2018/day34_feature_engineering.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Feature engineering with featuretools"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# pandas and numpy for data manipulation\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "# featuretools for automated feature engineering\n",
    "import featuretools as ft\n",
    "\n",
    "# ignore warnings from pandas\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Dataset\n",
    "An example dataset consisting of three tables:\n",
    "\n",
    "* `clients`: information about clients at a credit union\n",
    "* `loans`: previous loans taken out by the clients\n",
    "* `payments`: payments made/missed on the previous loans\n",
    "\n",
    "The general problem of feature engineering is taking disparate data, often distributed across multiple tables, and combining it into a single table that can be used for training a machine learning model. `featuretools` has the ability to do this for us, creating many new candidate features with minimal effort. These features are combined into a single table that can then be passed on to our model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Read in the data\n",
    "clients = pd.read_csv('data/clients.csv', parse_dates = ['joined'])\n",
    "loans = pd.read_csv('data/loans.csv', parse_dates = ['loan_start', 'loan_end'])\n",
    "payments = pd.read_csv('data/payments.csv', parse_dates = ['payment_date'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>client_id</th>\n",
       "      <th>joined</th>\n",
       "      <th>income</th>\n",
       "      <th>credit_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>46109</td>\n",
       "      <td>2002-04-16</td>\n",
       "      <td>172677</td>\n",
       "      <td>527</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>49545</td>\n",
       "      <td>2007-11-14</td>\n",
       "      <td>104564</td>\n",
       "      <td>770</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>41480</td>\n",
       "      <td>2013-03-11</td>\n",
       "      <td>122607</td>\n",
       "      <td>585</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>46180</td>\n",
       "      <td>2001-11-06</td>\n",
       "      <td>43851</td>\n",
       "      <td>562</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>25707</td>\n",
       "      <td>2006-10-06</td>\n",
       "      <td>211422</td>\n",
       "      <td>621</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   client_id     joined  income  credit_score\n",
       "0      46109 2002-04-16  172677           527\n",
       "1      49545 2007-11-14  104564           770\n",
       "2      41480 2013-03-11  122607           585\n",
       "3      46180 2001-11-06   43851           562\n",
       "4      25707 2006-10-06  211422           621"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clients.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "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>client_id</th>\n",
       "      <th>loan_type</th>\n",
       "      <th>loan_amount</th>\n",
       "      <th>repaid</th>\n",
       "      <th>loan_id</th>\n",
       "      <th>loan_start</th>\n",
       "      <th>loan_end</th>\n",
       "      <th>rate</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>288</th>\n",
       "      <td>32885</td>\n",
       "      <td>other</td>\n",
       "      <td>11142</td>\n",
       "      <td>1</td>\n",
       "      <td>11320</td>\n",
       "      <td>2009-05-29</td>\n",
       "      <td>2010-11-06</td>\n",
       "      <td>0.29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>69</th>\n",
       "      <td>46180</td>\n",
       "      <td>other</td>\n",
       "      <td>8643</td>\n",
       "      <td>0</td>\n",
       "      <td>10379</td>\n",
       "      <td>2000-11-03</td>\n",
       "      <td>2003-03-31</td>\n",
       "      <td>1.29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>346</th>\n",
       "      <td>39384</td>\n",
       "      <td>cash</td>\n",
       "      <td>11728</td>\n",
       "      <td>0</td>\n",
       "      <td>11700</td>\n",
       "      <td>2007-04-20</td>\n",
       "      <td>2009-11-09</td>\n",
       "      <td>5.78</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37</th>\n",
       "      <td>49545</td>\n",
       "      <td>home</td>\n",
       "      <td>10519</td>\n",
       "      <td>1</td>\n",
       "      <td>10425</td>\n",
       "      <td>2001-09-27</td>\n",
       "      <td>2004-03-04</td>\n",
       "      <td>1.14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>315</th>\n",
       "      <td>44387</td>\n",
       "      <td>home</td>\n",
       "      <td>14900</td>\n",
       "      <td>0</td>\n",
       "      <td>11614</td>\n",
       "      <td>2011-01-04</td>\n",
       "      <td>2013-02-05</td>\n",
       "      <td>4.75</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>335</th>\n",
       "      <td>39384</td>\n",
       "      <td>credit</td>\n",
       "      <td>2166</td>\n",
       "      <td>1</td>\n",
       "      <td>10274</td>\n",
       "      <td>2012-02-27</td>\n",
       "      <td>2013-08-30</td>\n",
       "      <td>1.56</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>321</th>\n",
       "      <td>44387</td>\n",
       "      <td>cash</td>\n",
       "      <td>2559</td>\n",
       "      <td>0</td>\n",
       "      <td>10690</td>\n",
       "      <td>2014-02-24</td>\n",
       "      <td>2016-06-11</td>\n",
       "      <td>3.27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>400</th>\n",
       "      <td>41472</td>\n",
       "      <td>home</td>\n",
       "      <td>1270</td>\n",
       "      <td>0</td>\n",
       "      <td>11855</td>\n",
       "      <td>2014-04-17</td>\n",
       "      <td>2015-10-28</td>\n",
       "      <td>9.82</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>210</th>\n",
       "      <td>26326</td>\n",
       "      <td>home</td>\n",
       "      <td>3597</td>\n",
       "      <td>1</td>\n",
       "      <td>11988</td>\n",
       "      <td>2001-04-14</td>\n",
       "      <td>2003-03-06</td>\n",
       "      <td>2.74</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>379</th>\n",
       "      <td>38537</td>\n",
       "      <td>credit</td>\n",
       "      <td>9355</td>\n",
       "      <td>1</td>\n",
       "      <td>10408</td>\n",
       "      <td>2004-04-22</td>\n",
       "      <td>2006-02-13</td>\n",
       "      <td>1.64</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     client_id loan_type  loan_amount  repaid  loan_id loan_start   loan_end  \\\n",
       "288      32885     other        11142       1    11320 2009-05-29 2010-11-06   \n",
       "69       46180     other         8643       0    10379 2000-11-03 2003-03-31   \n",
       "346      39384      cash        11728       0    11700 2007-04-20 2009-11-09   \n",
       "37       49545      home        10519       1    10425 2001-09-27 2004-03-04   \n",
       "315      44387      home        14900       0    11614 2011-01-04 2013-02-05   \n",
       "335      39384    credit         2166       1    10274 2012-02-27 2013-08-30   \n",
       "321      44387      cash         2559       0    10690 2014-02-24 2016-06-11   \n",
       "400      41472      home         1270       0    11855 2014-04-17 2015-10-28   \n",
       "210      26326      home         3597       1    11988 2001-04-14 2003-03-06   \n",
       "379      38537    credit         9355       1    10408 2004-04-22 2006-02-13   \n",
       "\n",
       "     rate  \n",
       "288  0.29  \n",
       "69   1.29  \n",
       "346  5.78  \n",
       "37   1.14  \n",
       "315  4.75  \n",
       "335  1.56  \n",
       "321  3.27  \n",
       "400  9.82  \n",
