model_training.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Model Training for Street Figther 6 Bracket Predictor\n",
    "This is the notebook demonstrating how to create the actual prediction model for the Street Figther 6 Bracket Predictor Model. For more on how to create the initial data, see the README for this project."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 1: Data Processing\n",
    "Before we can feed the data into the prediction model, we need to take our set data, which has winners and losers in their own rows, and join the data onto itself so that all relevant data for a given set are in the same row. We also need to join the player data in the form of ELO ratings and tournament participation data onto this set data too so the model has a better idea of each player's respective skill level."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import initial packages for data manipulation\n",
    "import pandas as pd\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 168,
   "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>set_id</th>\n",
       "      <th>entrant_id</th>\n",
       "      <th>entrant_name</th>\n",
       "      <th>standing</th>\n",
       "      <th>user_id</th>\n",
       "      <th>event_id</th>\n",
       "      <th>player_id</th>\n",
       "      <th>gamerTag</th>\n",
       "      <th>player_prefix</th>\n",
       "      <th>source</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>74141170.0</td>\n",
       "      <td>16483462</td>\n",
       "      <td>ADTerminal</td>\n",
       "      <td>2</td>\n",
       "      <td>221788.0</td>\n",
       "      <td>1129069</td>\n",
       "      <td>324331.0</td>\n",
       "      <td>ADTerminal</td>\n",
       "      <td>NaN</td>\n",
       "      <td>startgg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>74141170.0</td>\n",
       "      <td>16485711</td>\n",
       "      <td>Lyser</td>\n",
       "      <td>1</td>\n",
       "      <td>2568953.0</td>\n",
       "      <td>1129069</td>\n",
       "      <td>3998295.0</td>\n",
       "      <td>Lyser</td>\n",
       "      <td>NaN</td>\n",
       "      <td>startgg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>74141171.0</td>\n",
       "      <td>16485711</td>\n",
       "      <td>Lyser</td>\n",
       "      <td>1</td>\n",
       "      <td>2568953.0</td>\n",
       "      <td>1129069</td>\n",
       "      <td>3998295.0</td>\n",
       "      <td>Lyser</td>\n",
       "      <td>NaN</td>\n",
       "      <td>startgg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>74141171.0</td>\n",
       "      <td>16483462</td>\n",
       "      <td>ADTerminal</td>\n",
       "      <td>2</td>\n",
       "      <td>221788.0</td>\n",
       "      <td>1129069</td>\n",
       "      <td>324331.0</td>\n",
       "      <td>ADTerminal</td>\n",
       "      <td>NaN</td>\n",
       "      <td>startgg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>74141169.0</td>\n",
       "      <td>16483462</td>\n",
       "      <td>ADTerminal</td>\n",
       "      <td>1</td>\n",
       "      <td>221788.0</td>\n",
       "      <td>1129069</td>\n",
       "      <td>324331.0</td>\n",
       "      <td>ADTerminal</td>\n",
       "      <td>NaN</td>\n",
       "      <td>startgg</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       set_id  entrant_id entrant_name  standing    user_id  event_id  \\\n",
       "0  74141170.0    16483462   ADTerminal         2   221788.0   1129069   \n",
       "1  74141170.0    16485711        Lyser         1  2568953.0   1129069   \n",
       "2  74141171.0    16485711        Lyser         1  2568953.0   1129069   \n",
       "3  74141171.0    16483462   ADTerminal         2   221788.0   1129069   \n",
       "4  74141169.0    16483462   ADTerminal         1   221788.0   1129069   \n",
       "\n",
       "   player_id    gamerTag player_prefix   source  \n",
       "0   324331.0  ADTerminal           NaN  startgg  \n",
       "1  3998295.0       Lyser           NaN  startgg  \n",
       "2  3998295.0       Lyser           NaN  startgg  \n",
       "3   324331.0  ADTerminal           NaN  startgg  \n",
       "4   324331.0  ADTerminal           NaN  startgg  "
      ]
     },
     "execution_count": 168,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Import set data. Note that 1 signifies a win and 2 signifies a loss.\n",
