Spark Advanced Stats Model with 2 predictions.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pyspark\n",
    "from pyspark.sql import SparkSession\n",
    "from pyspark.sql import functions as f\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import time\n",
    "from itertools import chain"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from nba_api.stats.endpoints import leaguegamefinder\n",
    "from nba_api.stats.endpoints import teamdashboardbylastngames\n",
    "from nba_api.stats.endpoints import leaguedashteamstats\n",
    "from nba_api.stats.endpoints import boxscoreadvancedv2\n",
    "from nba_api.stats.static import teams"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Get a dataframe that contains team ids, abbervations, and nicknames for later use."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "team_df = pd.DataFrame(teams.get_teams())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "team_df = team_df[['id', 'abbreviation', 'nickname']]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Pull all games so we can store the game ids & build rows of matchups"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_games = leaguegamefinder.LeagueGameFinder(season_nullable='2020-21', league_id_nullable='00', season_type_nullable='Regular Season').get_data_frames()[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_game_ids = set(all_games['GAME_ID'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Use games ids to create a dataframe of advanced metrics for every game"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "season_advanced=[]\n",
    "for i in all_game_ids:\n",
    "    season_advanced.append(boxscoreadvancedv2.BoxScoreAdvancedV2(game_id=i).get_normalized_dict()['TeamStats'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "szn_adv_df = pd.DataFrame(list(chain.from_iterable(season_advanced)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(all_games) == len(szn_adv_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Implement Spark"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "spark = SparkSession.builder.appName('NBA_Advanced_Stats_Model').getOrCreate()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "team_spark = spark.createDataFrame(team_df)\n",
    "all_games_spark = spark.createDataFrame(all_games)\n",
    "szn_adv_spark = spark.createDataFrame(szn_adv_df)"
   ]
  },
  {
   "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>SEASON_ID</th>\n",
       "      <th>TEAM_ID</th>\n",
       "      <th>TEAM_ABBREVIATION</th>\n",
       "      <th>TEAM_NAME</th>\n",
       "      <th>GAME_ID</th>\n",
       "      <th>GAME_DATE</th>\n",
       "      <th>MATCHUP</th>\n",
       "      <th>WL</th>\n",
       "      <th>MIN</th>\n",
       "      <th>PTS</th>\n",
       "      <th>...</th>\n",
       "      <th>FT_PCT</th>\n",
       "      <th>OREB</th>\n",
       "      <th>DREB</th>\n",
       "      <th>REB</th>\n",
       "      <th>AST</th>\n",
       "      <th>STL</th>\n",
       "      <th>BLK</th>\n",
       "      <th>TOV</th>\n",
       "      <th>PF</th>\n",
       "      <th>PLUS_MINUS</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>22020</td>\n",
       "      <td>1610612759</td>\n",
       "      <td>SAS</td>\n",
       "      <td>San Antonio Spurs</td>\n",
       "      <td>0022001065</td>\n",
       "      <td>2021-05-15</td>\n",
       "      <td>SAS vs. PHX</td>\n",
       "      <td>L</td>\n",
       "      <td>240</td>\n",
       "      <td>103</td>\n",
       "      <td>...</td>\n",
       "      <td>0.875</td>\n",
       "      <td>7</td>\n",
       "      <td>34</td>\n",
       "      <td>41</td>\n",
       "      <td>23</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>16</td>\n",
       "      <td>17</td>\n",
       "      <td>-37.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>22020</td>\n",
       "      <td>1610612756</td>\n",
       "      <td>PHX</td>\n",
       "      <td>Phoenix Suns</td>\n",
       "      <td>0022001065</td>\n",
       "      <td>2021-05-15</td>\n",
       "      <td>PHX @ SAS</td>\n",
       "      <td>W</td>\n",
       "      <td>238</td>\n",
       "      <td>140</td>\n",
       "      <td>...</td>\n",
       "      <td>0.955</td>\n",
       "      <td>7</td>\n",
       "      <td>40</td>\n",
       "      <td>47</td>\n",
       "      <td>32</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "      <td>15</td>\n",
       "      <td>37.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>22020</td>\n",
       "      <td>1610612766</td>\n",
       "      <td>CHA</td>\n",
       "      <td>Charlotte Hornets</td>\n",
       "      <td>0022001064</td>\n",