       "210  2.74  \n",
       "379  1.64  "
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loans.sample(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "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>loan_id</th>\n",
       "      <th>payment_amount</th>\n",
       "      <th>payment_date</th>\n",
       "      <th>missed</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>738</th>\n",
       "      <td>10407</td>\n",
       "      <td>863</td>\n",
       "      <td>2011-06-23</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>459</th>\n",
       "      <td>11026</td>\n",
       "      <td>2306</td>\n",
       "      <td>2013-12-20</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>10990</td>\n",
       "      <td>2284</td>\n",
       "      <td>2006-04-27</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3433</th>\n",
       "      <td>10154</td>\n",
       "      <td>1367</td>\n",
       "      <td>2002-03-19</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3324</th>\n",
       "      <td>10891</td>\n",
       "      <td>1998</td>\n",
       "      <td>2013-05-26</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2815</th>\n",
       "      <td>11287</td>\n",
       "      <td>2069</td>\n",
       "      <td>2005-12-10</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2969</th>\n",
       "      <td>11361</td>\n",
       "      <td>1507</td>\n",
       "      <td>2007-09-19</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3110</th>\n",
       "      <td>10166</td>\n",
       "      <td>1990</td>\n",
       "      <td>2001-01-11</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2325</th>\n",
       "      <td>11975</td>\n",
       "      <td>510</td>\n",
       "      <td>2003-01-17</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>94</th>\n",
       "      <td>11867</td>\n",
       "      <td>416</td>\n",
       "      <td>2006-05-14</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      loan_id  payment_amount payment_date  missed\n",
       "738     10407             863   2011-06-23       1\n",
       "459     11026            2306   2013-12-20       0\n",
       "12      10990            2284   2006-04-27       1\n",
       "3433    10154            1367   2002-03-19       0\n",
       "3324    10891            1998   2013-05-26       1\n",
       "2815    11287            2069   2005-12-10       0\n",
       "2969    11361            1507   2007-09-19       1\n",
       "3110    10166            1990   2001-01-11       1\n",
       "2325    11975             510   2003-01-17       0\n",
       "94      11867             416   2006-05-14       0"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "payments.sample(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Manual Feature Engineering Examples"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "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>client_id</th>\n",
       "      <th>joined</th>\n",
       "      <th>income</th>\n",
       "      <th>credit_score</th>\n",
       "      <th>join_month</th>\n",
       "      <th>log_income</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>46109</td>\n",
       "      <td>2002-04-16</td>\n",
       "      <td>172677</td>\n",
       "      <td>527</td>\n",
       "      <td>4</td>\n",
       "      <td>12.059178</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>49545</td>\n",
       "      <td>2007-11-14</td>\n",
       "      <td>104564</td>\n",
       "      <td>770</td>\n",
       "      <td>11</td>\n",
       "      <td>11.557555</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>41480</td>\n",
       "      <td>2013-03-11</td>\n",
       "      <td>122607</td>\n",
       "      <td>585</td>\n",
       "      <td>3</td>\n",
       "      <td>11.716739</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>46180</td>\n",
       "      <td>2001-11-06</td>\n",
       "      <td>43851</td>\n",
       "      <td>562</td>\n",
       "      <td>11</td>\n",
       "      <td>10.688553</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>25707</td>\n",
       "      <td>2006-10-06</td>\n",
       "      <td>211422</td>\n",
       "      <td>621</td>\n",
       "      <td>10</td>\n",
       "      <td>12.261611</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   client_id     joined  income  credit_score  join_month  log_income\n",
       "0      46109 2002-04-16  172677           527           4   12.059178\n",
       "1      49545 2007-11-14  104564           770          11   11.557555\n",
       "2      41480 2013-03-11  122607           585           3   11.716739\n",
       "3      46180 2001-11-06   43851           562          11   10.688553\n",
       "4      25707 2006-10-06  211422           621          10   12.261611"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Create a month column\n",
    "clients['join_month'] = clients['joined'].dt.month\n",
    "\n",
    "# Create a log of income column\n",
    "clients['log_income'] = np.log(clients['income'])\n",
    "\n",
    "clients.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "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>mean_loan_amount</th>\n",
       "      <th>max_loan_amount</th>\n",
       "      <th>min_loan_amount</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>client_id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>25707</th>\n",
       "      <td>7963.950000</td>\n",
       "      <td>13913</td>\n",
       "      <td>1212</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26326</th>\n",
       "      <td>7270.062500</td>\n",
       "      <td>13464</td>\n",
       "      <td>1164</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26695</th>\n",
       "      <td>7824.722222</td>\n",
       "      <td>14865</td>\n",
       "      <td>2389</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26945</th>\n",
       "      <td>7125.933333</td>\n",
       "      <td>14593</td>\n",
       "      <td>653</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29841</th>\n",
       "      <td>9813.000000</td>\n",
       "      <td>14837</td>\n",
       "      <td>2778</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           mean_loan_amount  max_loan_amount  min_loan_amount\n",
       "client_id                                                    \n",
       "25707           7963.950000            13913             1212\n",
       "26326           7270.062500            13464             1164\n",
       "26695           7824.722222            14865             2389\n",
       "26945           7125.933333            14593              653\n",
       "29841           9813.000000            14837             2778"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Groupby client id and calculate mean, max, min previous loan size\n",
    "stats = loans.groupby('client_id')['loan_amount'].agg(['mean', 'max', 'min'])\n",
    "stats.columns = ['mean_loan_amount', 'max_loan_amount', 'min_loan_amount']\n",
    "stats.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "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>client_id</th>\n",
       "      <th>joined</th>\n",
       "      <th>income</th>\n",
       "      <th>credit_score</th>\n",
       "      <th>join_month</th>\n",
       "      <th>log_income</th>\n",
       "      <th>mean_loan_amount</th>\n",
       "      <th>max_loan_amount</th>\n",
       "      <th>min_loan_amount</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>46109</td>\n",
       "      <td>2002-04-16</td>\n",
       "      <td>172677</td>\n",
       "      <td>527</td>\n",
       "      <td>4</td>\n",
       "      <td>12.059178</td>\n",
       "      <td>8951.600000</td>\n",
       "      <td>14049</td>\n",
