automated-feature-engineering/automated-feature-engineering.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Automated feature engineering with [Deep Feature Synthesis]( https://docs.featuretools.com/)\n",
    "\n",
    "Learning resources:\n",
    "- https://github.com/WillKoehrsen/automated-feature-engineering/blob/master/walk_through/Automated_Feature_Engineering.ipynb\n",
    "- https://towardsdatascience.com/automated-feature-engineering-in-python-99baf11cc219"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import featuretools as ft\n",
    "import time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# utils\n",
    "class Timer:\n",
    "    def __init__(self, descr='Execution time'):\n",
    "        self.ts = None\n",
    "        self.descr = descr\n",
    "\n",
    "    def __enter__(self):\n",
    "        self.ts = time.time()\n",
    "\n",
    "    def __exit__(self, exc_type, exc_val, exc_tb):\n",
    "        timing = '%r  %2.2f minutes' % (self.descr, (time.time() - self.ts) / 60)\n",
    "        print(timing)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Define databse"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "players = pd.DataFrame({'player_id': [0,1,2,3], 'name': ['John', 'Rose', 'Amber', 'Tom']})\n",
    "\n",
    "transfers = pd.DataFrame({'transfer_id': np.arange(15), \n",
    "                          'player_id': np.random.choice([0,1,2,3], size=15), \n",
    "                          'year': np.random.choice(np.arange(2000, 2017), size=15), \n",
    "                          'contract_value ($)': np.random.choice(np.arange(2e5, 1e6, 1e3), size=15)})\n",
    "\n",
    "all_games = pd.DataFrame({'game_id': [0,1,2,3,4,5,6,7], \n",
    "                          'player_id': [0,3,3,1,2,2,1, 2],\n",
    "                          'opponent': [2,1,0,2,1,3,0, 1],\n",
    "                          'scored_points': [10,11,8,3,5,11,0,12],\n",
    "                          'lost_points': [3,8,9,15, 7, 15, 8, 13]\n",
    "                         })\n",
    "\n",
    "all_points = np.sum([all_games['scored_points'], all_games['lost_points']])\n",
    "\n",
    "game_details = pd.DataFrame({'detail_id': np.arange(all_points),\n",
    "                             'game_id': np.random.choice(all_games['game_id'], all_points),\n",
    "                             'score': np.random.choice([-1,2,3,4], all_points),\n",
    "                             'timestamp': np.random.randint(low=10e9, high=10e15, size=(all_points,)) +(time.time()*10e8)})\n",
    "game_details['timestamp'] =pd.to_datetime(game_details['timestamp'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>player_id</th>\n",
       "      <th>name</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>John</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>Rose</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>Amber</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>Tom</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   player_id   name\n",
       "0          0   John\n",
       "1          1   Rose\n",
       "2          2  Amber\n",
       "3          3    Tom"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "players"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>transfer_id</th>\n",
       "      <th>player_id</th>\n",
       "      <th>year</th>\n",
       "      <th>contract_value ($)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2007</td>\n",
       "      <td>282000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2012</td>\n",
       "      <td>842000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>2014</td>\n",
       "      <td>435000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>2013</td>\n",
       "      <td>910000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>2014</td>\n",
       "      <td>398000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "      <td>2013</td>\n",
       "      <td>863000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>6</td>\n",
       "      <td>2</td>\n",
       "      <td>2011</td>\n",
       "      <td>691000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>7</td>\n",
       "      <td>0</td>\n",
       "      <td>2006</td>\n",
       "      <td>705000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>8</td>\n",
       "      <td>2</td>\n",
       "      <td>2006</td>\n",
       "      <td>897000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>9</td>\n",
       "      <td>2</td>\n",
       "      <td>2000</td>\n",
       "      <td>953000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>2001</td>\n",
       "      <td>884000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>11</td>\n",
       "      <td>3</td>\n",
       "      <td>2001</td>\n",
       "      <td>471000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>12</td>\n",
       "      <td>3</td>\n",
       "      <td>2011</td>\n",
       "      <td>738000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>13</td>\n",
       "      <td>1</td>\n",
       "      <td>2013</td>\n",
       "      <td>585000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>14</td>\n",
       "      <td>3</td>\n",
       "      <td>2005</td>\n",
       "      <td>328000.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    transfer_id  player_id  year  contract_value ($)\n",
       "0             0          0  2007            282000.0\n",
       "1             1          0  2012            842000.0\n",
       "2             2          0  2014            435000.0\n",
       "3             3          1  2013            910000.0\n",
       "4             4          0  2014            398000.0\n",
       "5             5          3  2013            863000.0\n",
       "6             6          2  2011            691000.0\n",
       "7             7          0  2006            705000.0\n",