    "df = pd.read_csv('data\\\\all_sets.csv')\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "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>user_id</th>\n",
       "      <th>event_id</th>\n",
       "      <th>elo</th>\n",
       "      <th>tier1</th>\n",
       "      <th>tier2</th>\n",
       "      <th>tier3</th>\n",
       "      <th>tier5</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>59134.0</td>\n",
       "      <td>372265.0</td>\n",
       "      <td>143.211697</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>59135.0</td>\n",
       "      <td>372265.0</td>\n",
       "      <td>212.985028</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>59136.0</td>\n",
       "      <td>372265.0</td>\n",
       "      <td>202.474206</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>59137.0</td>\n",
       "      <td>372265.0</td>\n",
       "      <td>240.640067</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>59138.0</td>\n",
       "      <td>372265.0</td>\n",
       "      <td>200.689002</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   user_id  event_id         elo  tier1  tier2  tier3  tier5\n",
       "0  59134.0  372265.0  143.211697    0.0    0.0    0.0    1.0\n",
       "1  59135.0  372265.0  212.985028    0.0    0.0    0.0    1.0\n",
       "2  59136.0  372265.0  202.474206    0.0    0.0    0.0    1.0\n",
       "3  59137.0  372265.0  240.640067    0.0    0.0    0.0    1.0\n",
       "4  59138.0  372265.0  200.689002    0.0    0.0    0.0    1.0"
      ]
     },
     "execution_count": 169,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Import ELO data.\n",
    "elo = pd.read_csv('data\\\\elo_records.csv')\n",
    "elo.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### A note on tournament tiers\n",
    "An issue that was encountered in the early stages of this project was that ELO was not adequate to determine the true skill level of each player. It's suspected this was partially due to both a lack of data where not enough match result data could differentiate the top players against more average players. Including participation rates in higher level tournaments was thought to mediate this issue. \n",
    "\n",
    "Each \"tier\" column refers to the competition tier of each tournament a player is known to have participated in. These definitions are taken from the Liquipedia and are quoted from their [website](https://liquipedia.net/fighters/Street_Fighter_6/Tier_1_Tournaments) as follows:\n",
    "* \"**Tier 1 Tournaments**, sometimes called premier tournaments or super-majors, offer high prize pools and feature the best players from all over the world.\"\n",
    "* \"**Tier 2 Tournaments**, often called major tournaments or just majors, feature some top-tier players and offer good prize pools or seeding slots for premier events. They don't have the prestige or prize pools of a premier event, but they are usually notable enough to attract international attention.\"\n",
    "* \"**Tier 3 Tournaments**, often called minor tournaments or just minors, offer a smaller prize pool and less prestige than Tier 2 Tournaments, but still feature a significant level of competition. Smaller regional tournaments tend to fall into this category. \"\n",
    "\n",
    "However, to account for the amount of smaller, less formal events (such as local meetups) recorded from start.gg, a **Tier 5** category has been included to refer to these events."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Days Between Events\n",
    "I've also opted to include the days between each event for each player. The thought process behind including this variable is that players who participate more often in tournaments are more likely to win since they are playing more which is an indicator of experience."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 170,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "event_id                float64\n",
      "start_at    datetime64[ns, UTC]\n",
      "game                     object\n",
      "dtype: object\n"
     ]
    },
    {
     "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>user_id</th>\n",
       "      <th>event_id</th>\n",
       "      <th>elo</th>\n",
       "      <th>tier1</th>\n",
       "      <th>tier2</th>\n",
       "      <th>tier3</th>\n",
       "      <th>tier5</th>\n",
       "      <th>start_at</th>\n",
       "      <th>days_between_events</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>44328</th>\n",
       "      <td>0.0</td>\n",
       "      <td>867114.0</td>\n",