       "      <td>559</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>49545</td>\n",
       "      <td>2007-11-14</td>\n",
       "      <td>104564</td>\n",
       "      <td>770</td>\n",
       "      <td>11</td>\n",
       "      <td>11.557555</td>\n",
       "      <td>10289.300000</td>\n",
       "      <td>14971</td>\n",
       "      <td>3851</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>41480</td>\n",
       "      <td>2013-03-11</td>\n",
       "      <td>122607</td>\n",
       "      <td>585</td>\n",
       "      <td>3</td>\n",
       "      <td>11.716739</td>\n",
       "      <td>7894.850000</td>\n",
       "      <td>14399</td>\n",
       "      <td>811</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>46180</td>\n",
       "      <td>2001-11-06</td>\n",
       "      <td>43851</td>\n",
       "      <td>562</td>\n",
       "      <td>11</td>\n",
       "      <td>10.688553</td>\n",
       "      <td>7700.850000</td>\n",
       "      <td>14081</td>\n",
       "      <td>1607</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>25707</td>\n",
       "      <td>2006-10-06</td>\n",
       "      <td>211422</td>\n",
       "      <td>621</td>\n",
       "      <td>10</td>\n",
       "      <td>12.261611</td>\n",
       "      <td>7963.950000</td>\n",
       "      <td>13913</td>\n",
       "      <td>1212</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>39505</td>\n",
       "      <td>2011-10-14</td>\n",
       "      <td>153873</td>\n",
       "      <td>610</td>\n",
       "      <td>10</td>\n",
       "      <td>11.943883</td>\n",
       "      <td>7424.050000</td>\n",
       "      <td>14575</td>\n",
       "      <td>904</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>32726</td>\n",
       "      <td>2006-05-01</td>\n",
       "      <td>235705</td>\n",
       "      <td>730</td>\n",
       "      <td>5</td>\n",
       "      <td>12.370336</td>\n",
       "      <td>6633.263158</td>\n",
       "      <td>14802</td>\n",
       "      <td>851</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>35089</td>\n",
       "      <td>2010-03-01</td>\n",
       "      <td>131176</td>\n",
       "      <td>771</td>\n",
       "      <td>3</td>\n",
       "      <td>11.784295</td>\n",
       "      <td>6939.200000</td>\n",
       "      <td>13194</td>\n",
       "      <td>773</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>35214</td>\n",
       "      <td>2003-08-08</td>\n",
       "      <td>95849</td>\n",
       "      <td>696</td>\n",
       "      <td>8</td>\n",
       "      <td>11.470529</td>\n",
       "      <td>7173.555556</td>\n",
       "      <td>14767</td>\n",
       "      <td>667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>48177</td>\n",
       "      <td>2008-06-09</td>\n",
       "      <td>190632</td>\n",
       "      <td>769</td>\n",
       "      <td>6</td>\n",
       "      <td>12.158100</td>\n",
       "      <td>7424.368421</td>\n",
       "      <td>14740</td>\n",
       "      <td>659</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   client_id     joined  income  credit_score  join_month  log_income  \\\n",
       "0      46109 2002-04-16  172677           527           4   12.059178   \n",
       "1      49545 2007-11-14  104564           770          11   11.557555   \n",
       "2      41480 2013-03-11  122607           585           3   11.716739   \n",
       "3      46180 2001-11-06   43851           562          11   10.688553   \n",
       "4      25707 2006-10-06  211422           621          10   12.261611   \n",
       "5      39505 2011-10-14  153873           610          10   11.943883   \n",
       "6      32726 2006-05-01  235705           730           5   12.370336   \n",
       "7      35089 2010-03-01  131176           771           3   11.784295   \n",
       "8      35214 2003-08-08   95849           696           8   11.470529   \n",
       "9      48177 2008-06-09  190632           769           6   12.158100   \n",
       "\n",
       "   mean_loan_amount  max_loan_amount  min_loan_amount  \n",
       "0       8951.600000            14049              559  \n",
       "1      10289.300000            14971             3851  \n",
       "2       7894.850000            14399              811  \n",
       "3       7700.850000            14081             1607  \n",
       "4       7963.950000            13913             1212  \n",
       "5       7424.050000            14575              904  \n",
       "6       6633.263158            14802              851  \n",
       "7       6939.200000            13194              773  \n",
       "8       7173.555556            14767              667  \n",
       "9       7424.368421            14740              659  "
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Merge with the clients dataframe\n",
    "clients.merge(stats, left_on = 'client_id', right_index=True, how = 'left').head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Featuretools\n",
    "\n",
    "Featuretools operates on an idea known as Deep Feature Synthesis. The concept of Deep Feature Synthesis is to use basic building blocks known as feature primitives (like the transformations and aggregations done above) that can be stacked on top of each other to form new features. The depth of a \"deep feature\" is equal to the number of stacked primitives.\n",
    "\n",
    "The first part of Featuretools to understand is an `entity`. This is simply a table, or in pandas, a DataFrame. We corral multiple entities into a single object called an `EntitySet`. This is just a large data structure composed of many individual entities and the relationships between them.\n",
    "\n",
    "### EntitySet"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "es = ft.EntitySet(id = 'clients')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Entities"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create an entity from the client dataframe\n",
    "# This dataframe already has an index and a time index\n",
    "es = es.entity_from_dataframe(entity_id = 'clients', dataframe = clients, \n",
    "                              index = 'client_id', time_index = 'joined')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create an entity from the loans dataframe\n",
    "# This dataframe already has an index and a time index\n",
    "es = es.entity_from_dataframe(entity_id = 'loans', dataframe = loans, \n",
    "                              variable_types = {'repaid': ft.variable_types.Categorical},\n",
    "                              index = 'loan_id', \n",
    "                              time_index = 'loan_start')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create an entity from the payments dataframe\n",
    "# This does not yet have a unique index\n",
    "es = es.entity_from_dataframe(entity_id = 'payments', \n",
    "                              dataframe = payments,\n",
    "                              variable_types = {'missed': ft.variable_types.Categorical},\n",
    "                              make_index = True,\n",
    "                              index = 'payment_id',\n",
    "                              time_index = 'payment_date')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Entityset: clients\n",
       "  Entities:\n",
       "    clients [Rows: 25, Columns: 6]\n",
       "    loans [Rows: 443, Columns: 8]\n",
       "    payments [Rows: 3456, Columns: 5]\n",
       "  Relationships:\n",
       "    No relationships"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "es"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Entity: loans\n",
       "  Variables:\n",
       "    loan_id (dtype: index)\n",
       "    client_id (dtype: numeric)\n",
       "    loan_type (dtype: categorical)\n",