       "8             8          2  2006            897000.0\n",
       "9             9          2  2000            953000.0\n",
       "10           10          0  2001            884000.0\n",
       "11           11          3  2001            471000.0\n",
       "12           12          3  2011            738000.0\n",
       "13           13          1  2013            585000.0\n",
       "14           14          3  2005            328000.0"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "transfers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>game_id</th>\n",
       "      <th>player_id</th>\n",
       "      <th>opponent</th>\n",
       "      <th>scored_points</th>\n",
       "      <th>lost_points</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>10</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>11</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>8</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>11</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>6</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>7</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>12</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   game_id  player_id  opponent  scored_points  lost_points\n",
       "0        0          0         2             10            3\n",
       "1        1          3         1             11            8\n",
       "2        2          3         0              8            9\n",
       "3        3          1         2              3           15\n",
       "4        4          2         1              5            7\n",
       "5        5          2         3             11           15\n",
       "6        6          1         0              0            8\n",
       "7        7          2         1             12           13"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_games"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>detail_id</th>\n",
       "      <th>game_id</th>\n",
       "      <th>score</th>\n",
       "      <th>timestamp</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>2018-09-26 17:40:31.295806976</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "      <td>2018-10-01 16:38:42.781442048</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>2018-09-11 02:17:19.870081280</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "      <td>-1</td>\n",
       "      <td>2018-11-24 21:04:33.298283776</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>2018-12-15 23:33:55.886451200</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   detail_id  game_id  score                     timestamp\n",
       "0          0        4      2 2018-09-26 17:40:31.295806976\n",
       "1          1        5      4 2018-10-01 16:38:42.781442048\n",
       "2          2        2      3 2018-09-11 02:17:19.870081280\n",
       "3          3        7     -1 2018-11-24 21:04:33.298283776\n",
       "4          4        2      4 2018-12-15 23:33:55.886451200"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game_details.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Define new data structure\n",
    "\n",
    "#### EntitySet - is a collection of tables (entities) and the relationships between them"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "es = ft.EntitySet(id='database')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Entityset: database\n",
       "  Entities:\n",
       "  Relationships:\n",
       "    No relationships"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "es"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Create new entities (tables) from dataframes and add them to collection (EntitySet)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Entityset: database\n",
       "  Entities:\n",
       "    players [Rows: 4, Columns: 2]\n",
       "    transfers [Rows: 15, Columns: 4]\n",
       "    all_games [Rows: 8, Columns: 5]\n",
       "  Relationships:\n",
       "    No relationships"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "es.entity_from_dataframe(entity_id='players', dataframe=players, index='player_id')\n",
    "es.entity_from_dataframe(entity_id='transfers', dataframe=transfers, index='transfer_id')\n",
    "es.entity_from_dataframe(entity_id='all_games', dataframe=all_games, index='game_id')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "__ft__ automatically defines variable types however, identifying the correct variable types is important because Featuretools applies different operations to different data types (just as we do when manual feature engineering)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Entityset: database\n",
       "  Entities:\n",
       "    players [Rows: 4, Columns: 2]\n",
       "    transfers [Rows: 15, Columns: 4]\n",
       "    all_games [Rows: 8, Columns: 5]\n",
       "    game_details [Rows: 138, Columns: 4]\n",
       "  Relationships:\n",
       "    No relationships"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "es.entity_from_dataframe(entity_id='game_details', dataframe=game_details, \n",
    "                         index='detail_id', variable_types={\n",
    "                             'timestamp': ft.variable_types.DatetimeTimeIndex\n",
    "                         })"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<Variable: detail_id (dtype = index)>,\n",
       " <Variable: game_id (dtype = numeric)>,\n",
       " <Variable: score (dtype = numeric)>,\n",
       " <Variable: timestamp (dtype: datetime_time_index, format: None)>]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "es['game_details'].variables"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Define & apply relations between tables (entities)\n",
    "``` __init__(parent_variable, child_variable) ```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "linking_id = 'player_id'\n",