       "      <td>234.063434</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2023-08-18 09:00:00+00:00</td>\n",
       "      <td>999.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>174194</th>\n",
       "      <td>0.0</td>\n",
       "      <td>1088942.0</td>\n",
       "      <td>253.703530</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2024-03-10 20:00:00+00:00</td>\n",
       "      <td>999.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>192719</th>\n",
       "      <td>0.0</td>\n",
       "      <td>1118565.0</td>\n",
       "      <td>306.569468</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2024-04-17 01:45:00+00:00</td>\n",
       "      <td>999.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30569</th>\n",
       "      <td>1.0</td>\n",
       "      <td>864717.0</td>\n",
       "      <td>187.451839</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2023-08-04 15:00:00+00:00</td>\n",
       "      <td>999.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49950</th>\n",
       "      <td>1.0</td>\n",
       "      <td>963000.0</td>\n",
       "      <td>179.137104</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2023-08-20 21:00:00+00:00</td>\n",
       "      <td>999.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        user_id   event_id         elo  tier1  tier2  tier3  tier5  \\\n",
       "44328       0.0   867114.0  234.063434    1.0    0.0    0.0    0.0   \n",
       "174194      0.0  1088942.0  253.703530    1.0    0.0    0.0    1.0   \n",
       "192719      0.0  1118565.0  306.569468    1.0    0.0    0.0    2.0   \n",
       "30569       1.0   864717.0  187.451839    1.0    0.0    0.0    0.0   \n",
       "49950       1.0   963000.0  179.137104    1.0    0.0    0.0    1.0   \n",
       "\n",
       "                        start_at  days_between_events  \n",
       "44328  2023-08-18 09:00:00+00:00                999.0  \n",
       "174194 2024-03-10 20:00:00+00:00                999.0  \n",
       "192719 2024-04-17 01:45:00+00:00                999.0  \n",
       "30569  2023-08-04 15:00:00+00:00                999.0  \n",
       "49950  2023-08-20 21:00:00+00:00                999.0  "
      ]
     },
     "execution_count": 170,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Import events data\n",
    "events = pd.read_csv('data\\\\events.csv')[['event_id', 'start_at', 'game']]\n",
    "events['start_at'] = pd.to_datetime(events['start_at'], format='ISO8601')\n",
    "print(events.dtypes)\n",
    "events = events.loc[events['game'] == 'SF6', :].drop(columns=['game'])\n",
    "\n",
    "# Merge event data on ELO to calculate days between events for each player\n",
    "elo2 = elo.merge(events, on='event_id').sort_values(by=['user_id', 'start_at'],\n",
    "                                                    ascending= [True, True])\n",
    "elo2['days_between_events'] = elo2.groupby(['user_id', 'event_id'])['start_at'].diff().dt.days\n",
    "\n",
    "# Default value for days between events is 999 days as filler value\n",
    "elo2['days_between_events'] = elo2['days_between_events'].fillna(999)\n",
    "elo2.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 171,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\sarah\\AppData\\Local\\Temp\\ipykernel_31220\\3337745523.py:1: DtypeWarning: Columns (15) have mixed types. Specify dtype option on import or set low_memory=False.\n",
      "  players = pd.read_csv('data\\\\players.csv')\n"
     ]
    },
    {
     "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>user_id</th>\n",
       "      <th>event_id</th>\n",
       "      <th>elo</th>\n",
       "      <th>tier1</th>\n",
       "      <th>tier2</th>\n",
       "      <th>tier3</th>\n",
       "      <th>tier5</th>\n",
       "      <th>start_at</th>\n",
       "      <th>days_between_events</th>\n",
       "      <th>startgg_pid</th>\n",
       "      <th>liquidpedia_name</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0</td>\n",
       "      <td>867114.0</td>\n",
       "      <td>234.063434</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2023-08-18 09:00:00+00:00</td>\n",
       "      <td>999.0</td>\n",
       "      <td>324331.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.0</td>\n",
       "      <td>1088942.0</td>\n",
       "      <td>253.703530</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2024-03-10 20:00:00+00:00</td>\n",
       "      <td>999.0</td>\n",
