n",
    "\n",
    "df_train['edjefa'] = df_train['edjefa'].apply(replace_edjefa).astype(float)\n",
    "df_test['edjefa'] = df_test['edjefa'].apply(replace_edjefa).astype(float)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "_uuid": "8861fe2bde9e657f4a6989a6f3c453eed6ed0a0b"
   },
   "outputs": [],
   "source": [
    "# create feature with max education of either head of household\n",
    "df_train['edjef'] = np.max(df_train[['edjefa','edjefe']], axis=1)\n",
    "df_test['edjef'] = np.max(df_test[['edjefa','edjefe']], axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "5c550735bdfb9f17560b1b33421b40d4990d6905"
   },
   "source": [
    "## roof and electricity\n",
    "- I refered to https://www.kaggle.com/mineshjethva/exploratory-data-analysis-lightgbm. Thanks!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "_uuid": "c48f57d758c07fd47ec4c530e778eeada8976a7d"
   },
   "outputs": [],
   "source": [
    "df_train['roof_waste_material'] = np.nan\n",
    "df_test['roof_waste_material'] = np.nan\n",
    "df_train['electricity_other'] = np.nan\n",
    "df_test['electricity_other'] = np.nan\n",
    "\n",
    "def fill_roof_exception(x):\n",
    "    if (x['techozinc'] == 0) and (x['techoentrepiso'] == 0) and (x['techocane'] == 0) and (x['techootro'] == 0):\n",
    "        return 1\n",
    "    else:\n",
    "        return 0\n",
    "    \n",
    "def fill_no_electricity(x):\n",
    "    if (x['public'] == 0) and (x['planpri'] == 0) and (x['noelec'] == 0) and (x['coopele'] == 0):\n",
    "        return 1\n",
    "    else:\n",
    "        return 0\n",
    "\n",
    "df_train['roof_waste_material'] = df_train.apply(lambda x : fill_roof_exception(x),axis=1)\n",
    "df_test['roof_waste_material'] = df_test.apply(lambda x : fill_roof_exception(x),axis=1)\n",
    "df_train['electricity_other'] = df_train.apply(lambda x : fill_no_electricity(x),axis=1)\n",
    "df_test['electricity_other'] = df_test.apply(lambda x : fill_no_electricity(x),axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "9c913c59f17a28c5414c1e6d18ec8c98f5b09a13"
   },
   "source": [
    "## 2.2 Extract cat features\n",
    "- According to data scription, there are many binary category features."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "_uuid": "a20579541c69fdbef62329df3ffd115bd1fb2c1f"
   },
   "outputs": [],
   "source": [
    "binary_cat_features = [col for col in df_train.columns if df_train[col].value_counts().shape[0] == 2]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "62d933626a1eb27eee31722a174ceeb1f5260506"
   },
   "source": [
    "## 2.3 Make new features using continuous feature"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "_uuid": "0a06da54bb9640b8c2d0b20c182371a8dc4d7bb3"
   },
   "outputs": [],
   "source": [
    "continuous_features = [col for col in df_train.columns if col not in binary_cat_features]\n",
    "continuous_features = [col for col in continuous_features if col not in features_object]\n",
    "continuous_features = [col for col in continuous_features if col not in ['Id', 'Target', 'idhogar']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "_uuid": "2361e957db8bd5697aa3f14087a17bf48bfcdc6c",
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "There are 37 continuous features\n",
      "v2a1: [' Monthly rent payment']\n",
      "rooms: ['  number of all rooms in the house']\n",
      "v18q1: [' number of tablets household owns']\n",
      "r4h1: [' Males younger than 12 years of age']\n",
      "r4h2: [' Males 12 years of age and older']\n",
      "r4h3: [' Total males in the household']\n",
      "r4m1: [' Females younger than 12 years of age']\n",
      "r4m2: [' Females 12 years of age and older']\n",
      "r4m3: [' Total females in the household']\n",
      "r4t1: [' persons younger than 12 years of age']\n",
      "r4t2: [' persons 12 years of age and older']\n",
      "r4t3: [' Total persons in the household']\n",
      "tamhog: [' size of the household']\n",
      "tamviv: [' number of persons living in the household']\n",
      "escolari: [' years of schooling']\n",
      "rez_esc: [' Years behind in school']\n",
      "hhsize: [' household size']\n",
      "elimbasu5: [' =1 if rubbish disposal mainly by throwing in river,   creek or sea']\n",
      "hogar_nin: [' Number of children 0 to 19 in household']\n",
      "hogar_adul: [' Number of adults in household']\n",
      "hogar_mayor: [' # of individuals 65+ in the household']\n",
      "hogar_total: [' # of total individuals in the household']\n",
      "meaneduc: ['average years of education for adults (18+)']\n",
      "bedrooms: [' number of bedrooms']\n",
      "overcrowding: [' # persons per room']\n",
      "qmobilephone: [' # of mobile phones']\n",
      "age: [' Age in years']\n",
      "SQBescolari: [' escolari squared']\n",
      "SQBage: [' age squared']\n",
      "SQBhogar_total: [' hogar_total squared']\n",
      "SQBedjefe: [' edjefe squared']\n",
      "SQBhogar_nin: [' hogar_nin squared']\n",
      "SQBovercrowding: [' overcrowding squared']\n",
      "SQBdependency: [' dependency squared']\n",
      "SQBmeaned: [' meaned squared']\n",
      "agesq: [' Age squared']\n",
      "edjef: []\n"
     ]
    }
   ],
   "source": [
    "print('There are {} continuous features'.format(len(continuous_features)))\n",
    "for col in continuous_features:\n",
    "    print('{}: {}'.format(col, description.loc[description['varname'] == col, 'description'].values))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "a39897f86c6502c2467beaf824373870b94d51bf"
   },
   "source": [
    "- hhsize : household size\n",
    "- tamhog : size of the household\n",
    "\n",
    "What is different?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "0c63dadb664a2ea3c7fc3de989545100ef27570e"
   },
   "source": [
    "- As you can see, the meaning of two features are same but the exact number are different. Are they different?\n",
    "- I don't know.  For now, I decided to drop one feature 'tamhog'."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "_uuid": "3a34ab43e55dfbf649925dccb77c1528652ea744"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "6.0     2792\n",
       "11.0    1150\n",
       "9.0      723\n",
       "8.0      474\n",
       "15.0     473\n",
       "3.0      459\n",
       "0.0      435\n",
       "7.0      413\n",
       "4.0      400\n",
       "5.0      398\n",
       "14.0     328\n",
       "17.0     278\n",
       "2.0      278\n",
       "16.0     247\n",
       "10.0     207\n",
       "12.0     185\n",
       "13.0     155\n",
       "1.0       65\n",
       "21.0      48\n",
       "18.0      22\n",
       "19.0      18\n",
       "20.0       9\n",
       "Name: edjef, dtype: int64"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train['edjef'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "_uuid": "9dfb6abf3157d7ea9e50c13cb9d6f146e6bb9a1e"
   },
   "outputs": [],
   "source": [
    "df_train.drop('tamhog', axis=1, inplace=True)\n",
    "df_test.drop('tamhog', axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "7c030d812b81f5cabe9ba520fc936c7f6fa4f16b"
   },
   "source": [
    "### Family features\n",
    "- hogar_nin, hogar_adul, hogar_mayor, hogar_total, r4h1, r4h2, r4h3, r4m1, r4m2, r4m3, r4t1, r4t2, r4t3, tmbhog, tamvid, rez_esc, escolari"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "fb2fb6e49e524210213c34a48eab18ba9dac74e7"
   },
   "source": [
    "- Family size features (substract, ratio)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "_uuid": "43241b2ae7c0e3aefab06fbe4f8496ef666f365e"
   },
   "outputs": [],
   "source": [
    "df_train['adult'] = df_train['hogar_adul'] - df_train['hogar_mayor']\n",
    "df_train['dependency_count'] = df_train['hogar_nin'] + df_train['hogar_mayor']\n",
    "df_train['dependency'] = df_train['dependency_count'] / df_train['adult']\n",
    "df_train['child_percent'] = df_train['hogar_nin'] / df_train['hogar_total']\n",
    "df_train['elder_percent'] = df_train['hogar_mayor'] / df_train['hogar_total']\n",
    "df_train['adult_percent'] = df_train['hogar_adul'] / df_train['hogar_total']\n",
    "df_train['males_younger_12_years_percent'] = df_train['r4h1'] / df_train['hogar_total']\n",
    "df_train['males_older_12_years_percent'] = df_train['r4h2'] / df_train['hogar_total']\n",
    "df_train['males_percent'] = df_train['r4h3'] / df_train['hogar_total']\n",
    "df_train['females_younger_12_years_percent'] = df_train['r4m1'] / df_train['hogar_total']\n",
    "df_train['females_older_12_years_percent'] = df_train['r4m2'] / df_train['hogar_total']\n",