       "      <td>2021-05-15</td>\n",
       "      <td>CHA @ NYK</td>\n",
       "      <td>L</td>\n",
       "      <td>265</td>\n",
       "      <td>109</td>\n",
       "      <td>...</td>\n",
       "      <td>0.750</td>\n",
       "      <td>15</td>\n",
       "      <td>36</td>\n",
       "      <td>51</td>\n",
       "      <td>28</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>8</td>\n",
       "      <td>18</td>\n",
       "      <td>-9.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>22020</td>\n",
       "      <td>1610612752</td>\n",
       "      <td>NYK</td>\n",
       "      <td>New York Knicks</td>\n",
       "      <td>0022001064</td>\n",
       "      <td>2021-05-15</td>\n",
       "      <td>NYK vs. CHA</td>\n",
       "      <td>W</td>\n",
       "      <td>263</td>\n",
       "      <td>118</td>\n",
       "      <td>...</td>\n",
       "      <td>0.800</td>\n",
       "      <td>7</td>\n",
       "      <td>43</td>\n",
       "      <td>50</td>\n",
       "      <td>29</td>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "      <td>9</td>\n",
       "      <td>20</td>\n",
       "      <td>9.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>4 rows × 28 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   SEASON_ID     TEAM_ID TEAM_ABBREVIATION          TEAM_NAME     GAME_ID  \\\n",
       "0      22020  1610612759               SAS  San Antonio Spurs  0022001065   \n",
       "11     22020  1610612756               PHX       Phoenix Suns  0022001065   \n",
       "10     22020  1610612766               CHA  Charlotte Hornets  0022001064   \n",
       "5      22020  1610612752               NYK    New York Knicks  0022001064   \n",
       "\n",
       "     GAME_DATE      MATCHUP WL  MIN  PTS  ...  FT_PCT  OREB  DREB  REB  AST  \\\n",
       "0   2021-05-15  SAS vs. PHX  L  240  103  ...   0.875     7    34   41   23   \n",
       "11  2021-05-15    PHX @ SAS  W  238  140  ...   0.955     7    40   47   32   \n",
       "10  2021-05-15    CHA @ NYK  L  265  109  ...   0.750    15    36   51   28   \n",
       "5   2021-05-15  NYK vs. CHA  W  263  118  ...   0.800     7    43   50   29   \n",
       "\n",
       "    STL  BLK  TOV  PF  PLUS_MINUS  \n",
       "0     5    2   16  17       -37.0  \n",
       "11   10    1   10  15        37.0  \n",
       "10    5    2    8  18        -9.0  \n",
       "5     3    7    9  20         9.0  \n",
       "\n",
       "[4 rows x 28 columns]"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_games_spark.toPandas().sort_values('GAME_ID', ascending=False).head(4)"
   ]
  },
  {
   "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>GAME_ID</th>\n",
       "      <th>TEAM_ID</th>\n",
       "      <th>TEAM_NAME</th>\n",
       "      <th>TEAM_ABBREVIATION</th>\n",
       "      <th>TEAM_CITY</th>\n",
       "      <th>MIN</th>\n",
       "      <th>E_OFF_RATING</th>\n",
       "      <th>OFF_RATING</th>\n",
       "      <th>E_DEF_RATING</th>\n",
       "      <th>DEF_RATING</th>\n",
       "      <th>...</th>\n",
       "      <th>TM_TOV_PCT</th>\n",
       "      <th>EFG_PCT</th>\n",
       "      <th>TS_PCT</th>\n",
       "      <th>USG_PCT</th>\n",
       "      <th>E_USG_PCT</th>\n",
       "      <th>E_PACE</th>\n",
       "      <th>PACE</th>\n",
       "      <th>PACE_PER40</th>\n",
       "      <th>POSS</th>\n",
       "      <th>PIE</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>174</th>\n",
       "      <td>0022001065</td>\n",
       "      <td>1610612759</td>\n",
       "      <td>Spurs</td>\n",
       "      <td>SAS</td>\n",
       "      <td>San Antonio</td>\n",
       "      <td>240:00</td>\n",
       "      <td>95.8</td>\n",
       "      <td>96.3</td>\n",
       "      <td>131.2</td>\n",
       "      <td>132.1</td>\n",
       "      <td>...</td>\n",
       "      <td>15.0</td>\n",
       "      <td>0.466</td>\n",
       "      <td>0.523</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.199</td>\n",
       "      <td>107.12</td>\n",
       "      <td>106.50</td>\n",
       "      <td>88.75</td>\n",
       "      <td>107</td>\n",
       "      <td>0.348</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>175</th>\n",
       "      <td>0022001065</td>\n",
       "      <td>1610612756</td>\n",
       "      <td>Suns</td>\n",
       "      <td>PHX</td>\n",
       "      <td>Phoenix</td>\n",
       "      <td>240:00</td>\n",
       "      <td>131.2</td>\n",
       "      <td>132.1</td>\n",
       "      <td>95.8</td>\n",
       "      <td>96.3</td>\n",
       "      <td>...</td>\n",
       "      <td>9.4</td>\n",
       "      <td>0.633</td>\n",