       "      <td>324331.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0</td>\n",
       "      <td>1118565.0</td>\n",
       "      <td>306.569468</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2024-04-17 01:45:00+00:00</td>\n",
       "      <td>999.0</td>\n",
       "      <td>324331.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1.0</td>\n",
       "      <td>864717.0</td>\n",
       "      <td>187.451839</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2023-08-04 15:00:00+00:00</td>\n",
       "      <td>999.0</td>\n",
       "      <td>985993.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.0</td>\n",
       "      <td>963000.0</td>\n",
       "      <td>179.137104</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2023-08-20 21:00:00+00:00</td>\n",
       "      <td>999.0</td>\n",
       "      <td>985993.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   user_id   event_id         elo  tier1  tier2  tier3  tier5  \\\n",
       "0      0.0   867114.0  234.063434    1.0    0.0    0.0    0.0   \n",
       "1      0.0  1088942.0  253.703530    1.0    0.0    0.0    1.0   \n",
       "2      0.0  1118565.0  306.569468    1.0    0.0    0.0    2.0   \n",
       "3      1.0   864717.0  187.451839    1.0    0.0    0.0    0.0   \n",
       "4      1.0   963000.0  179.137104    1.0    0.0    0.0    1.0   \n",
       "\n",
       "                   start_at  days_between_events  startgg_pid liquidpedia_name  \n",
       "0 2023-08-18 09:00:00+00:00                999.0     324331.0              NaN  \n",
       "1 2024-03-10 20:00:00+00:00                999.0     324331.0              NaN  \n",
       "2 2024-04-17 01:45:00+00:00                999.0     324331.0              NaN  \n",
       "3 2023-08-04 15:00:00+00:00                999.0     985993.0              NaN  \n",
       "4 2023-08-20 21:00:00+00:00                999.0     985993.0              NaN  "
      ]
     },
     "execution_count": 171,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Import player data; This is neccesary to graft the elo data to event data\n",
    "players = pd.read_csv('data\\\\players.csv')\n",
    "players = players[['startgg_pid', 'uid', 'liquidpedia_name']]\n",
    "\n",
    "# Merge data and drop UID; including the specific UID as a category is highly likely to cause overfitting\n",
    "elo3 = elo2.merge(players, how='left', left_on=['user_id'], right_on=['uid'])\n",
    "elo3 = elo3.drop(columns=['uid'])\n",
    "elo3.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "After getting elo data for each event and event tier counts, it's important to shift the data down by each `user_id` group and include the default values for the first rows in each group. (We don't want to use the ELO calculated from data for that specific event since we shouldn't know that score yet!)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 172,
   "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>user_id</th>\n",
       "      <th>event_id</th>\n",
       "      <th>elo</th>\n",
       "      <th>tier1</th>\n",
       "      <th>tier2</th>\n",
       "      <th>tier3</th>\n",
       "      <th>tier5</th>\n",
       "      <th>start_at</th>\n",
       "      <th>days_between_events</th>\n",
       "      <th>startgg_pid</th>\n",
       "      <th>liquidpedia_name</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0</td>\n",
       "      <td>867114.0</td>\n",
       "      <td>200.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2023-08-18 09:00:00+00:00</td>\n",
       "      <td>999.0</td>\n",
       "      <td>324331.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.0</td>\n",
       "      <td>1088942.0</td>\n",
       "      <td>234.063434</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2024-03-10 20:00:00+00:00</td>\n",
       "      <td>999.0</td>\n",
       "      <td>324331.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0</td>\n",
       "      <td>1118565.0</td>\n",
       "      <td>253.703530</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2024-04-17 01:45:00+00:00</td>\n",
       "      <td>999.0</td>\n",
       "      <td>324331.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1.0</td>\n",
       "      <td>864717.0</td>\n",
       "      <td>200.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2023-08-04 15:00:00+00:00</td>\n",
       "      <td>999.0</td>\n",
       "      <td>985993.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.0</td>\n",