    "df_train['females_percent'] = df_train['r4m3'] / df_train['hogar_total']\n",
    "df_train['persons_younger_12_years_percent'] = df_train['r4t1'] / df_train['hogar_total']\n",
    "df_train['persons_older_12_years_percent'] = df_train['r4t2'] / df_train['hogar_total']\n",
    "df_train['persons_percent'] = df_train['r4t3'] / df_train['hogar_total']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "_uuid": "c4b3e88987e3810303007f158e8671b7463922be"
   },
   "outputs": [],
   "source": [
    "df_test['adult'] = df_test['hogar_adul'] - df_test['hogar_mayor']\n",
    "df_test['dependency_count'] = df_test['hogar_nin'] + df_test['hogar_mayor']\n",
    "df_test['dependency'] = df_test['dependency_count'] / df_test['adult']\n",
    "df_test['child_percent'] = df_test['hogar_nin'] / df_test['hogar_total']\n",
    "df_test['elder_percent'] = df_test['hogar_mayor'] / df_test['hogar_total']\n",
    "df_test['adult_percent'] = df_test['hogar_adul'] / df_test['hogar_total']\n",
    "df_test['males_younger_12_years_percent'] = df_test['r4h1'] / df_test['hogar_total']\n",
    "df_test['males_older_12_years_percent'] = df_test['r4h2'] / df_test['hogar_total']\n",
    "df_test['males_percent'] = df_test['r4h3'] / df_test['hogar_total']\n",
    "df_test['females_younger_12_years_percent'] = df_test['r4m1'] / df_test['hogar_total']\n",
    "df_test['females_older_12_years_percent'] = df_test['r4m2'] / df_test['hogar_total']\n",
    "df_test['females_percent'] = df_test['r4m3'] / df_test['hogar_total']\n",
    "df_test['persons_younger_12_years_percent'] = df_test['r4t1'] / df_test['hogar_total']\n",
    "df_test['persons_older_12_years_percent'] = df_test['r4t2'] / df_test['hogar_total']\n",
    "df_test['persons_percent'] = df_test['r4t3'] / df_test['hogar_total']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "_uuid": "bc1b3be9be5f6193bd716d109f88236fc6b2e824"
   },
   "outputs": [],
   "source": [
    "df_train['males_younger_12_years_in_household_size'] = df_train['r4h1'] / df_train['hhsize']\n",
    "df_train['males_older_12_years_in_household_size'] = df_train['r4h2'] / df_train['hhsize']\n",
    "df_train['males_in_household_size'] = df_train['r4h3'] / df_train['hhsize']\n",
    "df_train['females_younger_12_years_in_household_size'] = df_train['r4m1'] / df_train['hhsize']\n",
    "df_train['females_older_12_years_in_household_size'] = df_train['r4m2'] / df_train['hhsize']\n",
    "df_train['females_in_household_size'] = df_train['r4m3'] / df_train['hogar_total']\n",
    "df_train['persons_younger_12_years_in_household_size'] = df_train['r4t1'] / df_train['hhsize']\n",
    "df_train['persons_older_12_years_in_household_size'] = df_train['r4t2'] / df_train['hhsize']\n",
    "df_train['persons_in_household_size'] = df_train['r4t3'] / df_train['hhsize']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "_uuid": "ad7e56aa99bbc7c6aeaae9d203e41f05a044999b"
   },
   "outputs": [],
   "source": [
    "df_test['males_younger_12_years_in_household_size'] = df_test['r4h1'] / df_test['hhsize']\n",
    "df_test['males_older_12_years_in_household_size'] = df_test['r4h2'] / df_test['hhsize']\n",
    "df_test['males_in_household_size'] = df_test['r4h3'] / df_test['hhsize']\n",
    "df_test['females_younger_12_years_in_household_size'] = df_test['r4m1'] / df_test['hhsize']\n",
    "df_test['females_older_12_years_in_household_size'] = df_test['r4m2'] / df_test['hhsize']\n",
    "df_test['females_in_household_size'] = df_test['r4m3'] / df_test['hogar_total']\n",
    "df_test['persons_younger_12_years_in_household_size'] = df_test['r4t1'] / df_test['hhsize']\n",
    "df_test['persons_older_12_years_in_household_size'] = df_test['r4t2'] / df_test['hhsize']\n",
    "df_test['persons_in_household_size'] = df_test['r4t3'] / df_test['hhsize']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "_uuid": "fa2654cecf992480b6fc09e94102b407c561a93d"
   },
   "outputs": [],
   "source": [
    "df_train['overcrowding_room_and_bedroom'] = (df_train['hacdor'] + df_train['hacapo'])/2\n",
    "df_test['overcrowding_room_and_bedroom'] = (df_test['hacdor'] + df_test['hacapo'])/2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "_uuid": "388604d788a438b1125a6581dcc128fe89d1a2e9"
   },
   "outputs": [],
   "source": [
    "df_train['escolari_age'] = df_train['escolari']/df_train['age']\n",
    "df_test['escolari_age'] = df_test['escolari']/df_test['age']\n",
    "\n",
    "df_train['age_12_19'] = df_train['hogar_nin'] - df_train['r4t1']\n",
    "df_test['age_12_19'] = df_test['hogar_nin'] - df_test['r4t1']  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "_uuid": "692040a446fa5c6967c2b0ca508c8f23b567970a"
   },
   "outputs": [],
   "source": [
    "df_train['phones-per-capita'] = df_train['qmobilephone'] / df_train['tamviv']\n",
    "df_train['tablets-per-capita'] = df_train['v18q1'] / df_train['tamviv']\n",
    "df_train['rooms-per-capita'] = df_train['rooms'] / df_train['tamviv']\n",
    "df_train['rent-per-capita'] = df_train['v2a1'] / df_train['tamviv']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "_uuid": "04f8fedf12565739f1aad4eac1f323f430e792df"
   },
   "outputs": [],
   "source": [
    "df_test['phones-per-capita'] = df_test['qmobilephone'] / df_test['tamviv']\n",
    "df_test['tablets-per-capita'] = df_test['v18q1'] / df_test['tamviv']\n",
    "df_test['rooms-per-capita'] = df_test['rooms'] / df_test['tamviv']\n",
    "df_test['rent-per-capita'] = df_test['v2a1'] / df_test['tamviv']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "_uuid": "7c7d8a46b40b1647eb0abbe3a008778b16a480a0"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "9509"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(df_train['hogar_total'] == df_train['r4t3']).sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "88dcef80788a229d81ef03a71f6f8801e35e0b87"
   },
   "source": [
    "### Rent per  family features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "_uuid": "47b46ffaaaca67bdca60361a5c5c56966793bda4"
   },
   "outputs": [],
   "source": [
    "family_size_features = ['adult', 'hogar_adul', 'hogar_mayor', 'hogar_nin', 'hogar_total', 'r4h1', \n",
    "                        'r4h2', 'r4h3', 'r4m1', 'r4m2', 'r4m3', 'r4t1', 'r4t2', 'r4t3', 'hhsize']\n",
    "new_feats = []\n",
    "for col in family_size_features:\n",
    "    new_col_name = 'new_{}_per_{}'.format('v2a1', col)\n",
    "    new_feats.append(new_col_name)\n",
    "    df_train[new_col_name] = df_train['v2a1'] / df_train[col]\n",
    "    df_test[new_col_name] = df_test['v2a1'] / df_test[col]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "_uuid": "3a88293e44ccd48ea2b748a459f0176692b90678"
   },
   "outputs": [],
   "source": [
    "for col in new_feats:\n",
    "    df_train[col].replace([np.inf], np.nan, inplace=True)\n",
    "    df_train[col].fillna(0, inplace=True)\n",
    "    \n",
    "    df_test[col].replace([np.inf], np.nan, inplace=True)\n",
    "    df_test[col].fillna(0, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "b8a13813518255466844c9eec8f886ff952d2904"
   },
   "source": [
    "### Room per family features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "_uuid": "5d8cf393ff925a773c6bcfc2e580571402fe40d6"
   },
   "outputs": [],
   "source": [
    "new_feats = []\n",
    "for col in family_size_features:\n",
    "    new_col_name = 'new_{}_per_{}'.format('rooms', col)\n",
    "    new_feats.append(new_col_name)\n",
    "    df_train[new_col_name] = df_train['rooms'] / df_train[col]\n",
    "    df_test[new_col_name] = df_test['rooms'] / df_test[col]\n",
    "\n",
    "for col in new_feats:\n",
    "    df_train[col].replace([np.inf], np.nan, inplace=True)\n",
    "    df_train[col].fillna(0, inplace=True)\n",
    "    \n",
    "    df_test[col].replace([np.inf], np.nan, inplace=True)\n",
    "    df_test[col].fillna(0, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "8996ecc0399297daf4deec25e47bdbd8160ac7fe"
   },
   "source": [
    "### BedRoom per family features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "_uuid": "d9f49890f68d69f15ae8d7b415de5467c4e56e4a"
   },
   "outputs": [],
   "source": [
    "new_feats = []\n",
    "for col in family_size_features:\n",
    "    new_col_name = 'new_{}_per_{}'.format('bedrooms', col)\n",
    "    new_feats.append(new_col_name)\n",
    "    df_train[new_col_name] = df_train['bedrooms'] / df_train[col]\n",
    "    df_test[new_col_name] = df_test['bedrooms'] / df_test[col]\n",
    "\n",
    "for col in new_feats:\n",
    "    df_train[col].replace([np.inf], np.nan, inplace=True)\n",
    "    df_train[col].fillna(0, inplace=True)\n",