    "\n",
    "parent_variable=es['players'][linking_id]\n",
    "child_variable=es['transfers'][linking_id]\n",
    "\n",
    "players_transfers_relationship = ft.Relationship(parent_variable, child_variable)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Entityset: database\n",
       "  Entities:\n",
       "    players [Rows: 4, Columns: 2]\n",
       "    transfers [Rows: 15, Columns: 4]\n",
       "    all_games [Rows: 8, Columns: 5]\n",
       "    game_details [Rows: 138, Columns: 4]\n",
       "  Relationships:\n",
       "    transfers.player_id -> players.player_id\n",
       "    all_games.player_id -> players.player_id\n",
       "    game_details.game_id -> all_games.game_id"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "relationships = [players_transfers_relationship,\n",
    "                 ft.Relationship(es['players']['player_id'], es['all_games']['player_id']),\n",
    "                 ft.Relationship(es['all_games']['game_id'], es['game_details']['game_id'])\n",
    "                ]\n",
    "es.add_relationships(relationships)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Generate new features\n",
    "\n",
    "- Featuretools will automatically create many features from a set of related tables - using operations: transofmrations and aggregations.\n",
    "- Deep feature synthesis (dfs) - stacks multiple transformation and aggregation operations across tables. Its depth is defined by parameter. \n",
    "\n",
    "Operations in featuretools ecosystem are called [feature-primitives](https://docs.featuretools.com/automated_feature_engineering/primitives.html)\n",
    "- Transformation - acts on a single table by creating new features out of one or more of the existing columns. \n",
    "- Aggregations - using one-to-many relationship groups observations and then calculate statistics.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "agg_primitives = [\"sum\", \"std\", \"max\", \"skew\", \"min\", \"mean\", \"count\", \"percent_true\", \"mode\"] # defaults\n",
    "trans_primitives = ['time_since_previous', \"year\",'cum_count', 'diff', 'and', 'or']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "__Note:__ When dfs is called with features_only=True, only feature definitions are returned as output. By default this parameter is set to False. This parameter is used quickly inspect the feature definitions before the spending time calculating the feature matrix."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "'Execution time'  0.20 minutes\n"
     ]
    }
   ],
   "source": [
    "with Timer():\n",
    "    feature_defs = ft.dfs(entityset=es,\n",
    "                          features_only=True,\n",
    "                          target_entity='players',\n",
    "                          max_depth=4,\n",
    "                          agg_primitives=agg_primitives,\n",
    "                          trans_primitives=trans_primitives)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "'Execution time'  0.24 minutes\n"
     ]
    }
   ],
   "source": [
    "with Timer():\n",
    "    feature_matrix, feature_defs = ft.dfs(entityset=es,\n",
    "                          features_only=False,\n",
    "                          target_entity='players',\n",
    "                          max_depth=4,\n",
    "                          agg_primitives=agg_primitives,\n",
    "                          trans_primitives=trans_primitives)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Retured value is a Dataframe with new features.\n",
    "Not all of them will be crucial for model ale even make sense, however using this tool we can leverage automation. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "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>SUM(transfers.year)</th>\n",
       "      <th>SUM(transfers.contract_value ($))</th>\n",
       "      <th>STD(transfers.year)</th>\n",
       "      <th>STD(transfers.contract_value ($))</th>\n",
       "      <th>MAX(transfers.year)</th>\n",
       "      <th>MAX(transfers.contract_value ($))</th>\n",
       "      <th>SKEW(transfers.year)</th>\n",
       "      <th>SKEW(transfers.contract_value ($))</th>\n",
       "      <th>MIN(transfers.year)</th>\n",
       "      <th>...</th>\n",
       "      <th>MEAN(all_games.DIFF(STD(game_details.DIFF(score by game_id)) by player_id))</th>\n",
       "      <th>MEAN(all_games.DIFF(MEAN(game_details.CUM_COUNT(detail_id by game_id)) by player_id))</th>\n",
       "      <th>MEAN(all_games.DIFF(MEAN(game_details.DIFF(score by game_id)) by player_id))</th>\n",
       "      <th>MEAN(all_games.DIFF(SUM(game_details.time_since_previous_by_game_id) by player_id))</th>\n",
       "      <th>MEAN(all_games.DIFF(MIN(game_details.DIFF(score by game_id)) by player_id))</th>\n",
       "      <th>MEAN(all_games.DIFF(MEAN(game_details.time_since_previous_by_game_id) by player_id))</th>\n",
       "      <th>MEAN(all_games.DIFF(STD(game_details.CUM_COUNT(detail_id by game_id)) by player_id))</th>\n",
       "      <th>MEAN(all_games.DIFF(MAX(game_details.DIFF(score by game_id)) by player_id))</th>\n",
       "      <th>MEAN(all_games.DIFF(SKEW(game_details.time_since_previous_by_game_id) by player_id))</th>\n",
       "      <th>MEAN(all_games.DIFF(SKEW(game_details.DIFF(score by game_id)) by player_id))</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>player_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>0</th>\n",
       "      <td>John</td>\n",
       "      <td>12054</td>\n",
       "      <td>3546000.0</td>\n",
       "      <td>4.760952</td>\n",
       "      <td>230547.175216</td>\n",
       "      <td>2014</td>\n",
       "      <td>884000.0</td>\n",
       "      <td>-0.416995</td>\n",
       "      <td>0.026650</td>\n",
       "      <td>2001</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Rose</td>\n",