       "      <td>0.675</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.199</td>\n",
       "      <td>107.12</td>\n",
       "      <td>106.50</td>\n",
       "      <td>88.75</td>\n",
       "      <td>106</td>\n",
       "      <td>0.652</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1305</th>\n",
       "      <td>0022001064</td>\n",
       "      <td>1610612766</td>\n",
       "      <td>Hornets</td>\n",
       "      <td>CHA</td>\n",
       "      <td>Charlotte</td>\n",
       "      <td>265:00</td>\n",
       "      <td>103.3</td>\n",
       "      <td>110.1</td>\n",
       "      <td>120.9</td>\n",
       "      <td>119.2</td>\n",
       "      <td>...</td>\n",
       "      <td>8.1</td>\n",
       "      <td>0.446</td>\n",
       "      <td>0.484</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.197</td>\n",
       "      <td>92.00</td>\n",
       "      <td>89.66</td>\n",
       "      <td>74.72</td>\n",
       "      <td>99</td>\n",
       "      <td>0.425</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1304</th>\n",
       "      <td>0022001064</td>\n",
       "      <td>1610612752</td>\n",
       "      <td>Knicks</td>\n",
       "      <td>NYK</td>\n",
       "      <td>New York</td>\n",
       "      <td>265:00</td>\n",
       "      <td>120.9</td>\n",
       "      <td>119.2</td>\n",
       "      <td>103.3</td>\n",
       "      <td>110.1</td>\n",
       "      <td>...</td>\n",
       "      <td>10.1</td>\n",
       "      <td>0.602</td>\n",
       "      <td>0.624</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.199</td>\n",
       "      <td>92.00</td>\n",
       "      <td>89.66</td>\n",
       "      <td>74.72</td>\n",
       "      <td>99</td>\n",
       "      <td>0.575</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>4 rows × 29 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         GAME_ID     TEAM_ID TEAM_NAME TEAM_ABBREVIATION    TEAM_CITY     MIN  \\\n",
       "174   0022001065  1610612759     Spurs               SAS  San Antonio  240:00   \n",
       "175   0022001065  1610612756      Suns               PHX      Phoenix  240:00   \n",
       "1305  0022001064  1610612766   Hornets               CHA    Charlotte  265:00   \n",
       "1304  0022001064  1610612752    Knicks               NYK     New York  265:00   \n",
       "\n",
       "      E_OFF_RATING  OFF_RATING  E_DEF_RATING  DEF_RATING  ...  TM_TOV_PCT  \\\n",
       "174           95.8        96.3         131.2       132.1  ...        15.0   \n",
       "175          131.2       132.1          95.8        96.3  ...         9.4   \n",
       "1305         103.3       110.1         120.9       119.2  ...         8.1   \n",
       "1304         120.9       119.2         103.3       110.1  ...        10.1   \n",
       "\n",
       "      EFG_PCT  TS_PCT  USG_PCT  E_USG_PCT  E_PACE    PACE  PACE_PER40  POSS  \\\n",
       "174     0.466   0.523      1.0      0.199  107.12  106.50       88.75   107   \n",
       "175     0.633   0.675      1.0      0.199  107.12  106.50       88.75   106   \n",
       "1305    0.446   0.484      1.0      0.197   92.00   89.66       74.72    99   \n",
       "1304    0.602   0.624      1.0      0.199   92.00   89.66       74.72    99   \n",
       "\n",
       "        PIE  \n",
       "174   0.348  \n",
       "175   0.652  \n",
       "1305  0.425  \n",
       "1304  0.575  \n",
       "\n",
       "[4 rows x 29 columns]"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "szn_adv_spark.toPandas().sort_values('GAME_ID', ascending=False).head(4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "root\n",
      " |-- SEASON_ID: string (nullable = true)\n",
      " |-- TEAM_ID: long (nullable = true)\n",
      " |-- TEAM_ABBREVIATION: string (nullable = true)\n",
      " |-- TEAM_NAME: string (nullable = true)\n",
      " |-- GAME_ID: string (nullable = true)\n",
      " |-- GAME_DATE: string (nullable = true)\n",
      " |-- MATCHUP: string (nullable = true)\n",
      " |-- WL: string (nullable = true)\n",
      " |-- MIN: long (nullable = true)\n",
      " |-- PTS: long (nullable = true)\n",
      " |-- FGM: long (nullable = true)\n",
      " |-- FGA: long (nullable = true)\n",
      " |-- FG_PCT: double (nullable = true)\n",
      " |-- FG3M: long (nullable = true)\n",
      " |-- FG3A: long (nullable = true)\n",
      " |-- FG3_PCT: double (nullable = true)\n",
      " |-- FTM: long (nullable = true)\n",
      " |-- FTA: long (nullable = true)\n",
      " |-- FT_PCT: double (nullable = true)\n",
      " |-- OREB: long (nullable = true)\n",
      " |-- DREB: long (nullable = true)\n",
      " |-- REB: long (nullable = true)\n",
      " |-- AST: long (nullable = true)\n",