    "    \n",
    "    df_test[col].replace([np.inf], np.nan, inplace=True)\n",
    "    df_test[col].fillna(0, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "_uuid": "9aeb114529f877b728484dfdc0b60c6d69209127"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(9557, 220) (23856, 219)\n"
     ]
    }
   ],
   "source": [
    "print(df_train.shape, df_test.shape) # To check the same number of features between train and test (target is there in train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "aa67229c4b620b697e552dd83bc6b29451e9275a"
   },
   "source": [
    "### Tabulet per family features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "_uuid": "42fbeb4d81b8f63b94fcf27f3284ae0ea0a6eeb0"
   },
   "outputs": [],
   "source": [
    "new_feats = []\n",
    "for col in family_size_features:\n",
    "    new_col_name = 'new_{}_per_{}'.format('v18q1', col)\n",
    "    new_feats.append(new_col_name)\n",
    "    df_train[new_col_name] = df_train['v18q1'] / df_train[col]\n",
    "    df_test[new_col_name] = df_test['v18q1'] / df_test[col]\n",
    "\n",
    "for col in new_feats:\n",
    "    df_train[col].replace([np.inf], np.nan, inplace=True)\n",
    "    df_train[col].fillna(0, inplace=True)\n",
    "    \n",
    "    df_test[col].replace([np.inf], np.nan, inplace=True)\n",
    "    df_test[col].fillna(0, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "8a296e1a764b682daf24775835e6abede1ffaf1f"
   },
   "source": [
    "### phone per family features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "_uuid": "99592bf467959b6219965ae3705f473f952c7c8a"
   },
   "outputs": [],
   "source": [
    "new_feats = []\n",
    "for col in family_size_features:\n",
    "    new_col_name = 'new_{}_per_{}'.format('qmobilephone', col)\n",
    "    new_feats.append(new_col_name)\n",
    "    df_train[new_col_name] = df_train['qmobilephone'] / df_train[col]\n",
    "    df_test[new_col_name] = df_test['qmobilephone'] / df_test[col]\n",
    "\n",
    "for col in new_feats:\n",
    "    df_train[col].replace([np.inf], np.nan, inplace=True)\n",
    "    df_train[col].fillna(0, inplace=True)\n",
    "    \n",
    "    df_test[col].replace([np.inf], np.nan, inplace=True)\n",
    "    df_test[col].fillna(0, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "8ed9c9918d19dd3f7b797be2c5d76713903f62e2"
   },
   "source": [
    "### rez_esc(Years behind in school)  per family features "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "_uuid": "e6fa57417a86cbd1c4c5e9c5eb402fab4fef68e7"
   },
   "outputs": [],
   "source": [
    "new_feats = []\n",
    "for col in family_size_features:\n",
    "    new_col_name = 'new_{}_per_{}'.format('rez_esc', col)\n",
    "    new_feats.append(new_col_name)\n",
    "    df_train[new_col_name] = df_train['rez_esc'] / df_train[col]\n",
    "    df_test[new_col_name] = df_test['rez_esc'] / df_test[col]\n",
    "\n",
    "for col in new_feats:\n",
    "    df_train[col].replace([np.inf], np.nan, inplace=True)\n",
    "    df_train[col].fillna(0, inplace=True)\n",
    "    \n",
    "    df_test[col].replace([np.inf], np.nan, inplace=True)\n",
    "    df_test[col].fillna(0, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "_uuid": "5b9a3329657fe1a0733fed743d1e959e7fb7971e"
   },
   "outputs": [],
   "source": [
    "df_train['rez_esc_age'] = df_train['rez_esc'] / df_train['age']\n",
    "df_train['rez_esc_escolari'] = df_train['rez_esc'] / df_train['escolari']\n",
    "\n",
    "df_test['rez_esc_age'] = df_test['rez_esc'] / df_test['age']\n",
    "df_test['rez_esc_escolari'] = df_test['rez_esc'] / df_test['escolari']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "f21ffecaa311df3251930151f3e60950c57139a0"
   },
   "source": [
    "### Rich features\n",
    "- I think the more richer, the larger number of phones and tabulet"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "_uuid": "40d80e5f548c2861b28f385a7af41eb37d5124b5"
   },
   "outputs": [],
   "source": [
    "df_train['tabulet_x_qmobilephone'] = df_train['v18q1'] * df_train['qmobilephone']\n",
    "df_test['tabulet_x_qmobilephone'] = df_test['v18q1'] * df_test['qmobilephone']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "497b11ff192c7230568d79a32818922c9de5bf6c"
   },
   "source": [
    "- wall, roof, floor may be key factor.\n",
    "- Let's multiply each of them. Becuase they are binary cat features, so mulitification of each features generates new categorical features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "_uuid": "4b1da678fe03bcc328aae5806f0472fc46c88bd8"
   },
   "outputs": [],
   "source": [
    "# wall and roof\n",
    "for col1 in ['epared1', 'epared2', 'epared3']:\n",
    "    for col2 in ['etecho1', 'etecho2', 'etecho3']:\n",
    "        new_col_name = 'new_{}_x_{}'.format(col1, col2)\n",
    "        df_train[new_col_name] = df_train[col1] * df_train[col2]\n",
    "        df_test[new_col_name] = df_test[col1] * df_test[col2]\n",
    "        \n",
    "# wall and floor\n",
    "for col1 in ['epared1', 'epared2', 'epared3']:\n",
    "    for col2 in ['eviv1', 'eviv2', 'eviv3']:\n",
    "        new_col_name = 'new_{}_x_{}'.format(col1, col2)\n",
    "        df_train[new_col_name] = df_train[col1] * df_train[col2]\n",
    "        df_test[new_col_name] = df_test[col1] * df_test[col2]\n",
    "\n",
    "# roof and floor\n",
    "for col1 in ['etecho1', 'etecho2', 'etecho3']:\n",
    "    for col2 in ['eviv1', 'eviv2', 'eviv3']:\n",
    "        new_col_name = 'new_{}_x_{}'.format(col1, col2)\n",
    "        df_train[new_col_name] = df_train[col1] * df_train[col2]\n",
    "        df_test[new_col_name] = df_test[col1] * df_test[col2]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "e9085c19e1bc0f3909db69a2de83727fb5a9e1eb"
   },
   "source": [
    "- combination using three features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "_uuid": "10acc48210fa36d6cf294732222d5650f0fb9dd8"
   },
   "outputs": [],
   "source": [
    "for col1 in ['epared1', 'epared2', 'epared3']:\n",
    "    for col2 in ['etecho1', 'etecho2', 'etecho3']:\n",
    "        for col3 in ['eviv1', 'eviv2', 'eviv3']:\n",
    "            new_col_name = 'new_{}_x_{}_x_{}'.format(col1, col2, col3)\n",
    "            df_train[new_col_name] = df_train[col1] * df_train[col2] * df_train[col3]\n",
    "            df_test[new_col_name] = df_test[col1] * df_test[col2] * df_train[col3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "_uuid": "29987d9c3baaa375a35d5eae4b244f4d7bd10efc"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(9557, 322) (23856, 321)\n"
     ]
    }
   ],
   "source": [
    "print(df_train.shape, df_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "92c51e8598ea7221ee7ed3b93465be613569944b"
   },
   "source": [
    "- I want to mix electricity and energy features -> energy features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "_uuid": "b8ce3b259249a4b4947fcf7d42773d3b855967f7"
   },
   "outputs": [],
   "source": [
    "for col1 in ['public', 'planpri', 'noelec', 'coopele']:\n",
    "    for col2 in ['energcocinar1', 'energcocinar2', 'energcocinar3', 'energcocinar4']:\n",
    "        new_col_name = 'new_{}_x_{}'.format(col1, col2)\n",
    "        df_train[new_col_name] = df_train[col1] * df_train[col2]\n",
    "        df_test[new_col_name] = df_test[col1] * df_test[col2]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "1395cc67ed470303114a43c1cc0b5907cadb8c4d"
   },
   "source": [
    "- I want to mix toilet and rubbish disposal features -> other_infra features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "_uuid": "fcede16a4d1da244cb829b67e0abc1977b5e0d4d"
   },
   "outputs": [],
   "source": [
    "for col1 in ['sanitario1', 'sanitario2', 'sanitario3', 'sanitario5', 'sanitario6']:\n",
    "    for col2 in ['elimbasu1', 'elimbasu2', 'elimbasu3', 'elimbasu4', 'elimbasu5', 'elimbasu6']:\n",
    "        new_col_name = 'new_{}_x_{}'.format(col1, col2)\n",
    "        df_train[new_col_name] = df_train[col1] * df_train[col2]\n",
    "        df_test[new_col_name] = df_test[col1] * df_test[col2]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "ee29d4326b959fafcabb82d7b8a86e4a3c819548"
   },
   "source": [
    "- I want to mix toilet and water provision features -> water features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "_uuid": "9f539e6d25f9eafb37190fed3a278f4cfd92af3e"
   },
   "outputs": [],
   "source": [
    "for col1 in ['abastaguadentro', 'abastaguafuera', 'abastaguano']:\n",
    "    for col2 in ['sanitario1', 'sanitario2', 'sanitario3', 'sanitario5', 'sanitario6']:\n",