       "    loan_amount (dtype: numeric)\n",
       "    loan_start (dtype: datetime_time_index)\n",
       "    loan_end (dtype: datetime)\n",
       "    rate (dtype: numeric)\n",
       "    repaid (dtype: categorical)\n",
       "  Shape:\n",
       "    (Rows: 443, Columns: 8)"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "es['loans']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Entity: payments\n",
       "  Variables:\n",
       "    payment_id (dtype: index)\n",
       "    loan_id (dtype: numeric)\n",
       "    payment_amount (dtype: numeric)\n",
       "    payment_date (dtype: datetime_time_index)\n",
       "    missed (dtype: categorical)\n",
       "  Shape:\n",
       "    (Rows: 3456, Columns: 5)"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "es['payments']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Entity: clients\n",
       "  Variables:\n",
       "    client_id (dtype: index)\n",
       "    joined (dtype: datetime_time_index)\n",
       "    income (dtype: numeric)\n",
       "    credit_score (dtype: numeric)\n",
       "    join_month (dtype: numeric)\n",
       "    log_income (dtype: numeric)\n",
       "  Shape:\n",
       "    (Rows: 25, Columns: 6)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "es['clients']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Relationships"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Relationship between clients and previous loans\n",
    "r_client_previous = ft.Relationship(es['clients']['client_id'],\n",
    "                                    es['loans']['client_id'])\n",
    "\n",
    "# Add the relationship to the entity set\n",
    "es = es.add_relationship(r_client_previous)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Entityset: clients\n",
       "  Entities:\n",
       "    clients [Rows: 25, Columns: 6]\n",
       "    loans [Rows: 443, Columns: 8]\n",
       "    payments [Rows: 3456, Columns: 5]\n",
       "  Relationships:\n",
       "    loans.client_id -> clients.client_id\n",
       "    payments.loan_id -> loans.loan_id"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Relationship between previous loans and previous payments# Relati \n",
    "r_payments = ft.Relationship(es['loans']['loan_id'],\n",
    "                                      es['payments']['loan_id'])\n",
    "\n",
    "# Add the relationship to the entity set\n",
    "es = es.add_relationship(r_payments)\n",
    "\n",
    "es"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature Primitives"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "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>name</th>\n",
       "      <th>type</th>\n",
       "      <th>description</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>trend</td>\n",
       "      <td>aggregation</td>\n",
       "      <td>Calculates the slope of the linear trend of variable overtime.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>sum</td>\n",
       "      <td>aggregation</td>\n",
       "      <td>Counts the number of elements of a numeric or boolean feature.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>max</td>\n",
       "      <td>aggregation</td>\n",
       "      <td>Finds the maximum non-null value of a numeric feature.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>skew</td>\n",
       "      <td>aggregation</td>\n",
       "      <td>Computes the skewness of a data set.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>count</td>\n",
       "      <td>aggregation</td>\n",
       "      <td>Counts the number of non null values.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>min</td>\n",
       "      <td>aggregation</td>\n",
       "      <td>Finds the minimum non-null value of a numeric feature.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>time_since_last</td>\n",
       "      <td>aggregation</td>\n",
       "      <td>Time since last related instance.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>median</td>\n",
       "      <td>aggregation</td>\n",
       "      <td>Finds the median value of any feature with well-ordered values.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>mode</td>\n",
       "      <td>aggregation</td>\n",
       "      <td>Finds the most common element in a categorical feature.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>all</td>\n",
       "      <td>aggregation</td>\n",
       "      <td>Test if all values are 'True'.</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              name         type  \\\n",
       "0            trend  aggregation   \n",
       "1              sum  aggregation   \n",
       "2              max  aggregation   \n",
       "3             skew  aggregation   \n",
       "4            count  aggregation   \n",
       "5              min  aggregation   \n",
       "6  time_since_last  aggregation   \n",
       "7           median  aggregation   \n",
       "8             mode  aggregation   \n",
       "9              all  aggregation   \n",
       "\n",
       "                                                       description  \n",
       "0   Calculates the slope of the linear trend of variable overtime.  \n",
       "1   Counts the number of elements of a numeric or boolean feature.  \n",
       "2           Finds the maximum non-null value of a numeric feature.  \n",
       "3                             Computes the skewness of a data set.  \n",
       "4                            Counts the number of non null values.  \n",
       "5           Finds the minimum non-null value of a numeric feature.  \n",
       "6                                Time since last related instance.  \n",
       "7  Finds the median value of any feature with well-ordered values.  \n",
       "8          Finds the most common element in a categorical feature.  \n",
       "9                                   Test if all values are 'True'.  "
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "primitives = ft.list_primitives()\n",
    "pd.options.display.max_colwidth = 100\n",
    "primitives[primitives['type'] == 'aggregation'].head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "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>name</th>\n",
       "      <th>type</th>\n",
       "      <th>description</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>years</td>\n",
       "      <td>transform</td>\n",
       "      <td>Transform a Timedelta feature into the number of years.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>hour</td>\n",
       "      <td>transform</td>\n",
       "      <td>Transform a Datetime feature into the hour.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>mod</td>\n",
       "      <td>transform</td>\n",
       "      <td>Creates a transform feature that divides two features.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>not</td>\n",
       "      <td>transform</td>\n",
       "      <td>For each value of the base feature, negates the boolean value.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>months</td>\n",
       "      <td>transform</td>\n",
       "      <td>Transform a Timedelta feature into the number of months.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>cum_max</td>\n",
       "      <td>transform</td>\n",
       "      <td>Calculates the max of previous values of an instance for each value in a time-dependent entity.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>day</td>\n",
       "      <td>transform</td>\n",
       "      <td>Transform a Datetime feature into the day.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>multiply</td>\n",
       "      <td>transform</td>\n",
       "      <td>Creates a transform feature that multplies two features.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>isin</td>\n",