      " |-- STL: long (nullable = true)\n",
      " |-- BLK: long (nullable = true)\n",
      " |-- TOV: long (nullable = true)\n",
      " |-- PF: long (nullable = true)\n",
      " |-- PLUS_MINUS: double (nullable = true)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "all_games_spark.printSchema()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Combine DFs to get mathcup data for splitting and rejoinin home and away"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_games_spark.createOrReplaceTempView('all_games_V')\n",
    "szn_adv_spark.createOrReplaceTempView('szn_adv_V')\n",
    "df = spark.sql('SELECT a.MATCHUP, a.WL, a.PF, s.* FROM all_games_V a RIGHT JOIN szn_adv_V s \\\n",
    "                        ON a.GAME_ID = s.GAME_ID AND a.TEAM_ID = s.TEAM_ID')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "root\n",
      " |-- MATCHUP: string (nullable = true)\n",
      " |-- WL: string (nullable = true)\n",
      " |-- PF: long (nullable = true)\n",
      " |-- GAME_ID: string (nullable = true)\n",
      " |-- TEAM_ID: long (nullable = true)\n",
      " |-- TEAM_NAME: string (nullable = true)\n",
      " |-- TEAM_ABBREVIATION: string (nullable = true)\n",
      " |-- TEAM_CITY: string (nullable = true)\n",
      " |-- MIN: string (nullable = true)\n",
      " |-- E_OFF_RATING: double (nullable = true)\n",
      " |-- OFF_RATING: double (nullable = true)\n",
      " |-- E_DEF_RATING: double (nullable = true)\n",
      " |-- DEF_RATING: double (nullable = true)\n",
      " |-- E_NET_RATING: double (nullable = true)\n",
      " |-- NET_RATING: double (nullable = true)\n",
      " |-- AST_PCT: double (nullable = true)\n",
      " |-- AST_TOV: double (nullable = true)\n",
      " |-- AST_RATIO: double (nullable = true)\n",
      " |-- OREB_PCT: double (nullable = true)\n",
      " |-- DREB_PCT: double (nullable = true)\n",
      " |-- REB_PCT: double (nullable = true)\n",
      " |-- E_TM_TOV_PCT: double (nullable = true)\n",
      " |-- TM_TOV_PCT: double (nullable = true)\n",
      " |-- EFG_PCT: double (nullable = true)\n",
      " |-- TS_PCT: double (nullable = true)\n",
      " |-- USG_PCT: double (nullable = true)\n",
      " |-- E_USG_PCT: double (nullable = true)\n",
      " |-- E_PACE: double (nullable = true)\n",
      " |-- PACE: double (nullable = true)\n",
      " |-- PACE_PER40: double (nullable = true)\n",
      " |-- POSS: long (nullable = true)\n",
      " |-- PIE: double (nullable = true)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df.printSchema()"
   ]
  },
  {
   "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>MATCHUP</th>\n",
       "      <th>WL</th>\n",
       "      <th>PF</th>\n",
       "      <th>GAME_ID</th>\n",
       "      <th>TEAM_ID</th>\n",
       "      <th>TEAM_NAME</th>\n",
       "      <th>TEAM_ABBREVIATION</th>\n",
       "      <th>TEAM_CITY</th>\n",
       "      <th>MIN</th>\n",
       "      <th>E_OFF_RATING</th>\n",
       "      <th>...</th>\n",
       "      <th>TM_TOV_PCT</th>\n",
       "      <th>EFG_PCT</th>\n",
       "      <th>TS_PCT</th>\n",
       "      <th>USG_PCT</th>\n",
       "      <th>E_USG_PCT</th>\n",
       "      <th>E_PACE</th>\n",
       "      <th>PACE</th>\n",
       "      <th>PACE_PER40</th>\n",
       "      <th>POSS</th>\n",
       "      <th>PIE</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>TOR @ SAS</td>\n",
       "      <td>L</td>\n",
       "      <td>23</td>\n",
       "      <td>0022000027</td>\n",
       "      <td>1610612761</td>\n",
       "      <td>Raptors</td>\n",
       "      <td>TOR</td>\n",
       "      <td>Toronto</td>\n",
       "      <td>240:00</td>\n",
       "      <td>109.0</td>\n",
       "      <td>...</td>\n",
       "      <td>11.5</td>\n",
       "      <td>0.536</td>\n",
       "      <td>0.556</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.199</td>\n",
       "      <td>106.82</td>\n",
       "      <td>104.0</td>\n",
       "      <td>86.67</td>\n",
       "      <td>104</td>\n",
       "      <td>0.491</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ORL @ MIN</td>\n",
       "      <td>W</td>\n",
       "      <td>17</td>\n",
       "      <td>0022000221</td>\n",
       "      <td>1610612753</td>\n",
       "      <td>Magic</td>\n",
       "      <td>ORL</td>\n",
       "      <td>Orlando</td>\n",
       "      <td>240:00</td>\n",
       "      <td>96.8</td>\n",
       "      <td>...</td>\n",
       "      <td>12.5</td>\n",
       "      <td>0.446</td>\n",
       "      <td>0.479</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.198</td>\n",
       "      <td>99.70</td>\n",
       "      <td>95.5</td>\n",
       "      <td>79.58</td>\n",