    "        new_col_name = 'new_{}_x_{}'.format(col1, col2)\n",
    "        df_train[new_col_name] = df_train[col1] * df_train[col2]\n",
    "        df_test[new_col_name] = df_test[col1] * df_test[col2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "_uuid": "1bfe5a24ad2775593b9ac193c2e5d2df34409a4d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(9557, 383) (23856, 382)\n"
     ]
    }
   ],
   "source": [
    "print(df_train.shape, df_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "e9a662850fae45b7841e44b6975b3c567b64c6bb"
   },
   "source": [
    "- I want mix education and area features -> education_zone_features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {
    "_uuid": "a8815d3f93460c0a3fa2f21c4e3d0f3db9e8e714"
   },
   "outputs": [],
   "source": [
    "for col1 in ['area1', 'area2']:\n",
    "    for col2 in ['instlevel1', 'instlevel2', 'instlevel3', 'instlevel4', 'instlevel5', 'instlevel6', 'instlevel7', 'instlevel8', 'instlevel9']:\n",
    "        new_col_name = 'new_{}_x_{}'.format(col1, col2)\n",
    "        df_train[new_col_name] = df_train[col1] * df_train[col2]\n",
    "        df_test[new_col_name] = df_test[col1] * df_test[col2]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "11f50d3234faf68fd28501f11cb38f181ef7001b"
   },
   "source": [
    "- Mix region and education"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {
    "_uuid": "abf8ef60e2f3ad77d78329a96b68b728969c37ab"
   },
   "outputs": [],
   "source": [
    "for col1 in ['lugar1', 'lugar2', 'lugar3', 'lugar4', 'lugar5', 'lugar6']:\n",
    "    for col2 in ['instlevel1', 'instlevel2', 'instlevel3', 'instlevel4', 'instlevel5', 'instlevel6', 'instlevel7', 'instlevel8', 'instlevel9']:\n",
    "        new_col_name = 'new_{}_x_{}'.format(col1, col2)\n",
    "        df_train[new_col_name] = df_train[col1] * df_train[col2]\n",
    "        df_test[new_col_name] = df_test[col1] * df_test[col2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {
    "_uuid": "6afffae902157da07565806da902eada635e86c8"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(9557, 455) (23856, 454)\n"
     ]
    }
   ],
   "source": [
    "print(df_train.shape, df_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "d5fd48e31c759584522282cea8c7c01be8dde430"
   },
   "source": [
    "- Multiply television / mobilephone / computer / tabulet / refrigerator -> electornics features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {
    "_uuid": "2f2b4b02509471c2ee2edbffb5b0f8415e2d9f32"
   },
   "outputs": [],
   "source": [
    "df_train['electronics'] = df_train['computer'] * df_train['mobilephone'] * df_train['television'] * df_train['v18q'] * df_train['refrig']\n",
    "df_test['electronics'] = df_test['computer'] * df_test['mobilephone'] * df_test['television'] * df_test['v18q'] * df_test['refrig']\n",
    "\n",
    "df_train['no_appliances'] = df_train['refrig'] + df_train['computer'] + df_train['television'] + df_train['mobilephone']\n",
    "df_test['no_appliances'] = df_test['refrig'] + df_test['computer'] + df_test['television'] + df_test['mobilephone']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "55bf2544aa4a56d1fa0d3b8d47a7ddba73d3e9e5"
   },
   "source": [
    "- Mix wall material of roof, floor, wall"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {
    "_uuid": "aef35c2d709fa6e165f55916a5c781a18faf8751"
   },
   "outputs": [],
   "source": [
    "for col1 in ['paredblolad', 'paredzocalo', 'paredpreb', 'pareddes', 'paredmad', 'paredzinc', 'paredfibras', 'paredother']:\n",
    "    for col2 in ['pisomoscer', 'pisocemento', 'pisoother', 'pisonatur', 'pisonotiene', 'pisomadera']:\n",
    "        new_col_name = 'new_{}_x_{}'.format(col1, col2)\n",
    "        df_train[new_col_name] = df_train[col1] * df_train[col2]\n",
    "        df_test[new_col_name] = df_test[col1] * df_test[col2]\n",
    "\n",
    "for col1 in ['pisomoscer', 'pisocemento', 'pisoother', 'pisonatur', 'pisonotiene', 'pisomadera']:\n",
    "    for col1 in ['techozinc', 'techoentrepiso', 'techocane', 'techootro']:\n",
    "        new_col_name = 'new_{}_x_{}'.format(col1, col2)\n",
    "        df_train[new_col_name] = df_train[col1] * df_train[col2]\n",
    "        df_test[new_col_name] = df_test[col1] * df_test[col2]\n",
    "        \n",
    "for col1 in ['paredblolad', 'paredzocalo', 'paredpreb', 'pareddes', 'paredmad', 'paredzinc', 'paredfibras', 'paredother']:\n",
    "    for col2 in ['techozinc', 'techoentrepiso', 'techocane', 'techootro']:\n",
    "        new_col_name = 'new_{}_x_{}'.format(col1, col2)\n",
    "        df_train[new_col_name] = df_train[col1] * df_train[col2]\n",
    "        df_test[new_col_name] = df_test[col1] * df_test[col2]        \n",
    "        \n",
    "for col1 in ['paredblolad', 'paredzocalo', 'paredpreb', 'pareddes', 'paredmad', 'paredzinc', 'paredfibras', 'paredother']:\n",
    "    for col2 in ['pisomoscer', 'pisocemento', 'pisoother', 'pisonatur', 'pisonotiene', 'pisomadera']:\n",
    "        for col3 in ['techozinc', 'techoentrepiso', 'techocane', 'techootro']:\n",
    "            new_col_name = 'new_{}_x_{}_x_{}'.format(col1, col2, col3)\n",
    "            df_train[new_col_name] = df_train[col1] * df_train[col2] * df_train[col3]\n",
    "            df_test[new_col_name] = df_test[col1] * df_test[col2] * df_train[col3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {
    "_uuid": "35be9d556f0c0ffb16ecdb7c15d1181309e0c7e6"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(9557, 733) (23856, 732)\n"
     ]
    }
   ],
   "source": [
    "print(df_train.shape, df_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "a473a83424be54e6658544f5b0456618cf0e912f"
   },
   "source": [
    "### Remove feature with only one value"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {
    "_uuid": "12605c0a589e7e6c11ef46b773aae56ca4bb8b83",
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "elimbasu5\n",
      "new_planpri_x_energcocinar1\n",
      "new_planpri_x_energcocinar2\n",
      "new_planpri_x_energcocinar3\n",
      "new_planpri_x_energcocinar4\n",
      "new_noelec_x_energcocinar2\n",
      "new_sanitario1_x_elimbasu4\n",
      "new_sanitario1_x_elimbasu5\n",
      "new_sanitario1_x_elimbasu6\n",
      "new_sanitario2_x_elimbasu4\n",
      "new_sanitario2_x_elimbasu5\n",
      "new_sanitario2_x_elimbasu6\n",
      "new_sanitario3_x_elimbasu5\n",
      "new_sanitario5_x_elimbasu4\n",
      "new_sanitario5_x_elimbasu5\n",
      "new_sanitario5_x_elimbasu6\n",
      "new_sanitario6_x_elimbasu2\n",
      "new_sanitario6_x_elimbasu4\n",
      "new_sanitario6_x_elimbasu5\n",
      "new_sanitario6_x_elimbasu6\n",
      "new_abastaguafuera_x_sanitario6\n",
      "new_abastaguano_x_sanitario2\n",
      "new_abastaguano_x_sanitario6\n",
      "new_paredblolad_x_pisonatur\n",
      "new_paredblolad_x_pisonotiene\n",
      "new_paredzocalo_x_pisoother\n",
      "new_paredzocalo_x_pisonatur\n",
      "new_paredpreb_x_pisonatur\n",
      "new_pareddes_x_pisoother\n",
      "new_pareddes_x_pisonatur\n",
      "new_paredmad_x_pisoother\n",
      "new_paredmad_x_pisonatur\n",
      "new_paredzinc_x_pisoother\n",
      "new_paredzinc_x_pisonatur\n",
      "new_paredfibras_x_pisoother\n",
      "new_paredfibras_x_pisonatur\n",
      "new_paredfibras_x_pisonotiene\n",
      "new_paredfibras_x_pisomadera\n",
      "new_paredother_x_pisoother\n",
      "new_paredother_x_pisonatur\n",
      "new_paredother_x_pisonotiene\n",
      "new_paredother_x_pisomadera\n",
      "new_techocane_x_pisomadera\n",
      "new_techootro_x_pisomadera\n",
      "new_paredzocalo_x_techoentrepiso\n",
      "new_paredzocalo_x_techocane\n",
      "new_paredzocalo_x_techootro\n",
      "new_paredpreb_x_techootro\n",
      "new_pareddes_x_techoentrepiso\n",
      "new_pareddes_x_techocane\n",
      "new_pareddes_x_techootro\n",
      "new_paredmad_x_techocane\n",
      "new_paredmad_x_techootro\n",
      "new_paredzinc_x_techoentrepiso\n",
      "new_paredzinc_x_techocane\n",
      "new_paredzinc_x_techootro\n",
      "new_paredfibras_x_techoentrepiso\n",
      "new_paredfibras_x_techootro\n",
      "new_paredother_x_techoentrepiso\n",
      "new_paredother_x_techocane\n",
      "new_paredother_x_techootro\n",
      "new_paredblolad_x_pisocemento_x_techocane\n",
      "new_paredblolad_x_pisocemento_x_techootro\n",
      "new_paredblolad_x_pisoother_x_techoentrepiso\n",
      "new_paredblolad_x_pisoother_x_techocane\n",
      "new_paredblolad_x_pisoother_x_techootro\n",
      "new_paredblolad_x_pisonatur_x_techozinc\n",
      "new_paredblolad_x_pisonatur_x_techoentrepiso\n",