       "      <td>transform</td>\n",
       "      <td>For each value of the base feature, checks whether it is in a provided list.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>weeks</td>\n",
       "      <td>transform</td>\n",
       "      <td>Transform a Timedelta feature into the number of weeks.</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        name       type  \\\n",
       "19     years  transform   \n",
       "20      hour  transform   \n",
       "21       mod  transform   \n",
       "22       not  transform   \n",
       "23    months  transform   \n",
       "24   cum_max  transform   \n",
       "25       day  transform   \n",
       "26  multiply  transform   \n",
       "27      isin  transform   \n",
       "28     weeks  transform   \n",
       "\n",
       "                                                                                        description  \n",
       "19                                          Transform a Timedelta feature into the number of years.  \n",
       "20                                                      Transform a Datetime feature into the hour.  \n",
       "21                                           Creates a transform feature that divides two features.  \n",
       "22                                   For each value of the base feature, negates the boolean value.  \n",
       "23                                         Transform a Timedelta feature into the number of months.  \n",
       "24  Calculates the max of previous values of an instance for each value in a time-dependent entity.  \n",
       "25                                                       Transform a Datetime feature into the day.  \n",
       "26                                         Creates a transform feature that multplies two features.  \n",
       "27                     For each value of the base feature, checks whether it is in a provided list.  \n",
       "28                                          Transform a Timedelta feature into the number of weeks.  "
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "primitives[primitives['type'] == 'transform'].head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create new features using specified primitives\n",
    "features, feature_names = ft.dfs(entityset = es, target_entity = 'clients', \n",
    "                                 agg_primitives = ['mean', 'max', 'percent_true', 'last'],\n",
    "                                 trans_primitives = ['years', 'month', 'subtract', 'divide'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "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>MONTH(joined)</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>client_id</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>25707</th>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26326</th>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26695</th>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26945</th>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29841</th>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           MONTH(joined)\n",
       "client_id               \n",
       "25707                 10\n",
       "26326                  5\n",
       "26695                  8\n",
       "26945                 11\n",
       "29841                  8"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame(features['MONTH(joined)'].head())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>income</th>\n",
       "      <th>credit_score</th>\n",
       "      <th>join_month</th>\n",
       "      <th>log_income</th>\n",
       "      <th>MEAN(loans.loan_amount)</th>\n",
       "      <th>MEAN(loans.rate)</th>\n",
       "      <th>MAX(loans.loan_amount)</th>\n",
       "      <th>MAX(loans.rate)</th>\n",
       "      <th>LAST(loans.loan_type)</th>\n",
       "      <th>LAST(loans.loan_amount)</th>\n",
       "      <th>...</th>\n",
       "      <th>income - log_income / credit_score</th>\n",
       "      <th>MEAN(loans.loan_amount) / MEAN(payments.payment_amount)</th>\n",
       "      <th>MEAN(loans.loan_amount) / credit_score - income</th>\n",
       "      <th>LAST(loans.rate) / log_income - income</th>\n",
       "      <th>LAST(payments.payment_amount) / credit_score - join_month</th>\n",
       "      <th>credit_score - join_month / credit_score - income</th>\n",
       "      <th>MAX(loans.loan_amount) / join_month - log_income</th>\n",
       "      <th>log_income - income / MAX(payments.payment_amount)</th>\n",
       "      <th>log_income / credit_score - join_month</th>\n",
       "      <th>credit_score / income - log_income</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>client_id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>25707</th>\n",
       "      <td>211422</td>\n",
       "      <td>621</td>\n",
       "      <td>10</td>\n",
       "      <td>12.261611</td>\n",
       "      <td>7963.950000</td>\n",
       "      <td>3.477000</td>\n",
       "      <td>13913</td>\n",
       "      <td>9.44</td>\n",
       "      <td>home</td>\n",
       "      <td>2203</td>\n",
       "      <td>...</td>\n",
       "      <td>340.434361</td>\n",
       "      <td>6.757398</td>\n",
       "      <td>-0.037779</td>\n",
       "      <td>-0.000035</td>\n",
       "      <td>0.394435</td>\n",
       "      <td>-0.002898</td>\n",
       "      <td>-6151.808355</td>\n",
       "      <td>-78.184075</td>\n",
       "      <td>0.020068</td>\n",
       "      <td>0.002937</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26326</th>\n",
       "      <td>227920</td>\n",
       "      <td>633</td>\n",
       "      <td>5</td>\n",
       "      <td>12.336750</td>\n",
       "      <td>7270.062500</td>\n",
       "      <td>2.517500</td>\n",
       "      <td>13464</td>\n",
       "      <td>6.73</td>\n",
       "      <td>credit</td>\n",
       "      <td>5275</td>\n",
       "      <td>...</td>\n",
       "      <td>360.043702</td>\n",
       "      <td>6.231107</td>\n",
       "      <td>-0.031986</td>\n",
       "      <td>-0.000006</td>\n",
       "      <td>1.484076</td>\n",
       "      <td>-0.002763</td>\n",
       "      <td>-1835.144997</td>\n",
       "      <td>-85.744042</td>\n",
       "      <td>0.019645</td>\n",
       "      <td>0.002777</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26695</th>\n",
       "      <td>174532</td>\n",
       "      <td>680</td>\n",
       "      <td>8</td>\n",
       "      <td>12.069863</td>\n",
       "      <td>7824.722222</td>\n",
       "      <td>2.466111</td>\n",
       "      <td>14865</td>\n",
       "      <td>6.51</td>\n",
       "      <td>other</td>\n",
       "      <td>13918</td>\n",
       "      <td>...</td>\n",
       "      <td>256.646956</td>\n",
       "      <td>6.480456</td>\n",
       "      <td>-0.045008</td>\n",
       "      <td>-0.000005</td>\n",
       "      <td>3.101190</td>\n",
       "      <td>-0.003865</td>\n",
       "      <td>-3652.456752</td>\n",
       "      <td>-59.522486</td>\n",
       "      <td>0.017961</td>\n",
       "      <td>0.003896</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26945</th>\n",
       "      <td>214516</td>\n",
       "      <td>806</td>\n",
       "      <td>11</td>\n",
       "      <td>12.276140</td>\n",
       "      <td>7125.933333</td>\n",
       "      <td>2.855333</td>\n",
       "      <td>14593</td>\n",
       "      <td>5.65</td>\n",
       "      <td>cash</td>\n",
       "      <td>9249</td>\n",
       "      <td>...</td>\n",
       "      <td>266.133652</td>\n",
       "      <td>6.422808</td>\n",
       "      <td>-0.033344</td>\n",
       "      <td>-0.000013</td>\n",
       "      <td>2.008805</td>\n",