       "      <td>96</td>\n",
       "      <td>0.505</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>IND @ MIL</td>\n",
       "      <td>L</td>\n",
       "      <td>13</td>\n",
       "      <td>0022000329</td>\n",
       "      <td>1610612754</td>\n",
       "      <td>Pacers</td>\n",
       "      <td>IND</td>\n",
       "      <td>Indiana</td>\n",
       "      <td>240:00</td>\n",
       "      <td>102.2</td>\n",
       "      <td>...</td>\n",
       "      <td>13.0</td>\n",
       "      <td>0.511</td>\n",
       "      <td>0.541</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.201</td>\n",
       "      <td>108.88</td>\n",
       "      <td>107.5</td>\n",
       "      <td>89.58</td>\n",
       "      <td>108</td>\n",
       "      <td>0.424</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>HOU @ NOP</td>\n",
       "      <td>L</td>\n",
       "      <td>23</td>\n",
       "      <td>0022000377</td>\n",
       "      <td>1610612745</td>\n",
       "      <td>Rockets</td>\n",
       "      <td>HOU</td>\n",
       "      <td>Houston</td>\n",
       "      <td>240:00</td>\n",
       "      <td>100.8</td>\n",
       "      <td>...</td>\n",
       "      <td>12.4</td>\n",
       "      <td>0.489</td>\n",
       "      <td>0.504</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.198</td>\n",
       "      <td>100.78</td>\n",
       "      <td>98.0</td>\n",
       "      <td>81.67</td>\n",
       "      <td>97</td>\n",
       "      <td>0.324</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>POR vs. PHI</td>\n",
       "      <td>W</td>\n",
       "      <td>21</td>\n",
       "      <td>0022000394</td>\n",
       "      <td>1610612757</td>\n",
       "      <td>Trail Blazers</td>\n",
       "      <td>POR</td>\n",
       "      <td>Portland</td>\n",
       "      <td>240:00</td>\n",
       "      <td>122.8</td>\n",
       "      <td>...</td>\n",
       "      <td>8.6</td>\n",
       "      <td>0.522</td>\n",
       "      <td>0.572</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.200</td>\n",
       "      <td>96.44</td>\n",
       "      <td>94.0</td>\n",
       "      <td>78.33</td>\n",
       "      <td>93</td>\n",
       "      <td>0.474</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 32 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       MATCHUP WL  PF     GAME_ID     TEAM_ID      TEAM_NAME  \\\n",
       "0    TOR @ SAS  L  23  0022000027  1610612761        Raptors   \n",
       "1    ORL @ MIN  W  17  0022000221  1610612753          Magic   \n",
       "2    IND @ MIL  L  13  0022000329  1610612754         Pacers   \n",
       "3    HOU @ NOP  L  23  0022000377  1610612745        Rockets   \n",
       "4  POR vs. PHI  W  21  0022000394  1610612757  Trail Blazers   \n",
       "\n",
       "  TEAM_ABBREVIATION TEAM_CITY     MIN  E_OFF_RATING  ...  TM_TOV_PCT  EFG_PCT  \\\n",
       "0               TOR   Toronto  240:00         109.0  ...        11.5    0.536   \n",
       "1               ORL   Orlando  240:00          96.8  ...        12.5    0.446   \n",
       "2               IND   Indiana  240:00         102.2  ...        13.0    0.511   \n",
       "3               HOU   Houston  240:00         100.8  ...        12.4    0.489   \n",
       "4               POR  Portland  240:00         122.8  ...         8.6    0.522   \n",
       "\n",
       "   TS_PCT  USG_PCT  E_USG_PCT  E_PACE   PACE  PACE_PER40  POSS    PIE  \n",
       "0   0.556      1.0      0.199  106.82  104.0       86.67   104  0.491  \n",
       "1   0.479      1.0      0.198   99.70   95.5       79.58    96  0.505  \n",
       "2   0.541      1.0      0.201  108.88  107.5       89.58   108  0.424  \n",
       "3   0.504      1.0      0.198  100.78   98.0       81.67    97  0.324  \n",
       "4   0.572      1.0      0.200   96.44   94.0       78.33    93  0.474  \n",
       "\n",
       "[5 rows x 32 columns]"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.toPandas().head(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Split Home and Away"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "home_df = df.filter(df.MATCHUP.contains('.'))\n",
    "away_df = df.filter(df.MATCHUP.contains('@'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Create HomeWin column comprised of binary response: 1 for home team win, 0 for home team loss."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "home_df = home_df.withColumn('HomeWin', f.when(f.col('WL') == 'W', 1).otherwise(0))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "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>WL</th>\n",