      "new_paredblolad_x_pisonatur_x_techocane\n",
      "new_paredblolad_x_pisonatur_x_techootro\n",
      "new_paredblolad_x_pisonotiene_x_techozinc\n",
      "new_paredblolad_x_pisonotiene_x_techoentrepiso\n",
      "new_paredblolad_x_pisonotiene_x_techocane\n",
      "new_paredblolad_x_pisonotiene_x_techootro\n",
      "new_paredblolad_x_pisomadera_x_techocane\n",
      "new_paredblolad_x_pisomadera_x_techootro\n",
      "new_paredzocalo_x_pisomoscer_x_techoentrepiso\n",
      "new_paredzocalo_x_pisomoscer_x_techocane\n",
      "new_paredzocalo_x_pisomoscer_x_techootro\n",
      "new_paredzocalo_x_pisocemento_x_techoentrepiso\n",
      "new_paredzocalo_x_pisocemento_x_techocane\n",
      "new_paredzocalo_x_pisocemento_x_techootro\n",
      "new_paredzocalo_x_pisoother_x_techozinc\n",
      "new_paredzocalo_x_pisoother_x_techoentrepiso\n",
      "new_paredzocalo_x_pisoother_x_techocane\n",
      "new_paredzocalo_x_pisoother_x_techootro\n",
      "new_paredzocalo_x_pisonatur_x_techozinc\n",
      "new_paredzocalo_x_pisonatur_x_techoentrepiso\n",
      "new_paredzocalo_x_pisonatur_x_techocane\n",
      "new_paredzocalo_x_pisonatur_x_techootro\n",
      "new_paredzocalo_x_pisonotiene_x_techoentrepiso\n",
      "new_paredzocalo_x_pisonotiene_x_techocane\n",
      "new_paredzocalo_x_pisonotiene_x_techootro\n",
      "new_paredzocalo_x_pisomadera_x_techoentrepiso\n",
      "new_paredzocalo_x_pisomadera_x_techocane\n",
      "new_paredzocalo_x_pisomadera_x_techootro\n",
      "new_paredpreb_x_pisocemento_x_techoentrepiso\n",
      "new_paredpreb_x_pisocemento_x_techocane\n",
      "new_paredpreb_x_pisocemento_x_techootro\n",
      "new_paredpreb_x_pisoother_x_techoentrepiso\n",
      "new_paredpreb_x_pisoother_x_techocane\n",
      "new_paredpreb_x_pisoother_x_techootro\n",
      "new_paredpreb_x_pisonatur_x_techozinc\n",
      "new_paredpreb_x_pisonatur_x_techoentrepiso\n",
      "new_paredpreb_x_pisonatur_x_techocane\n",
      "new_paredpreb_x_pisonatur_x_techootro\n",
      "new_paredpreb_x_pisonotiene_x_techozinc\n",
      "new_paredpreb_x_pisonotiene_x_techoentrepiso\n",
      "new_paredpreb_x_pisonotiene_x_techocane\n",
      "new_paredpreb_x_pisonotiene_x_techootro\n",
      "new_paredpreb_x_pisomadera_x_techoentrepiso\n",
      "new_paredpreb_x_pisomadera_x_techocane\n",
      "new_paredpreb_x_pisomadera_x_techootro\n",
      "new_pareddes_x_pisomoscer_x_techozinc\n",
      "new_pareddes_x_pisomoscer_x_techoentrepiso\n",
      "new_pareddes_x_pisomoscer_x_techocane\n",
      "new_pareddes_x_pisomoscer_x_techootro\n",
      "new_pareddes_x_pisocemento_x_techoentrepiso\n",
      "new_pareddes_x_pisocemento_x_techocane\n",
      "new_pareddes_x_pisocemento_x_techootro\n",
      "new_pareddes_x_pisoother_x_techozinc\n",
      "new_pareddes_x_pisoother_x_techoentrepiso\n",
      "new_pareddes_x_pisoother_x_techocane\n",
      "new_pareddes_x_pisoother_x_techootro\n",
      "new_pareddes_x_pisonatur_x_techozinc\n",
      "new_pareddes_x_pisonatur_x_techoentrepiso\n",
      "new_pareddes_x_pisonatur_x_techocane\n",
      "new_pareddes_x_pisonatur_x_techootro\n",
      "new_pareddes_x_pisonotiene_x_techoentrepiso\n",
      "new_pareddes_x_pisonotiene_x_techocane\n",
      "new_pareddes_x_pisonotiene_x_techootro\n",
      "new_pareddes_x_pisomadera_x_techozinc\n",
      "new_pareddes_x_pisomadera_x_techoentrepiso\n",
      "new_pareddes_x_pisomadera_x_techocane\n",
      "new_pareddes_x_pisomadera_x_techootro\n",
      "new_paredmad_x_pisomoscer_x_techocane\n",
      "new_paredmad_x_pisomoscer_x_techootro\n",
      "new_paredmad_x_pisocemento_x_techoentrepiso\n",
      "new_paredmad_x_pisocemento_x_techocane\n",
      "new_paredmad_x_pisocemento_x_techootro\n",
      "new_paredmad_x_pisoother_x_techozinc\n",
      "new_paredmad_x_pisoother_x_techoentrepiso\n",
      "new_paredmad_x_pisoother_x_techocane\n",
      "new_paredmad_x_pisoother_x_techootro\n",
      "new_paredmad_x_pisonatur_x_techozinc\n",
      "new_paredmad_x_pisonatur_x_techoentrepiso\n",
      "new_paredmad_x_pisonatur_x_techocane\n",
      "new_paredmad_x_pisonatur_x_techootro\n",
      "new_paredmad_x_pisonotiene_x_techoentrepiso\n",
      "new_paredmad_x_pisonotiene_x_techocane\n",
      "new_paredmad_x_pisonotiene_x_techootro\n",
      "new_paredmad_x_pisomadera_x_techocane\n",
      "new_paredzinc_x_pisomoscer_x_techoentrepiso\n",
      "new_paredzinc_x_pisomoscer_x_techocane\n",
      "new_paredzinc_x_pisomoscer_x_techootro\n",
      "new_paredzinc_x_pisocemento_x_techoentrepiso\n",
      "new_paredzinc_x_pisocemento_x_techocane\n",
      "new_paredzinc_x_pisocemento_x_techootro\n",
      "new_paredzinc_x_pisoother_x_techozinc\n",
      "new_paredzinc_x_pisoother_x_techoentrepiso\n",
      "new_paredzinc_x_pisoother_x_techocane\n",
      "new_paredzinc_x_pisoother_x_techootro\n",
      "new_paredzinc_x_pisonatur_x_techozinc\n",
      "new_paredzinc_x_pisonatur_x_techoentrepiso\n",
      "new_paredzinc_x_pisonatur_x_techocane\n",
      "new_paredzinc_x_pisonatur_x_techootro\n",
      "new_paredzinc_x_pisonotiene_x_techoentrepiso\n",
      "new_paredzinc_x_pisonotiene_x_techocane\n",
      "new_paredzinc_x_pisonotiene_x_techootro\n",
      "new_paredzinc_x_pisomadera_x_techoentrepiso\n",
      "new_paredzinc_x_pisomadera_x_techocane\n",
      "new_paredzinc_x_pisomadera_x_techootro\n",
      "new_paredfibras_x_pisomoscer_x_techoentrepiso\n",
      "new_paredfibras_x_pisomoscer_x_techocane\n",
      "new_paredfibras_x_pisomoscer_x_techootro\n",
      "new_paredfibras_x_pisocemento_x_techoentrepiso\n",
      "new_paredfibras_x_pisocemento_x_techocane\n",
      "new_paredfibras_x_pisocemento_x_techootro\n",
      "new_paredfibras_x_pisoother_x_techozinc\n",
      "new_paredfibras_x_pisoother_x_techoentrepiso\n",
      "new_paredfibras_x_pisoother_x_techocane\n",
      "new_paredfibras_x_pisoother_x_techootro\n",
      "new_paredfibras_x_pisonatur_x_techozinc\n",
      "new_paredfibras_x_pisonatur_x_techoentrepiso\n",
      "new_paredfibras_x_pisonatur_x_techocane\n",
      "new_paredfibras_x_pisonatur_x_techootro\n",
      "new_paredfibras_x_pisonotiene_x_techozinc\n",
      "new_paredfibras_x_pisonotiene_x_techoentrepiso\n",
      "new_paredfibras_x_pisonotiene_x_techocane\n",
      "new_paredfibras_x_pisonotiene_x_techootro\n",
      "new_paredfibras_x_pisomadera_x_techozinc\n",
      "new_paredfibras_x_pisomadera_x_techoentrepiso\n",
      "new_paredfibras_x_pisomadera_x_techocane\n",
      "new_paredfibras_x_pisomadera_x_techootro\n",
      "new_paredother_x_pisomoscer_x_techozinc\n",
      "new_paredother_x_pisomoscer_x_techoentrepiso\n",
      "new_paredother_x_pisomoscer_x_techocane\n",
      "new_paredother_x_pisomoscer_x_techootro\n",
      "new_paredother_x_pisocemento_x_techoentrepiso\n",
      "new_paredother_x_pisocemento_x_techocane\n",
      "new_paredother_x_pisocemento_x_techootro\n",
      "new_paredother_x_pisoother_x_techozinc\n",
      "new_paredother_x_pisoother_x_techoentrepiso\n",
      "new_paredother_x_pisoother_x_techocane\n",
      "new_paredother_x_pisoother_x_techootro\n",
      "new_paredother_x_pisonatur_x_techozinc\n",
      "new_paredother_x_pisonatur_x_techoentrepiso\n",
      "new_paredother_x_pisonatur_x_techocane\n",
      "new_paredother_x_pisonatur_x_techootro\n",
      "new_paredother_x_pisonotiene_x_techozinc\n",
      "new_paredother_x_pisonotiene_x_techoentrepiso\n",
      "new_paredother_x_pisonotiene_x_techocane\n",
      "new_paredother_x_pisonotiene_x_techootro\n",
      "new_paredother_x_pisomadera_x_techozinc\n",
      "new_paredother_x_pisomadera_x_techoentrepiso\n",
      "new_paredother_x_pisomadera_x_techocane\n",
      "new_paredother_x_pisomadera_x_techootro\n"
     ]
    }
   ],
   "source": [
    "cols_with_only_one_value = []\n",
    "for col in df_train.columns:\n",
    "    if col == 'Target':\n",
    "        continue\n",
    "    if df_train[col].value_counts().shape[0] == 1 or df_test[col].value_counts().shape[0] == 1:\n",
    "        print(col)\n",
    "        cols_with_only_one_value.append(col)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "23753ad58e2e62a3f871cc9889c754ec88f1b96e"
   },
   "source": [
    "- Let's remove them!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {
    "_uuid": "f3d31d3ad469fe2213a6922c5750b14e4cfe51c9"
   },
   "outputs": [],
   "source": [
    "df_train.drop(cols_with_only_one_value, axis=1, inplace=True)\n",
    "df_test.drop(cols_with_only_one_value, axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_train.to_csv(\"train_p.csv\", mode='w')\n",