       "      <td>4026</td>\n",
       "      <td>1495000.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>162500.000000</td>\n",
       "      <td>2013</td>\n",
       "      <td>910000.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>2013</td>\n",
       "      <td>...</td>\n",
       "      <td>0.191535</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-9.184019e+05</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-24714.153038</td>\n",
       "      <td>0.289056</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Amber</td>\n",
       "      <td>6017</td>\n",
       "      <td>2541000.0</td>\n",
       "      <td>4.496913</td>\n",
       "      <td>112652.859114</td>\n",
       "      <td>2011</td>\n",
       "      <td>953000.0</td>\n",
       "      <td>-0.110780</td>\n",
       "      <td>-0.578327</td>\n",
       "      <td>2000</td>\n",
       "      <td>...</td>\n",
       "      <td>0.317837</td>\n",
       "      <td>-0.5</td>\n",
       "      <td>0.125000</td>\n",
       "      <td>-7.339127e+04</td>\n",
       "      <td>0.5</td>\n",
       "      <td>-15435.513384</td>\n",
       "      <td>-0.289690</td>\n",
       "      <td>0.5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Tom</td>\n",
       "      <td>8030</td>\n",
       "      <td>2400000.0</td>\n",
       "      <td>4.769696</td>\n",
       "      <td>211446.210654</td>\n",
       "      <td>2013</td>\n",
       "      <td>863000.0</td>\n",
       "      <td>-0.186618</td>\n",
       "      <td>-0.038367</td>\n",
       "      <td>2001</td>\n",
       "      <td>...</td>\n",
       "      <td>0.619988</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.025974</td>\n",
       "      <td>8.649261e+06</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>410506.142760</td>\n",
       "      <td>2.312756</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>4 rows × 480 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            name  SUM(transfers.year)  SUM(transfers.contract_value ($))  \\\n",
       "player_id                                                                  \n",
       "0           John                12054                          3546000.0   \n",
       "1           Rose                 4026                          1495000.0   \n",
       "2          Amber                 6017                          2541000.0   \n",
       "3            Tom                 8030                          2400000.0   \n",
       "\n",
       "           STD(transfers.year)  STD(transfers.contract_value ($))  \\\n",
       "player_id                                                           \n",
       "0                     4.760952                      230547.175216   \n",
       "1                     0.000000                      162500.000000   \n",
       "2                     4.496913                      112652.859114   \n",
       "3                     4.769696                      211446.210654   \n",
       "\n",
       "           MAX(transfers.year)  MAX(transfers.contract_value ($))  \\\n",
       "player_id                                                           \n",
       "0                         2014                           884000.0   \n",
       "1                         2013                           910000.0   \n",
       "2                         2011                           953000.0   \n",
       "3                         2013                           863000.0   \n",
       "\n",
       "           SKEW(transfers.year)  SKEW(transfers.contract_value ($))  \\\n",
       "player_id                                                             \n",
       "0                     -0.416995                            0.026650   \n",
       "1                      0.000000                            0.000000   \n",
       "2                     -0.110780                           -0.578327   \n",
       "3                     -0.186618                           -0.038367   \n",
       "\n",
       "           MIN(transfers.year)  \\\n",
       "player_id                        \n",
       "0                         2001   \n",
       "1                         2013   \n",
       "2                         2000   \n",
       "3                         2001   \n",
       "\n",
       "                                               ...                                       \\\n",
       "player_id                                      ...                                        \n",
       "0                                              ...                                        \n",
       "1                                              ...                                        \n",
       "2                                              ...                                        \n",
       "3                                              ...                                        \n",
       "\n",
       "           MEAN(all_games.DIFF(STD(game_details.DIFF(score by game_id)) by player_id))  \\\n",
       "player_id                                                                                \n",
       "0                                                        NaN                             \n",
       "1                                                   0.191535                             \n",
       "2                                                   0.317837                             \n",
       "3                                                   0.619988                             \n",
       "\n",
       "           MEAN(all_games.DIFF(MEAN(game_details.CUM_COUNT(detail_id by game_id)) by player_id))  \\\n",
       "player_id                                                                                          \n",