       "      <th>HomeWin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>W</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>L</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>W</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  WL  HomeWin\n",
       "0  W        1\n",
       "1  L        0\n",
       "2  W        1"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "home_df.toPandas()[['WL', 'HomeWin']].head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Prepare columns for `home_` and `away_` prefixes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "away_drop = ['MATCHUP', 'WL', 'GAME_ID', 'TEAM_CITY', 'MIN', 'USG_PCT']\n",
    "home_drop = away_drop + ['HomeWin']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "away_prep = away_df.drop(*away_drop)\n",
    "home_prep = home_df.drop(*home_drop)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "away_col_names = [f\"away_{i}\" for i in away_prep.columns]\n",
    "home_col_names = [f\"home_{i}\" for i in home_prep.columns]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "away_prep_df = away_prep.toDF(*away_col_names)\n",
    "home_prep_df = home_prep.toDF(*home_col_names)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Cannot concatenate horizonatlly in PySpark. So, we'll convert some DFs to pandas, concatenate, and convert back to Spark DF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "final_away = spark.createDataFrame(pd.concat([away_df.toPandas()['GAME_ID'], away_prep_df.toPandas()], axis=1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "final_home = spark.createDataFrame(pd.concat([home_df.toPandas()[['GAME_ID', 'HomeWin']], home_prep_df.toPandas()], axis=1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "# This join method avoids duplicating the 'GAME_ID' column.\n",
    "final_df = final_home.join(final_away, 'GAME_ID')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "root\n",
      " |-- GAME_ID: string (nullable = true)\n",
      " |-- HomeWin: long (nullable = true)\n",
      " |-- home_PF: long (nullable = true)\n",
      " |-- home_TEAM_ID: long (nullable = true)\n",
      " |-- home_TEAM_NAME: string (nullable = true)\n",
      " |-- home_TEAM_ABBREVIATION: string (nullable = true)\n",
      " |-- home_E_OFF_RATING: double (nullable = true)\n",
      " |-- home_OFF_RATING: double (nullable = true)\n",
      " |-- home_E_DEF_RATING: double (nullable = true)\n",
      " |-- home_DEF_RATING: double (nullable = true)\n",
      " |-- home_E_NET_RATING: double (nullable = true)\n",
      " |-- home_NET_RATING: double (nullable = true)\n",
      " |-- home_AST_PCT: double (nullable = true)\n",
      " |-- home_AST_TOV: double (nullable = true)\n",
      " |-- home_AST_RATIO: double (nullable = true)\n",
      " |-- home_OREB_PCT: double (nullable = true)\n",
      " |-- home_DREB_PCT: double (nullable = true)\n",
      " |-- home_REB_PCT: double (nullable = true)\n",
      " |-- home_E_TM_TOV_PCT: double (nullable = true)\n",
      " |-- home_TM_TOV_PCT: double (nullable = true)\n",
      " |-- home_EFG_PCT: double (nullable = true)\n",
      " |-- home_TS_PCT: double (nullable = true)\n",
      " |-- home_E_USG_PCT: double (nullable = true)\n",
      " |-- home_E_PACE: double (nullable = true)\n",
      " |-- home_PACE: double (nullable = true)\n",
      " |-- home_PACE_PER40: double (nullable = true)\n",
      " |-- home_POSS: long (nullable = true)\n",
      " |-- home_PIE: double (nullable = true)\n",
      " |-- away_PF: long (nullable = true)\n",
      " |-- away_TEAM_ID: long (nullable = true)\n",
      " |-- away_TEAM_NAME: string (nullable = true)\n",
      " |-- away_TEAM_ABBREVIATION: string (nullable = true)\n",
      " |-- away_E_OFF_RATING: double (nullable = true)\n",
      " |-- away_OFF_RATING: double (nullable = true)\n",
      " |-- away_E_DEF_RATING: double (nullable = true)\n",
      " |-- away_DEF_RATING: double (nullable = true)\n",
      " |-- away_E_NET_RATING: double (nullable = true)\n",
      " |-- away_NET_RATING: double (nullable = true)\n",
      " |-- away_AST_PCT: double (nullable = true)\n",
      " |-- away_AST_TOV: double (nullable = true)\n",
      " |-- away_AST_RATIO: double (nullable = true)\n",
      " |-- away_OREB_PCT: double (nullable = true)\n",
      " |-- away_DREB_PCT: double (nullable = true)\n",
      " |-- away_REB_PCT: double (nullable = true)\n",
      " |-- away_E_TM_TOV_PCT: double (nullable = true)\n",
      " |-- away_TM_TOV_PCT: double (nullable = true)\n",
      " |-- away_EFG_PCT: double (nullable = true)\n",
      " |-- away_TS_PCT: double (nullable = true)\n",
      " |-- away_E_USG_PCT: double (nullable = true)\n",
      " |-- away_E_PACE: double (nullable = true)\n",