    "df_test.to_csv(\"test_p.csv\", mode='w')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "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>Id</th>\n",
       "      <th>v2a1</th>\n",
       "      <th>hacdor</th>\n",
       "      <th>rooms</th>\n",
       "      <th>hacapo</th>\n",
       "      <th>v14a</th>\n",
       "      <th>refrig</th>\n",
       "      <th>v18q</th>\n",
       "      <th>v18q1</th>\n",
       "      <th>r4h1</th>\n",
       "      <th>...</th>\n",
       "      <th>new_paredmad_x_pisomadera_x_techozinc</th>\n",
       "      <th>new_paredmad_x_pisomadera_x_techoentrepiso</th>\n",
       "      <th>new_paredmad_x_pisomadera_x_techootro</th>\n",
       "      <th>new_paredzinc_x_pisomoscer_x_techozinc</th>\n",
       "      <th>new_paredzinc_x_pisocemento_x_techozinc</th>\n",
       "      <th>new_paredzinc_x_pisonotiene_x_techozinc</th>\n",
       "      <th>new_paredzinc_x_pisomadera_x_techozinc</th>\n",
       "      <th>new_paredfibras_x_pisomoscer_x_techozinc</th>\n",
       "      <th>new_paredfibras_x_pisocemento_x_techozinc</th>\n",
       "      <th>new_paredother_x_pisocemento_x_techozinc</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ID_279628684</td>\n",
       "      <td>190000.0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ID_f29eb3ddd</td>\n",
       "      <td>135000.0</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ID_68de51c94</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>8</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ID_d671db89c</td>\n",
       "      <td>180000.0</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>ID_d56d6f5f5</td>\n",
       "      <td>180000.0</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 516 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "             Id      v2a1  hacdor  rooms  hacapo  v14a  refrig  v18q  v18q1  \\\n",
       "0  ID_279628684  190000.0       0      3       0     1       1     0    0.0   \n",
       "1  ID_f29eb3ddd  135000.0       0      4       0     1       1     1    1.0   \n",
       "2  ID_68de51c94       0.0       0      8       0     1       1     0    0.0   \n",
       "3  ID_d671db89c  180000.0       0      5       0     1       1     1    1.0   \n",
       "4  ID_d56d6f5f5  180000.0       0      5       0     1       1     1    1.0   \n",
       "\n",
       "   r4h1  ...  new_paredmad_x_pisomadera_x_techozinc  \\\n",
       "0     0  ...                                      0   \n",
       "1     0  ...                                      1   \n",
       "2     0  ...                                      0   \n",
       "3     0  ...                                      0   \n",
       "4     0  ...                                      0   \n",
       "\n",
       "   new_paredmad_x_pisomadera_x_techoentrepiso  \\\n",
       "0                                           0   \n",
       "1                                           0   \n",
       "2                                           0   \n",
       "3                                           0   \n",
       "4                                           0   \n",
       "\n",
       "   new_paredmad_x_pisomadera_x_techootro  \\\n",
       "0                                      0   \n",
       "1                                      0   \n",
       "2                                      0   \n",
       "3                                      0   \n",
       "4                                      0   \n",
       "\n",
       "   new_paredzinc_x_pisomoscer_x_techozinc  \\\n",
       "0                                       0   \n",
       "1                                       0   \n",
       "2                                       0   \n",
       "3                                       0   \n",
       "4                                       0   \n",
       "\n",
       "   new_paredzinc_x_pisocemento_x_techozinc  \\\n",
       "0                                        0   \n",
       "1                                        0   \n",
       "2                                        0   \n",
       "3                                        0   \n",
       "4                                        0   \n",
       "\n",
       "   new_paredzinc_x_pisonotiene_x_techozinc  \\\n",
       "0                                        0   \n",
       "1                                        0   \n",
       "2                                        0   \n",
       "3                                        0   \n",
       "4                                        0   \n",
       "\n",
       "   new_paredzinc_x_pisomadera_x_techozinc  \\\n",
       "0                                       0   \n",
       "1                                       0   \n",
       "2                                       0   \n",
       "3                                       0   \n",
       "4                                       0   \n",
       "\n",
       "   new_paredfibras_x_pisomoscer_x_techozinc  \\\n",
       "0                                         0   \n",
       "1                                         0   \n",
       "2                                         0   \n",
       "3                                         0   \n",
       "4                                         0   \n",
       "\n",
       "   new_paredfibras_x_pisocemento_x_techozinc  \\\n",
       "0                                          0   \n",
       "1                                          0   \n",
       "2                                          0   \n",
       "3                                          0   \n",
       "4                                          0   \n",
       "\n",
       "   new_paredother_x_pisocemento_x_techozinc  \n",
       "0                                         0  \n",
       "1                                         0  \n",
       "2                                         0  \n",
       "3                                         0  \n",
       "4                                         0  \n",
       "\n",
       "[5 rows x 516 columns]"
      ]
     },
     "execution_count": 70,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['Id',\n",
       " 'v2a1',\n",
       " 'hacdor',\n",
       " 'rooms',\n",
       " 'hacapo',\n",
       " 'v14a',\n",
       " 'refrig',\n",
       " 'v18q',\n",
       " 'v18q1',\n",
       " 'r4h1',\n",
       " 'r4h2',\n",
       " 'r4h3',\n",
       " 'r4m1',\n",
       " 'r4m2',\n",
       " 'r4m3',\n",
       " 'r4t1',\n",
       " 'r4t2',\n",
       " 'r4t3',\n",
       " 'tamviv',\n",
       " 'escolari',\n",
       " 'rez_esc',\n",
       " 'hhsize',\n",
       " 'paredblolad',\n",
       " 'paredzocalo',\n",
       " 'paredpreb',\n",
       " 'pareddes',\n",
       " 'paredmad',\n",
       " 'paredzinc',\n",
       " 'paredfibras',\n",
       " 'paredother',\n",
       " 'pisomoscer',\n",
       " 'pisocemento',\n",
       " 'pisoother',\n",
       " 'pisonatur',\n",
       " 'pisonotiene',\n",
       " 'pisomadera',\n",
       " 'techozinc',\n",
       " 'techoentrepiso',\n",
       " 'techocane',\n",
       " 'techootro',\n",
       " 'cielorazo',\n",
       " 'abastaguadentro',\n",
       " 'abastaguafuera',\n",
       " 'abastaguano',\n",
       " 'public',\n",
       " 'planpri',\n",
       " 'noelec',\n",
       " 'coopele',\n",
       " 'sanitario1',\n",
       " 'sanitario2',\n",
       " 'sanitario3',\n",
       " 'sanitario5',\n",
       " 'sanitario6',\n",
       " 'energcocinar1',\n",
       " 'energcocinar2',\n",
       " 'energcocinar3',\n",
       " 'energcocinar4',\n",
       " 'elimbasu1',\n",
       " 'elimbasu2',\n",
       " 'elimbasu3',\n",
       " 'elimbasu4',\n",
       " 'elimbasu6',\n",
       " 'epared1',\n",
       " 'epared2',\n",
       " 'epared3',\n",
       " 'etecho1',\n",
       " 'etecho2',\n",
       " 'etecho3',\n",
       " 'eviv1',\n",
       " 'eviv2',\n",
       " 'eviv3',\n",
       " 'dis',\n",
       " 'male',\n",
       " 'female',\n",
       " 'estadocivil1',\n",
       " 'estadocivil2',\n",
       " 'estadocivil3',\n",
       " 'estadocivil4',\n",
       " 'estadocivil5',\n",
       " 'estadocivil6',\n",
       " 'estadocivil7',\n",
       " 'parentesco1',\n",
       " 'parentesco2',\n",
       " 'parentesco3',\n",
       " 'parentesco4',\n",
       " 'parentesco5',\n",
       " 'parentesco6',\n",
       " 'parentesco7',\n",
       " 'parentesco8',\n",
       " 'parentesco9',\n",
       " 'parentesco10',\n",
       " 'parentesco11',\n",
       " 'parentesco12',\n",
       " 'idhogar',\n",
       " 'hogar_nin',\n",