       "      <td>-0.003720</td>\n",
       "      <td>-11435.269248</td>\n",
       "      <td>-77.494120</td>\n",
       "      <td>0.015442</td>\n",
       "      <td>0.003758</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29841</th>\n",
       "      <td>38354</td>\n",
       "      <td>523</td>\n",
       "      <td>8</td>\n",
       "      <td>10.554614</td>\n",
       "      <td>9813.000000</td>\n",
       "      <td>3.445000</td>\n",
       "      <td>14837</td>\n",
       "      <td>6.76</td>\n",
       "      <td>home</td>\n",
       "      <td>7223</td>\n",
       "      <td>...</td>\n",
       "      <td>73.314427</td>\n",
       "      <td>6.817266</td>\n",
       "      <td>-0.259390</td>\n",
       "      <td>-0.000133</td>\n",
       "      <td>1.553398</td>\n",
       "      <td>-0.013613</td>\n",
       "      <td>-5807.922213</td>\n",
       "      <td>-13.231003</td>\n",
       "      <td>0.020494</td>\n",
       "      <td>0.013640</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 797 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           income  credit_score  join_month  log_income  \\\n",
       "client_id                                                 \n",
       "25707      211422           621          10   12.261611   \n",
       "26326      227920           633           5   12.336750   \n",
       "26695      174532           680           8   12.069863   \n",
       "26945      214516           806          11   12.276140   \n",
       "29841       38354           523           8   10.554614   \n",
       "\n",
       "           MEAN(loans.loan_amount)  MEAN(loans.rate)  MAX(loans.loan_amount)  \\\n",
       "client_id                                                                      \n",
       "25707                  7963.950000          3.477000                   13913   \n",
       "26326                  7270.062500          2.517500                   13464   \n",
       "26695                  7824.722222          2.466111                   14865   \n",
       "26945                  7125.933333          2.855333                   14593   \n",
       "29841                  9813.000000          3.445000                   14837   \n",
       "\n",
       "           MAX(loans.rate) LAST(loans.loan_type)  LAST(loans.loan_amount)  \\\n",
       "client_id                                                                   \n",
       "25707                 9.44                  home                     2203   \n",
       "26326                 6.73                credit                     5275   \n",
       "26695                 6.51                 other                    13918   \n",
       "26945                 5.65                  cash                     9249   \n",
       "29841                 6.76                  home                     7223   \n",
       "\n",
       "                          ...                  \\\n",
       "client_id                 ...                   \n",
       "25707                     ...                   \n",
       "26326                     ...                   \n",
       "26695                     ...                   \n",
       "26945                     ...                   \n",
       "29841                     ...                   \n",
       "\n",
       "           income - log_income / credit_score  \\\n",
       "client_id                                       \n",
       "25707                              340.434361   \n",
       "26326                              360.043702   \n",
       "26695                              256.646956   \n",
       "26945                              266.133652   \n",
       "29841                               73.314427   \n",
       "\n",
       "           MEAN(loans.loan_amount) / MEAN(payments.payment_amount)  \\\n",
       "client_id                                                            \n",
       "25707                                                     6.757398   \n",
       "26326                                                     6.231107   \n",
       "26695                                                     6.480456   \n",
       "26945                                                     6.422808   \n",
       "29841                                                     6.817266   \n",
       "\n",
       "           MEAN(loans.loan_amount) / credit_score - income  \\\n",
       "client_id                                                    \n",
       "25707                                            -0.037779   \n",
       "26326                                            -0.031986   \n",
       "26695                                            -0.045008   \n",
       "26945                                            -0.033344   \n",
       "29841                                            -0.259390   \n",
       "\n",
       "           LAST(loans.rate) / log_income - income  \\\n",
       "client_id                                           \n",
       "25707                                   -0.000035   \n",
       "26326                                   -0.000006   \n",
       "26695                                   -0.000005   \n",
       "26945                                   -0.000013   \n",
       "29841                                   -0.000133   \n",
       "\n",
       "           LAST(payments.payment_amount) / credit_score - join_month  \\\n",
       "client_id                                                              \n",
       "25707                                                       0.394435   \n",
       "26326                                                       1.484076   \n",
       "26695                                                       3.101190   \n",
       "26945                                                       2.008805   \n",
       "29841                                                       1.553398   \n",
       "\n",
       "           credit_score - join_month / credit_score - income  \\\n",
       "client_id                                                      \n",
       "25707                                              -0.002898   \n",
       "26326                                              -0.002763   \n",
       "26695                                              -0.003865   \n",
       "26945                                              -0.003720   \n",
       "29841                                              -0.013613   \n",
       "\n",
       "           MAX(loans.loan_amount) / join_month - log_income  \\\n",
       "client_id                                                     \n",
       "25707                                          -6151.808355   \n",
       "26326                                          -1835.144997   \n",
       "26695                                          -3652.456752   \n",
       "26945                                         -11435.269248   \n",
       "29841                                          -5807.922213   \n",
       "\n",
       "           log_income - income / MAX(payments.payment_amount)  \\\n",
       "client_id                                                       \n",
       "25707                                              -78.184075   \n",
       "26326                                              -85.744042   \n",
       "26695                                              -59.522486   \n",
       "26945                                              -77.494120   \n",
       "29841                                              -13.231003   \n",
       "\n",
       "           log_income / credit_score - join_month  \\\n",
       "client_id                                           \n",
       "25707                                    0.020068   \n",