      " |-- away_PACE: double (nullable = true)\n",
      " |-- away_PACE_PER40: double (nullable = true)\n",
      " |-- away_POSS: long (nullable = true)\n",
      " |-- away_PIE: double (nullable = true)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "final_df.printSchema()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Remove unnecessary columns and columns that provide too much certainty."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "no_e_cols = ['GAME_ID', 'home_TEAM_ID', 'home_TEAM_NAME', 'home_TEAM_ABBREVIATION', 'home_OFF_RATING', 'home_DEF_RATING', \n",
    "             'home_NET_RATING', 'away_TEAM_ID', 'away_TEAM_NAME', 'away_TEAM_ABBREVIATION', 'away_OFF_RATING', 'away_DEF_RATING', \n",
    "             'away_NET_RATING','home_E_OFF_RATING', 'home_E_DEF_RATING', 'home_E_NET_RATING', 'home_E_TOV_PCT', 'home_E_USG_PCT', \n",
    "             'home_E_PACE', 'home_E_TM_TOV_PCT', 'away_E_OFF_RATING', 'away_E_DEF_RATING', 'away_E_NET_RATING', 'away_E_TOV_PCT', \n",
    "             'away_E_USG_PCT', 'away_E_PACE', 'away_E_TM_TOV_PCT', 'home_EFG_PCT', 'away_EFG_PCT', 'home_PIE', 'away_PIE']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "no_e_cols = [i for i in final_df.columns if i not in no_e_cols]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "no_e_df = final_df[no_e_cols]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x1008 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize = (14,14))\n",
    "sns.heatmap(no_e_df.toPandas().corr())\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Correlation of HomeWin = 1.0\n",
      "Correlation of home_PF = -0.008284735439198762\n",
      "Correlation of home_AST_PCT = 0.04512517473790482\n",
      "Correlation of home_AST_TOV = 0.15493191902642559\n",
      "Correlation of home_AST_RATIO = 0.2908741385447218\n",
      "Correlation of home_OREB_PCT = 0.13985806971408493\n",
      "Correlation of home_DREB_PCT = 0.14934096292746155\n",
      "Correlation of home_REB_PCT = 0.43301543883520854\n",
      "Correlation of home_TM_TOV_PCT = -0.05905952204330835\n",
      "Correlation of home_TS_PCT = 0.5251072389555801\n",
      "Correlation of home_PACE = 0.016185270832408595\n",
      "Correlation of home_PACE_PER40 = 0.01617025568799348\n",
      "Correlation of home_POSS = 0.010319361211448828\n",
      "Correlation of away_PF = 0.053666713047682495\n",
      "Correlation of away_AST_PCT = -0.05971826802708149\n",
      "Correlation of away_AST_TOV = -0.1903854079505966\n",
      "Correlation of away_AST_RATIO = -0.3089610461479396\n",
      "Correlation of away_OREB_PCT = -0.14947385344000336\n",
      "Correlation of away_DREB_PCT = -0.13985276479550438\n",
      "Correlation of away_REB_PCT = -0.4330060994760807\n",
      "Correlation of away_TM_TOV_PCT = 0.038946538508922185\n",
      "Correlation of away_TS_PCT = -0.4811062918086309\n",
      "Correlation of away_PACE = 0.016185270832408595\n",
      "Correlation of away_PACE_PER40 = 0.01617025568799348\n",
      "Correlation of away_POSS = -0.008246459046394717\n"
     ]
    }
   ],
   "source": [
    "corr_df = no_e_df.toPandas().corr()\n",
    "j=0\n",
    "for i in corr_df.columns:\n",
    "    print(f\"Correlation of {i} = {corr_df.iloc[0][j]}\")\n",
    "    j+=1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### We'll drop some of these columns that have very extremely low correlations."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "low_corr=[]\n",
    "for i in range(len(corr_df.columns)):\n",
    "    if abs(corr_df.iloc[0][i]) < .1:\n",
    "        low_corr.append(corr_df.columns[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "no_e_df = no_e_df.drop(*low_corr)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Prepare data with VectorAssembler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pyspark.ml.feature import VectorAssembler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "inputCols = no_e_df.toPandas().drop('HomeWin', axis=1).columns.tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "va = VectorAssembler(inputCols = inputCols, outputCol = 'features')\n",
    "va_df = va.transform(no_e_df)\n",
    "va_df = va_df.select(['features', 'HomeWin'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Standardize Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pyspark.ml.feature import StandardScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_test_split = va_df.randomSplit([0.8, 0.2], seed=76)\n",
    "train_df = train_test_split[0]\n",
    "test_df = train_test_split[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "scaler = StandardScaler(inputCol='features', outputCol='scaledFeats')\n",