       " 'hogar_adul',\n",
       " 'hogar_mayor',\n",
       " 'hogar_total',\n",
       " 'dependency',\n",
       " 'edjefe',\n",
       " 'edjefa',\n",
       " 'meaneduc',\n",
       " 'instlevel1',\n",
       " 'instlevel2',\n",
       " 'instlevel3',\n",
       " 'instlevel4',\n",
       " 'instlevel5',\n",
       " 'instlevel6',\n",
       " 'instlevel7',\n",
       " 'instlevel8',\n",
       " 'instlevel9',\n",
       " 'bedrooms',\n",
       " 'overcrowding',\n",
       " 'tipovivi1',\n",
       " 'tipovivi2',\n",
       " 'tipovivi3',\n",
       " 'tipovivi4',\n",
       " 'tipovivi5',\n",
       " 'computer',\n",
       " 'television',\n",
       " 'mobilephone',\n",
       " 'qmobilephone',\n",
       " 'lugar1',\n",
       " 'lugar2',\n",
       " 'lugar3',\n",
       " 'lugar4',\n",
       " 'lugar5',\n",
       " 'lugar6',\n",
       " 'area1',\n",
       " 'area2',\n",
       " 'age',\n",
       " 'SQBescolari',\n",
       " 'SQBage',\n",
       " 'SQBhogar_total',\n",
       " 'SQBedjefe',\n",
       " 'SQBhogar_nin',\n",
       " 'SQBovercrowding',\n",
       " 'SQBdependency',\n",
       " 'SQBmeaned',\n",
       " 'agesq',\n",
       " 'Target',\n",
       " 'edjef',\n",
       " 'roof_waste_material',\n",
       " 'electricity_other',\n",
       " 'adult',\n",
       " 'dependency_count',\n",
       " 'child_percent',\n",
       " 'elder_percent',\n",
       " 'adult_percent',\n",
       " 'males_younger_12_years_percent',\n",
       " 'males_older_12_years_percent',\n",
       " 'males_percent',\n",
       " 'females_younger_12_years_percent',\n",
       " 'females_older_12_years_percent',\n",
       " 'females_percent',\n",
       " 'persons_younger_12_years_percent',\n",
       " 'persons_older_12_years_percent',\n",
       " 'persons_percent',\n",
       " 'males_younger_12_years_in_household_size',\n",
       " 'males_older_12_years_in_household_size',\n",
       " 'males_in_household_size',\n",
       " 'females_younger_12_years_in_household_size',\n",
       " 'females_older_12_years_in_household_size',\n",
       " 'females_in_household_size',\n",
       " 'persons_younger_12_years_in_household_size',\n",
       " 'persons_older_12_years_in_household_size',\n",
       " 'persons_in_household_size',\n",
       " 'overcrowding_room_and_bedroom',\n",
       " 'escolari_age',\n",
       " 'age_12_19',\n",
       " 'phones-per-capita',\n",
       " 'tablets-per-capita',\n",
       " 'rooms-per-capita',\n",
       " 'rent-per-capita',\n",
       " 'new_v2a1_per_adult',\n",
       " 'new_v2a1_per_hogar_adul',\n",
       " 'new_v2a1_per_hogar_mayor',\n",
       " 'new_v2a1_per_hogar_nin',\n",
       " 'new_v2a1_per_hogar_total',\n",
       " 'new_v2a1_per_r4h1',\n",
       " 'new_v2a1_per_r4h2',\n",
       " 'new_v2a1_per_r4h3',\n",
       " 'new_v2a1_per_r4m1',\n",
       " 'new_v2a1_per_r4m2',\n",
       " 'new_v2a1_per_r4m3',\n",
       " 'new_v2a1_per_r4t1',\n",
       " 'new_v2a1_per_r4t2',\n",
       " 'new_v2a1_per_r4t3',\n",
       " 'new_v2a1_per_hhsize',\n",
       " 'new_rooms_per_adult',\n",
       " 'new_rooms_per_hogar_adul',\n",
       " 'new_rooms_per_hogar_mayor',\n",
       " 'new_rooms_per_hogar_nin',\n",
       " 'new_rooms_per_hogar_total',\n",
       " 'new_rooms_per_r4h1',\n",
       " 'new_rooms_per_r4h2',\n",
       " 'new_rooms_per_r4h3',\n",
       " 'new_rooms_per_r4m1',\n",
       " 'new_rooms_per_r4m2',\n",
       " 'new_rooms_per_r4m3',\n",
       " 'new_rooms_per_r4t1',\n",
       " 'new_rooms_per_r4t2',\n",
       " 'new_rooms_per_r4t3',\n",
       " 'new_rooms_per_hhsize',\n",
       " 'new_bedrooms_per_adult',\n",
       " 'new_bedrooms_per_hogar_adul',\n",
       " 'new_bedrooms_per_hogar_mayor',\n",
       " 'new_bedrooms_per_hogar_nin',\n",
       " 'new_bedrooms_per_hogar_total',\n",
       " 'new_bedrooms_per_r4h1',\n",
       " 'new_bedrooms_per_r4h2',\n",
       " 'new_bedrooms_per_r4h3',\n",
       " 'new_bedrooms_per_r4m1',\n",
       " 'new_bedrooms_per_r4m2',\n",
       " 'new_bedrooms_per_r4m3',\n",
       " 'new_bedrooms_per_r4t1',\n",
       " 'new_bedrooms_per_r4t2',\n",
       " 'new_bedrooms_per_r4t3',\n",
       " 'new_bedrooms_per_hhsize',\n",
       " 'new_v18q1_per_adult',\n",
       " 'new_v18q1_per_hogar_adul',\n",
       " 'new_v18q1_per_hogar_mayor',\n",
       " 'new_v18q1_per_hogar_nin',\n",
       " 'new_v18q1_per_hogar_total',\n",
       " 'new_v18q1_per_r4h1',\n",
       " 'new_v18q1_per_r4h2',\n",
       " 'new_v18q1_per_r4h3',\n",
       " 'new_v18q1_per_r4m1',\n",
       " 'new_v18q1_per_r4m2',\n",
       " 'new_v18q1_per_r4m3',\n",
       " 'new_v18q1_per_r4t1',\n",
       " 'new_v18q1_per_r4t2',\n",
       " 'new_v18q1_per_r4t3',\n",
       " 'new_v18q1_per_hhsize',\n",
       " 'new_qmobilephone_per_adult',\n",
       " 'new_qmobilephone_per_hogar_adul',\n",
       " 'new_qmobilephone_per_hogar_mayor',\n",
       " 'new_qmobilephone_per_hogar_nin',\n",
       " 'new_qmobilephone_per_hogar_total',\n",
       " 'new_qmobilephone_per_r4h1',\n",
       " 'new_qmobilephone_per_r4h2',\n",
       " 'new_qmobilephone_per_r4h3',\n",
       " 'new_qmobilephone_per_r4m1',\n",
       " 'new_qmobilephone_per_r4m2',\n",
       " 'new_qmobilephone_per_r4m3',\n",
       " 'new_qmobilephone_per_r4t1',\n",
       " 'new_qmobilephone_per_r4t2',\n",
       " 'new_qmobilephone_per_r4t3',\n",
       " 'new_qmobilephone_per_hhsize',\n",
       " 'new_rez_esc_per_adult',\n",
       " 'new_rez_esc_per_hogar_adul',\n",
       " 'new_rez_esc_per_hogar_mayor',\n",
       " 'new_rez_esc_per_hogar_nin',\n",
       " 'new_rez_esc_per_hogar_total',\n",
       " 'new_rez_esc_per_r4h1',\n",
       " 'new_rez_esc_per_r4h2',\n",
       " 'new_rez_esc_per_r4h3',\n",
       " 'new_rez_esc_per_r4m1',\n",
       " 'new_rez_esc_per_r4m2',\n",
       " 'new_rez_esc_per_r4m3',\n",
       " 'new_rez_esc_per_r4t1',\n",
       " 'new_rez_esc_per_r4t2',\n",
       " 'new_rez_esc_per_r4t3',\n",
       " 'new_rez_esc_per_hhsize',\n",
       " 'rez_esc_age',\n",
       " 'rez_esc_escolari',\n",
       " 'tabulet_x_qmobilephone',\n",
       " 'new_epared1_x_etecho1',\n",
       " 'new_epared1_x_etecho2',\n",
       " 'new_epared1_x_etecho3',\n",
       " 'new_epared2_x_etecho1',\n",
       " 'new_epared2_x_etecho2',\n",
       " 'new_epared2_x_etecho3',\n",
       " 'new_epared3_x_etecho1',\n",
       " 'new_epared3_x_etecho2',\n",
       " 'new_epared3_x_etecho3',\n",
       " 'new_epared1_x_eviv1',\n",
       " 'new_epared1_x_eviv2',\n",
       " 'new_epared1_x_eviv3',\n",
       " 'new_epared2_x_eviv1',\n",
       " 'new_epared2_x_eviv2',\n",
       " 'new_epared2_x_eviv3',\n",
       " 'new_epared3_x_eviv1',\n",
       " 'new_epared3_x_eviv2',\n",
       " 'new_epared3_x_eviv3',\n",
       " 'new_etecho1_x_eviv1',\n",
       " 'new_etecho1_x_eviv2',\n",
       " 'new_etecho1_x_eviv3',\n",
       " 'new_etecho2_x_eviv1',\n",
       " 'new_etecho2_x_eviv2',\n",
       " 'new_etecho2_x_eviv3',\n",
       " 'new_etecho3_x_eviv1',\n",
       " 'new_etecho3_x_eviv2',\n",
       " 'new_etecho3_x_eviv3',\n",
       " 'new_epared1_x_etecho1_x_eviv1',\n",
       " 'new_epared1_x_etecho1_x_eviv2',\n",
       " 'new_epared1_x_etecho1_x_eviv3',\n",
       " 'new_epared1_x_etecho2_x_eviv1',\n",
       " 'new_epared1_x_etecho2_x_eviv2',\n",
       " 'new_epared1_x_etecho2_x_eviv3',\n",
       " 'new_epared1_x_etecho3_x_eviv1',\n",
       " 'new_epared1_x_etecho3_x_eviv2',\n",
       " 'new_epared1_x_etecho3_x_eviv3',\n",
       " 'new_epared2_x_etecho1_x_eviv1',\n",
       " 'new_epared2_x_etecho1_x_eviv2',\n",
       " 'new_epared2_x_etecho1_x_eviv3',\n",
       " 'new_epared2_x_etecho2_x_eviv1',\n",
       " 'new_epared2_x_etecho2_x_eviv2',\n",
       " 'new_epared2_x_etecho2_x_eviv3',\n",
       " 'new_epared2_x_etecho3_x_eviv1',\n",
       " 'new_epared2_x_etecho3_x_eviv2',\n",
       " 'new_epared2_x_etecho3_x_eviv3',\n",
       " 'new_epared3_x_etecho1_x_eviv1',\n",
       " 'new_epared3_x_etecho1_x_eviv2',\n",
       " 'new_epared3_x_etecho1_x_eviv3',\n",
       " 'new_epared3_x_etecho2_x_eviv1',\n",
       " 'new_epared3_x_etecho2_x_eviv2',\n",
       " 'new_epared3_x_etecho2_x_eviv3',\n",
       " 'new_epared3_x_etecho3_x_eviv1',\n",
       " 'new_epared3_x_etecho3_x_eviv2',\n",
       " 'new_epared3_x_etecho3_x_eviv3',\n",
       " 'new_public_x_energcocinar1',\n",
       " 'new_public_x_energcocinar2',\n",
       " 'new_public_x_energcocinar3',\n",
       " 'new_public_x_energcocinar4',\n",
       " 'new_noelec_x_energcocinar1',\n",
       " 'new_noelec_x_energcocinar3',\n",
       " 'new_noelec_x_energcocinar4',\n",