       "26326                                    0.019645   \n",
       "26695                                    0.017961   \n",
       "26945                                    0.015442   \n",
       "29841                                    0.020494   \n",
       "\n",
       "           credit_score / income - log_income  \n",
       "client_id                                      \n",
       "25707                                0.002937  \n",
       "26326                                0.002777  \n",
       "26695                                0.003896  \n",
       "26945                                0.003758  \n",
       "29841                                0.013640  \n",
       "\n",
       "[5 rows x 797 columns]"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Deep Feature Synthesis\n",
    "\n",
    "The depth of a feature is simply the number of primitives required to make a feature. So, a feature that relies on a single aggregation would be a deep feature with a depth of 1, a feature that stacks two primitives would have a depth of 2 and so on."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>MEAN(loans.loan_amount)</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>client_id</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>25707</th>\n",
       "      <td>7963.950000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26326</th>\n",
       "      <td>7270.062500</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26695</th>\n",
       "      <td>7824.722222</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26945</th>\n",
       "      <td>7125.933333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29841</th>\n",
       "      <td>9813.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32726</th>\n",
       "      <td>6633.263158</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32885</th>\n",
       "      <td>9920.400000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32961</th>\n",
       "      <td>7882.235294</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35089</th>\n",
       "      <td>6939.200000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35214</th>\n",
       "      <td>7173.555556</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           MEAN(loans.loan_amount)\n",
       "client_id                         \n",
       "25707                  7963.950000\n",
       "26326                  7270.062500\n",
       "26695                  7824.722222\n",
       "26945                  7125.933333\n",
       "29841                  9813.000000\n",
       "32726                  6633.263158\n",
       "32885                  9920.400000\n",
       "32961                  7882.235294\n",
       "35089                  6939.200000\n",
       "35214                  7173.555556"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Show a feature with a depth of 1\n",
    "pd.DataFrame(features['MEAN(loans.loan_amount)'].head(10))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "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>LAST(loans.MEAN(payments.payment_amount))</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>client_id</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>25707</th>\n",
       "      <td>293.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26326</th>\n",
       "      <td>977.375000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26695</th>\n",
       "      <td>1769.166667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26945</th>\n",
       "      <td>1598.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29841</th>\n",
       "      <td>1125.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32726</th>\n",
       "      <td>799.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32885</th>\n",
       "      <td>1729.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32961</th>\n",
       "      <td>282.600000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35089</th>\n",
       "      <td>110.400000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35214</th>\n",
       "      <td>1410.250000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           LAST(loans.MEAN(payments.payment_amount))\n",
       "client_id                                           \n",
       "25707                                     293.500000\n",
       "26326                                     977.375000\n",
       "26695                                    1769.166667\n",
       "26945                                    1598.666667\n",
       "29841                                    1125.500000\n",
       "32726                                     799.500000\n",
       "32885                                    1729.000000\n",
       "32961                                     282.600000\n",
       "35089                                     110.400000\n",
       "35214                                    1410.250000"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Show a feature with a depth of 2\n",
    "pd.DataFrame(features['LAST(loans.MEAN(payments.payment_amount))'].head(10))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Automated Deep Feature Synthesis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Perform deep feature synthesis without specifying primitives\n",
    "features, feature_names = ft.dfs(entityset=es, target_entity='clients', \n",
    "                                 max_depth = 2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>SUM(loans.loan_amount)</th>\n",
       "      <th>SUM(loans.rate)</th>\n",
       "      <th>STD(loans.loan_amount)</th>\n",
       "      <th>STD(loans.rate)</th>\n",
       "      <th>MAX(loans.loan_amount)</th>\n",
       "      <th>MAX(loans.rate)</th>\n",
       "      <th>SKEW(loans.loan_amount)</th>\n",
       "      <th>SKEW(loans.rate)</th>\n",
       "      <th>MIN(loans.loan_amount)</th>\n",
       "      <th>MIN(loans.rate)</th>\n",
       "      <th>...</th>\n",
       "      <th>NUM_UNIQUE(loans.WEEKDAY(loan_end))</th>\n",
       "      <th>MODE(loans.MODE(payments.missed))</th>\n",
       "      <th>MODE(loans.DAY(loan_start))</th>\n",
       "      <th>MODE(loans.DAY(loan_end))</th>\n",
       "      <th>MODE(loans.YEAR(loan_start))</th>\n",
       "      <th>MODE(loans.YEAR(loan_end))</th>\n",
       "      <th>MODE(loans.MONTH(loan_start))</th>\n",
       "      <th>MODE(loans.MONTH(loan_end))</th>\n",
       "      <th>MODE(loans.WEEKDAY(loan_start))</th>\n",
       "      <th>MODE(loans.WEEKDAY(loan_end))</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>client_id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>25707</th>\n",
       "      <td>159279</td>\n",
       "      <td>69.54</td>\n",
       "      <td>4044.418728</td>\n",
       "      <td>2.421285</td>\n",
       "      <td>13913</td>\n",
       "      <td>9.44</td>\n",
       "      <td>-0.172074</td>\n",
       "      <td>0.679118</td>\n",
       "      <td>1212</td>\n",
       "      <td>0.33</td>\n",
       "      <td>...</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>27</td>\n",
       "      <td>1</td>\n",
       "      <td>2010</td>\n",
       "      <td>2007</td>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26326</th>\n",
       "      <td>116321</td>\n",
       "      <td>40.28</td>\n",
       "      <td>4254.149422</td>\n",
       "      <td>1.991819</td>\n",
       "      <td>13464</td>\n",
       "      <td>6.73</td>\n",
       "      <td>0.135246</td>\n",
       "      <td>1.067853</td>\n",
       "      <td>1164</td>\n",
       "      <td>0.50</td>\n",
       "      <td>...</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>6</td>\n",
       "      <td>2003</td>\n",
       "      <td>2005</td>\n",
       "      <td>4</td>\n",