    "train_scaled = scaler.fit(train_df).transform(train_df)\n",
    "test_scaled = scaler.fit(train_df).transform(test_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Modeling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pyspark.ml.classification import LogisticRegression\n",
    "from pyspark.ml.evaluation import BinaryClassificationEvaluator"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "lr = LogisticRegression(featuresCol='scaledFeats', labelCol='HomeWin', maxIter=10, regParam=0.3, elasticNetParam=0.8)\n",
    "lr_model = lr.fit(train_scaled)\n",
    "lr_predictions = lr_model.transform(test_scaled)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9066784175479826"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lr_evaluator = BinaryClassificationEvaluator(labelCol='HomeWin')\n",
    "lr_evaluator.evaluate(lr_predictions)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Get a dataframe of last 5 games advanced stats"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "# DF for competive window before playoff positioning locks\n",
    "# team_adv = leaguedashteamstats.LeagueDashTeamStats(measure_type_detailed_defense='Advanced',\n",
    "#                                                    date_from_nullable='05/06/2021', date_to_nullable='05/14/2021').get_data_frames()[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "# DF of last 5\n",
    "team_adv = leaguedashteamstats.LeagueDashTeamStats(last_n_games=5, measure_type_detailed_defense='Advanced').get_data_frames()[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Build some functoins to avoid a further mess of local variables than we already have"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "def nickname_to_id(nickname=''):\n",
    "    return team_df[team_df['nickname'] == nickname].iloc[0][0]\n",
    "\n",
    "def gen_data(nickname=''):\n",
    "    return team_adv[team_adv['TEAM_ID'] == nickname_to_id(nickname)]\n",
    "\n",
    "def away_cols(nickname=''):\n",
    "    new_df = gen_data(nickname)\n",
    "    new_df = new_df.rename(columns={'AST_TO': 'AST_TOV'})\n",
    "    new_df = new_df.add_prefix('away_')\n",
    "    return new_df\n",
    "    \n",
    "def home_cols(nickname=''):\n",
    "    new_df = gen_data(nickname)\n",
    "    new_df = new_df.rename(columns={'AST_TO': 'AST_TOV'})\n",
    "    new_df = new_df.add_prefix('home_')\n",
    "    return new_df\n",
    "\n",
    "def matchup(home='', away=''):\n",
    "    home_team = home_cols(home)\n",
    "    away_team = away_cols(away)\n",
    "    home_data = [i for i in no_e_df.columns if i in home_cols(home).columns]\n",
    "    away_data = [i for i in no_e_df.columns if i in away_cols(away).columns]\n",
    "    final_home = home_team[home_data].reset_index(drop=True)\n",
    "    final_away = away_team[away_data].reset_index(drop=True)\n",
    "    final_line = pd.concat([final_home, final_away], axis=1)\n",
    "    return final_line\n",
    "\n",
    "def winner(home='', away=''):\n",
    "    match_df = matchup(home,away)\n",
    "    spark_match = spark.createDataFrame(match_df)\n",
    "    va = VectorAssembler(inputCols = spark_match.columns, outputCol = 'features')\n",
    "    va_df = va.transform(spark_match)\n",
    "    game = va_df.select('features')\n",
    "    return game"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Scale the data for the mathcup according to our training data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 05/16/21 - Predict Trail Blazers vs. Nuggets 9:00PM Tip-Off, 12:45PM Prediction\n",
    "* 1.0 indicates model predicts Home team win"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+----------+\n",
      "|prediction|\n",
      "+----------+\n",
      "|       1.0|\n",
      "+----------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "lr_model.transform(scaler.fit(train_df).transform(winner('Trail Blazers', 'Nuggets'))).select('prediction').show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Model predicts Trail Blazers to win."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 05/16/21 - Predict Timberwolves vs. Mavericks 9:00PM Tip-Off, 12:45PM Prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+----------+\n",
      "|prediction|\n",
      "+----------+\n",
      "|       1.0|\n",
      "+----------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "lr_model.transform(scaler.fit(train_df).transform(winner('Timberwolves', 'Mavericks'))).select('prediction').show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Model predicts Timberwolves to win."
   ]
  }
 ],
 "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": "3.8.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}