       " 'new_coopele_x_energcocinar1',\n",
       " 'new_coopele_x_energcocinar2',\n",
       " 'new_coopele_x_energcocinar3',\n",
       " 'new_coopele_x_energcocinar4',\n",
       " 'new_sanitario1_x_elimbasu1',\n",
       " 'new_sanitario1_x_elimbasu2',\n",
       " 'new_sanitario1_x_elimbasu3',\n",
       " 'new_sanitario2_x_elimbasu1',\n",
       " 'new_sanitario2_x_elimbasu2',\n",
       " 'new_sanitario2_x_elimbasu3',\n",
       " 'new_sanitario3_x_elimbasu1',\n",
       " 'new_sanitario3_x_elimbasu2',\n",
       " 'new_sanitario3_x_elimbasu3',\n",
       " 'new_sanitario3_x_elimbasu4',\n",
       " 'new_sanitario3_x_elimbasu6',\n",
       " 'new_sanitario5_x_elimbasu1',\n",
       " 'new_sanitario5_x_elimbasu2',\n",
       " 'new_sanitario5_x_elimbasu3',\n",
       " 'new_sanitario6_x_elimbasu1',\n",
       " 'new_sanitario6_x_elimbasu3',\n",
       " 'new_abastaguadentro_x_sanitario1',\n",
       " 'new_abastaguadentro_x_sanitario2',\n",
       " 'new_abastaguadentro_x_sanitario3',\n",
       " 'new_abastaguadentro_x_sanitario5',\n",
       " 'new_abastaguadentro_x_sanitario6',\n",
       " 'new_abastaguafuera_x_sanitario1',\n",
       " 'new_abastaguafuera_x_sanitario2',\n",
       " 'new_abastaguafuera_x_sanitario3',\n",
       " 'new_abastaguafuera_x_sanitario5',\n",
       " 'new_abastaguano_x_sanitario1',\n",
       " 'new_abastaguano_x_sanitario3',\n",
       " 'new_abastaguano_x_sanitario5',\n",
       " 'new_area1_x_instlevel1',\n",
       " 'new_area1_x_instlevel2',\n",
       " 'new_area1_x_instlevel3',\n",
       " 'new_area1_x_instlevel4',\n",
       " 'new_area1_x_instlevel5',\n",
       " 'new_area1_x_instlevel6',\n",
       " 'new_area1_x_instlevel7',\n",
       " 'new_area1_x_instlevel8',\n",
       " 'new_area1_x_instlevel9',\n",
       " 'new_area2_x_instlevel1',\n",
       " 'new_area2_x_instlevel2',\n",
       " 'new_area2_x_instlevel3',\n",
       " 'new_area2_x_instlevel4',\n",
       " 'new_area2_x_instlevel5',\n",
       " 'new_area2_x_instlevel6',\n",
       " 'new_area2_x_instlevel7',\n",
       " 'new_area2_x_instlevel8',\n",
       " 'new_area2_x_instlevel9',\n",
       " 'new_lugar1_x_instlevel1',\n",
       " 'new_lugar1_x_instlevel2',\n",
       " 'new_lugar1_x_instlevel3',\n",
       " 'new_lugar1_x_instlevel4',\n",
       " 'new_lugar1_x_instlevel5',\n",
       " 'new_lugar1_x_instlevel6',\n",
       " 'new_lugar1_x_instlevel7',\n",
       " 'new_lugar1_x_instlevel8',\n",
       " 'new_lugar1_x_instlevel9',\n",
       " 'new_lugar2_x_instlevel1',\n",
       " 'new_lugar2_x_instlevel2',\n",
       " 'new_lugar2_x_instlevel3',\n",
       " 'new_lugar2_x_instlevel4',\n",
       " 'new_lugar2_x_instlevel5',\n",
       " 'new_lugar2_x_instlevel6',\n",
       " 'new_lugar2_x_instlevel7',\n",
       " 'new_lugar2_x_instlevel8',\n",
       " 'new_lugar2_x_instlevel9',\n",
       " 'new_lugar3_x_instlevel1',\n",
       " 'new_lugar3_x_instlevel2',\n",
       " 'new_lugar3_x_instlevel3',\n",
       " 'new_lugar3_x_instlevel4',\n",
       " 'new_lugar3_x_instlevel5',\n",
       " 'new_lugar3_x_instlevel6',\n",
       " 'new_lugar3_x_instlevel7',\n",
       " 'new_lugar3_x_instlevel8',\n",
       " 'new_lugar3_x_instlevel9',\n",
       " 'new_lugar4_x_instlevel1',\n",
       " 'new_lugar4_x_instlevel2',\n",
       " 'new_lugar4_x_instlevel3',\n",
       " 'new_lugar4_x_instlevel4',\n",
       " 'new_lugar4_x_instlevel5',\n",
       " 'new_lugar4_x_instlevel6',\n",
       " 'new_lugar4_x_instlevel7',\n",
       " 'new_lugar4_x_instlevel8',\n",
       " 'new_lugar4_x_instlevel9',\n",
       " 'new_lugar5_x_instlevel1',\n",
       " 'new_lugar5_x_instlevel2',\n",
       " 'new_lugar5_x_instlevel3',\n",
       " 'new_lugar5_x_instlevel4',\n",
       " 'new_lugar5_x_instlevel5',\n",
       " 'new_lugar5_x_instlevel6',\n",
       " 'new_lugar5_x_instlevel7',\n",
       " 'new_lugar5_x_instlevel8',\n",
       " 'new_lugar5_x_instlevel9',\n",
       " 'new_lugar6_x_instlevel1',\n",
       " 'new_lugar6_x_instlevel2',\n",
       " 'new_lugar6_x_instlevel3',\n",
       " 'new_lugar6_x_instlevel4',\n",
       " 'new_lugar6_x_instlevel5',\n",
       " 'new_lugar6_x_instlevel6',\n",
       " 'new_lugar6_x_instlevel7',\n",
       " 'new_lugar6_x_instlevel8',\n",
       " 'new_lugar6_x_instlevel9',\n",
       " 'electronics',\n",
       " 'no_appliances',\n",
       " 'new_paredblolad_x_pisomoscer',\n",
       " 'new_paredblolad_x_pisocemento',\n",
       " 'new_paredblolad_x_pisoother',\n",
       " 'new_paredblolad_x_pisomadera',\n",
       " 'new_paredzocalo_x_pisomoscer',\n",
       " 'new_paredzocalo_x_pisocemento',\n",
       " 'new_paredzocalo_x_pisonotiene',\n",
       " 'new_paredzocalo_x_pisomadera',\n",
       " 'new_paredpreb_x_pisomoscer',\n",
       " 'new_paredpreb_x_pisocemento',\n",
       " 'new_paredpreb_x_pisoother',\n",
       " 'new_paredpreb_x_pisonotiene',\n",
       " 'new_paredpreb_x_pisomadera',\n",
       " 'new_pareddes_x_pisomoscer',\n",
       " 'new_pareddes_x_pisocemento',\n",
       " 'new_pareddes_x_pisonotiene',\n",
       " 'new_pareddes_x_pisomadera',\n",
       " 'new_paredmad_x_pisomoscer',\n",
       " 'new_paredmad_x_pisocemento',\n",
       " 'new_paredmad_x_pisonotiene',\n",
       " 'new_paredmad_x_pisomadera',\n",
       " 'new_paredzinc_x_pisomoscer',\n",
       " 'new_paredzinc_x_pisocemento',\n",
       " 'new_paredzinc_x_pisonotiene',\n",
       " 'new_paredzinc_x_pisomadera',\n",
       " 'new_paredfibras_x_pisomoscer',\n",
       " 'new_paredfibras_x_pisocemento',\n",
       " 'new_paredother_x_pisomoscer',\n",
       " 'new_paredother_x_pisocemento',\n",
       " 'new_techozinc_x_pisomadera',\n",
       " 'new_techoentrepiso_x_pisomadera',\n",
       " 'new_paredblolad_x_techozinc',\n",
       " 'new_paredblolad_x_techoentrepiso',\n",
       " 'new_paredblolad_x_techocane',\n",
       " 'new_paredblolad_x_techootro',\n",
       " 'new_paredzocalo_x_techozinc',\n",
       " 'new_paredpreb_x_techozinc',\n",
       " 'new_paredpreb_x_techoentrepiso',\n",
       " 'new_paredpreb_x_techocane',\n",
       " 'new_pareddes_x_techozinc',\n",
       " 'new_paredmad_x_techozinc',\n",
       " 'new_paredmad_x_techoentrepiso',\n",
       " 'new_paredzinc_x_techozinc',\n",
       " 'new_paredfibras_x_techozinc',\n",
       " 'new_paredfibras_x_techocane',\n",
       " 'new_paredother_x_techozinc',\n",
       " 'new_paredblolad_x_pisomoscer_x_techozinc',\n",
       " 'new_paredblolad_x_pisomoscer_x_techoentrepiso',\n",
       " 'new_paredblolad_x_pisomoscer_x_techocane',\n",
       " 'new_paredblolad_x_pisomoscer_x_techootro',\n",
       " 'new_paredblolad_x_pisocemento_x_techozinc',\n",
       " 'new_paredblolad_x_pisocemento_x_techoentrepiso',\n",
       " 'new_paredblolad_x_pisoother_x_techozinc',\n",
       " 'new_paredblolad_x_pisomadera_x_techozinc',\n",
       " 'new_paredblolad_x_pisomadera_x_techoentrepiso',\n",
       " 'new_paredzocalo_x_pisomoscer_x_techozinc',\n",
       " 'new_paredzocalo_x_pisocemento_x_techozinc',\n",
       " 'new_paredzocalo_x_pisonotiene_x_techozinc',\n",
       " 'new_paredzocalo_x_pisomadera_x_techozinc',\n",
       " 'new_paredpreb_x_pisomoscer_x_techozinc',\n",
       " 'new_paredpreb_x_pisomoscer_x_techoentrepiso',\n",
       " 'new_paredpreb_x_pisomoscer_x_techocane',\n",
       " 'new_paredpreb_x_pisomoscer_x_techootro',\n",
       " 'new_paredpreb_x_pisocemento_x_techozinc',\n",
       " 'new_paredpreb_x_pisoother_x_techozinc',\n",
       " 'new_paredpreb_x_pisomadera_x_techozinc',\n",
       " 'new_pareddes_x_pisocemento_x_techozinc',\n",
       " 'new_pareddes_x_pisonotiene_x_techozinc',\n",
       " 'new_paredmad_x_pisomoscer_x_techozinc',\n",
       " 'new_paredmad_x_pisomoscer_x_techoentrepiso',\n",
       " 'new_paredmad_x_pisocemento_x_techozinc',\n",
       " 'new_paredmad_x_pisonotiene_x_techozinc',\n",
       " 'new_paredmad_x_pisomadera_x_techozinc',\n",
       " 'new_paredmad_x_pisomadera_x_techoentrepiso',\n",
       " 'new_paredmad_x_pisomadera_x_techootro',\n",
       " 'new_paredzinc_x_pisomoscer_x_techozinc',\n",
       " 'new_paredzinc_x_pisocemento_x_techozinc',\n",
       " 'new_paredzinc_x_pisonotiene_x_techozinc',\n",
       " 'new_paredzinc_x_pisomadera_x_techozinc',\n",
       " 'new_paredfibras_x_pisomoscer_x_techozinc',\n",
       " 'new_paredfibras_x_pisocemento_x_techozinc',\n",
       " 'new_paredother_x_pisocemento_x_techozinc']"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.set_option('display.max_rows', 1000)\n",
    "df_train.columns.to_list()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}