       "      <td>963000.0</td>\n",
       "      <td>187.451839</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2023-08-20 21:00:00+00:00</td>\n",
       "      <td>999.0</td>\n",
       "      <td>985993.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   user_id   event_id         elo  tier1  tier2  tier3  tier5  \\\n",
       "0      0.0   867114.0  200.000000    0.0    0.0    0.0    0.0   \n",
       "1      0.0  1088942.0  234.063434    1.0    0.0    0.0    0.0   \n",
       "2      0.0  1118565.0  253.703530    1.0    0.0    0.0    1.0   \n",
       "3      1.0   864717.0  200.000000    0.0    0.0    0.0    0.0   \n",
       "4      1.0   963000.0  187.451839    1.0    0.0    0.0    0.0   \n",
       "\n",
       "                   start_at  days_between_events  startgg_pid liquidpedia_name  \n",
       "0 2023-08-18 09:00:00+00:00                999.0     324331.0              NaN  \n",
       "1 2024-03-10 20:00:00+00:00                999.0     324331.0              NaN  \n",
       "2 2024-04-17 01:45:00+00:00                999.0     324331.0              NaN  \n",
       "3 2023-08-04 15:00:00+00:00                999.0     985993.0              NaN  \n",
       "4 2023-08-20 21:00:00+00:00                999.0     985993.0              NaN  "
      ]
     },
     "execution_count": 172,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "default_elo = 200\n",
    "elo3['elo'] = elo3.groupby('user_id')['elo'].shift(1).fillna(default_elo)\n",
    "for tier in ['tier1', 'tier2', 'tier3', 'tier5']:\n",
    "    elo3[tier] = elo3.groupby('user_id')[tier].shift(1).fillna(0)\n",
    "elo3.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we're ready to merge all of the data onto itself so that we end up with the data for each player and their set result."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 173,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\sarah\\AppData\\Local\\Temp\\ipykernel_31220\\1336078828.py:9: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  df5['match_name'] = df['entrant_name'].str.lower().str.replace(' ','', regex=True)\n"
     ]
    },
    {
     "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>set_id</th>\n",
       "      <th>entrant_id_x</th>\n",
       "      <th>entrant_name_x</th>\n",
       "      <th>standing_x</th>\n",
       "      <th>user_id_x</th>\n",
       "      <th>event_id</th>\n",
       "      <th>player_id_x</th>\n",
       "      <th>gamerTag_x</th>\n",
       "      <th>player_prefix_x</th>\n",
       "      <th>source_x</th>\n",
       "      <th>...</th>\n",
       "      <th>tier1_y</th>\n",
       "      <th>tier2_y</th>\n",
       "      <th>tier3_y</th>\n",
       "      <th>tier5_y</th>\n",
       "      <th>start_at_y</th>\n",
       "      <th>days_between_events_y</th>\n",
       "      <th>startgg_pid</th>\n",
       "      <th>liquidpedia_name</th>\n",
       "      <th>event_id_x</th>\n",
       "      <th>event_id_y</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>73485780.0</td>\n",
       "      <td>16337647.0</td>\n",
       "      <td>AlukardNY</td>\n",
       "      <td>2.0</td>\n",
       "      <td>196.0</td>\n",
       "      <td>1117616.0</td>\n",
       "      <td>3484.0</td>\n",
       "      <td>AlukardNY</td>\n",
       "      <td>NaN</td>\n",
       "      <td>startgg</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaT</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>73485594.0</td>\n",
       "      <td>16337647.0</td>\n",
       "      <td>AlukardNY</td>\n",
       "      <td>2.0</td>\n",
       "      <td>196.0</td>\n",
       "      <td>1117616.0</td>\n",
       "      <td>3484.0</td>\n",
       "      <td>AlukardNY</td>\n",
       "      <td>NaN</td>\n",
       "      <td>startgg</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaT</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>2</th>\n",
       "      <td>72612176.0</td>\n",
       "      <td>16118157.0</td>\n",
       "      <td>AlukardNY</td>\n",
       "      <td>2.0</td>\n",
       "      <td>196.0</td>\n",
       "      <td>1102572.0</td>\n",
       "      <td>3484.0</td>\n",
       "      <td>AlukardNY</td>\n",
       "      <td>NaN</td>\n",
       "      <td>startgg</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaT</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>3</th>\n",
       "      <td>72612483.0</td>\n",
       "      <td>16118157.0</td>\n",
       "      <td>AlukardNY</td>\n",