       "      <td>7</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26695</th>\n",
       "      <td>140845</td>\n",
       "      <td>44.39</td>\n",
       "      <td>4078.228493</td>\n",
       "      <td>1.517660</td>\n",
       "      <td>14865</td>\n",
       "      <td>6.51</td>\n",
       "      <td>0.154467</td>\n",
       "      <td>0.820060</td>\n",
       "      <td>2389</td>\n",
       "      <td>0.22</td>\n",
       "      <td>...</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "      <td>2003</td>\n",
       "      <td>2005</td>\n",
       "      <td>9</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26945</th>\n",
       "      <td>106889</td>\n",
       "      <td>42.83</td>\n",
       "      <td>4389.555657</td>\n",
       "      <td>1.564795</td>\n",
       "      <td>14593</td>\n",
       "      <td>5.65</td>\n",
       "      <td>0.156534</td>\n",
       "      <td>-0.001998</td>\n",
       "      <td>653</td>\n",
       "      <td>0.13</td>\n",
       "      <td>...</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>16</td>\n",
       "      <td>1</td>\n",
       "      <td>2002</td>\n",
       "      <td>2004</td>\n",
       "      <td>12</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29841</th>\n",
       "      <td>176634</td>\n",
       "      <td>62.01</td>\n",
       "      <td>4090.630609</td>\n",
       "      <td>2.063092</td>\n",
       "      <td>14837</td>\n",
       "      <td>6.76</td>\n",
       "      <td>-0.212397</td>\n",
       "      <td>0.050600</td>\n",
       "      <td>2778</td>\n",
       "      <td>0.26</td>\n",
       "      <td>...</td>\n",
       "      <td>7</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>15</td>\n",
       "      <td>2005</td>\n",
       "      <td>2007</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 90 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           SUM(loans.loan_amount)  SUM(loans.rate)  STD(loans.loan_amount)  \\\n",
       "client_id                                                                    \n",
       "25707                      159279            69.54             4044.418728   \n",
       "26326                      116321            40.28             4254.149422   \n",
       "26695                      140845            44.39             4078.228493   \n",
       "26945                      106889            42.83             4389.555657   \n",
       "29841                      176634            62.01             4090.630609   \n",
       "\n",
       "           STD(loans.rate)  MAX(loans.loan_amount)  MAX(loans.rate)  \\\n",
       "client_id                                                             \n",
       "25707             2.421285                   13913             9.44   \n",
       "26326             1.991819                   13464             6.73   \n",
       "26695             1.517660                   14865             6.51   \n",
       "26945             1.564795                   14593             5.65   \n",
       "29841             2.063092                   14837             6.76   \n",
       "\n",
       "           SKEW(loans.loan_amount)  SKEW(loans.rate)  MIN(loans.loan_amount)  \\\n",
       "client_id                                                                      \n",
       "25707                    -0.172074          0.679118                    1212   \n",
       "26326                     0.135246          1.067853                    1164   \n",
       "26695                     0.154467          0.820060                    2389   \n",
       "26945                     0.156534         -0.001998                     653   \n",
       "29841                    -0.212397          0.050600                    2778   \n",
       "\n",
       "           MIN(loans.rate)              ...                \\\n",
       "client_id                               ...                 \n",
       "25707                 0.33              ...                 \n",
       "26326                 0.50              ...                 \n",
       "26695                 0.22              ...                 \n",
       "26945                 0.13              ...                 \n",
       "29841                 0.26              ...                 \n",
       "\n",
       "           NUM_UNIQUE(loans.WEEKDAY(loan_end))  \\\n",
       "client_id                                        \n",
       "25707                                        6   \n",
       "26326                                        5   \n",
       "26695                                        6   \n",
       "26945                                        6   \n",
       "29841                                        7   \n",
       "\n",
       "           MODE(loans.MODE(payments.missed))  MODE(loans.DAY(loan_start))  \\\n",
       "client_id                                                                   \n",
       "25707                                      0                           27   \n",
       "26326                                      0                            6   \n",
       "26695                                      0                            3   \n",
       "26945                                      0                           16   \n",
       "29841                                      1                            1   \n",
       "\n",
       "           MODE(loans.DAY(loan_end))  MODE(loans.YEAR(loan_start))  \\\n",
       "client_id                                                            \n",
       "25707                              1                          2010   \n",
       "26326                              6                          2003   \n",
       "26695                             14                          2003   \n",
       "26945                              1                          2002   \n",
       "29841                             15                          2005   \n",
       "\n",
       "          MODE(loans.YEAR(loan_end))  MODE(loans.MONTH(loan_start))  \\\n",
       "client_id                                                             \n",
       "25707                           2007                              1   \n",
       "26326                           2005                              4   \n",
       "26695                           2005                              9   \n",
       "26945                           2004                             12   \n",
       "29841                           2007                              3   \n",
       "\n",
       "           MODE(loans.MONTH(loan_end))  MODE(loans.WEEKDAY(loan_start))  \\\n",
       "client_id                                                                 \n",
       "25707                                8                                3   \n",
       "26326                                7                                5   \n",
       "26695                                4                                1   \n",
       "26945                                5                                0   \n",
       "29841                                2                                5   \n",
       "\n",
       "           MODE(loans.WEEKDAY(loan_end))  \n",
       "client_id                                 \n",
       "25707                                  0  \n",
       "26326                                  2  \n",
       "26695                                  1  \n",
       "26945                                  1  \n",
       "29841                                  1  \n",
       "\n",
       "[5 rows x 90 columns]"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features.iloc[:, 4:].head()"
   ]
  },
  {
   "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": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}