       "0                                                        NaN                                       \n",
       "1                                                        0.5                                       \n",
       "2                                                       -0.5                                       \n",
       "3                                                        4.0                                       \n",
       "\n",
       "           MEAN(all_games.DIFF(MEAN(game_details.DIFF(score by game_id)) by player_id))  \\\n",
       "player_id                                                                                 \n",
       "0                                                        NaN                              \n",
       "1                                                   0.000000                              \n",
       "2                                                   0.125000                              \n",
       "3                                                   0.025974                              \n",
       "\n",
       "           MEAN(all_games.DIFF(SUM(game_details.time_since_previous_by_game_id) by player_id))  \\\n",
       "player_id                                                                                        \n",
       "0                                                        NaN                                     \n",
       "1                                              -9.184019e+05                                     \n",
       "2                                              -7.339127e+04                                     \n",
       "3                                               8.649261e+06                                     \n",
       "\n",
       "           MEAN(all_games.DIFF(MIN(game_details.DIFF(score by game_id)) by player_id))  \\\n",
       "player_id                                                                                \n",
       "0                                                        NaN                             \n",
       "1                                                        0.0                             \n",
       "2                                                        0.5                             \n",
       "3                                                       -1.0                             \n",
       "\n",
       "           MEAN(all_games.DIFF(MEAN(game_details.time_since_previous_by_game_id) by player_id))  \\\n",
       "player_id                                                                                         \n",
       "0                                                        NaN                                      \n",
       "1                                              -24714.153038                                      \n",
       "2                                              -15435.513384                                      \n",
       "3                                              410506.142760                                      \n",
       "\n",
       "           MEAN(all_games.DIFF(STD(game_details.CUM_COUNT(detail_id by game_id)) by player_id))  \\\n",
       "player_id                                                                                         \n",
       "0                                                        NaN                                      \n",
       "1                                                   0.289056                                      \n",
       "2                                                  -0.289690                                      \n",
       "3                                                   2.312756                                      \n",
       "\n",
       "           MEAN(all_games.DIFF(MAX(game_details.DIFF(score by game_id)) by player_id))  \\\n",
       "player_id                                                                                \n",
       "0                                                        NaN                             \n",
       "1                                                        1.0                             \n",
       "2                                                        0.5                             \n",
       "3                                                        1.0                             \n",
       "\n",
       "           MEAN(all_games.DIFF(SKEW(game_details.time_since_previous_by_game_id) by player_id))  \\\n",
       "player_id                                                                                         \n",
       "0                                                        NaN                                      \n",
       "1                                                        NaN                                      \n",
       "2                                                        NaN                                      \n",
       "3                                                        NaN                                      \n",
       "\n",
       "           MEAN(all_games.DIFF(SKEW(game_details.DIFF(score by game_id)) by player_id))  \n",
       "player_id                                                                                \n",
       "0                                                        NaN                             \n",
       "1                                                        NaN                             \n",
       "2                                                        NaN                             \n",
       "3                                                        NaN                             \n",
       "\n",
       "[4 rows x 480 columns]"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "feature_matrix"
   ]
  }
 ],
 "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": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}