       "      <td>2.0</td>\n",
       "      <td>196.0</td>\n",
       "      <td>1102572.0</td>\n",
       "      <td>3484.0</td>\n",
       "      <td>AlukardNY</td>\n",
       "      <td>NaN</td>\n",
       "      <td>startgg</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaT</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>4</th>\n",
       "      <td>72612162.0</td>\n",
       "      <td>16118157.0</td>\n",
       "      <td>AlukardNY</td>\n",
       "      <td>1.0</td>\n",
       "      <td>196.0</td>\n",
       "      <td>1102572.0</td>\n",
       "      <td>3484.0</td>\n",
       "      <td>AlukardNY</td>\n",
       "      <td>NaN</td>\n",
       "      <td>startgg</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaT</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 37 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       set_id  entrant_id_x entrant_name_x  standing_x  user_id_x   event_id  \\\n",
       "0  73485780.0    16337647.0      AlukardNY         2.0      196.0  1117616.0   \n",
       "1  73485594.0    16337647.0      AlukardNY         2.0      196.0  1117616.0   \n",
       "2  72612176.0    16118157.0      AlukardNY         2.0      196.0  1102572.0   \n",
       "3  72612483.0    16118157.0      AlukardNY         2.0      196.0  1102572.0   \n",
       "4  72612162.0    16118157.0      AlukardNY         1.0      196.0  1102572.0   \n",
       "\n",
       "   player_id_x gamerTag_x player_prefix_x source_x  ...  tier1_y  tier2_y  \\\n",
       "0       3484.0  AlukardNY             NaN  startgg  ...      0.0      0.0   \n",
       "1       3484.0  AlukardNY             NaN  startgg  ...      0.0      0.0   \n",
       "2       3484.0  AlukardNY             NaN  startgg  ...      0.0      0.0   \n",
       "3       3484.0  AlukardNY             NaN  startgg  ...      0.0      0.0   \n",
       "4       3484.0  AlukardNY             NaN  startgg  ...      0.0      0.0   \n",
       "\n",
       "   tier3_y  tier5_y  start_at_y days_between_events_y  startgg_pid  \\\n",
       "0      0.0      0.0         NaT                   NaN          NaN   \n",
       "1      0.0      0.0         NaT                   NaN          NaN   \n",
       "2      0.0      0.0         NaT                   NaN          NaN   \n",
       "3      0.0      0.0         NaT                   NaN          NaN   \n",
       "4      0.0      0.0         NaT                   NaN          NaN   \n",
       "\n",
       "   liquidpedia_name event_id_x  event_id_y  \n",
       "0               NaN        NaN         NaN  \n",
       "1               NaN        NaN         NaN  \n",
       "2               NaN        NaN         NaN  \n",
       "3               NaN        NaN         NaN  \n",
       "4               NaN        NaN         NaN  \n",
       "\n",
       "[5 rows x 37 columns]"
      ]
     },
     "execution_count": 173,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df3 = df.merge(elo2, on=['user_id', 'event_id'])\n",
    "df3 = df3.merge(df, left_on = ['set_id', 'event_id'], right_on=['set_id', 'event_id'])\n",
    "df3 = df3[(df3['standing_y'] != df3['standing_x'])].drop_duplicates(subset=['set_id', 'event_id'], keep='first')\n",
    "\n",
    "df4 = df3.loc[df3['event_id'] > 60, :]\n",
    "df4 = df4.merge(elo3, how='left', left_on=['user_id_y', 'event_id'], right_on=['startgg_pid', 'event_id'])\n",
    "\n",
    "df5 = df3.loc[df['event_id'] <= 60, :]\n",
    "df5['match_name'] = df['entrant_name'].str.lower().str.replace(' ','', regex=True)\n",
    "elo4 = elo3.copy()\n",
    "elo4['match_name'] = elo4['liquidpedia_name'].str.lower().str.replace(' ','', regex=True)\n",
    "df5 = df5.merge(elo4, how='left', on=['match_name'])\n",
    "df5 = df5.drop(columns=['match_name'])\n",
    "\n",
    "df3 = pd.concat([df4, df5], ignore_index=True)\n",
    "\n",
    "for val in ['elo_x', 'elo_y']:\n",
    "    df3[val] = df3[val].fillna(200)\n",
    "\n",
    "for val in ['tier1_x', 'tier1_y', 'tier2_x', 'tier2_y',\n",
    "            'tier3_x', 'tier3_y', 'tier5_x', 'tier5_y']:\n",
    "    df3[val] = df3[val].fillna(0)\n",
    "\n",
    "df3.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "At this point, we can filter out unneccesary columns and calculate a new column that includes all competitions as a potential column to use for our model training."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "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>set_id</th>\n",
       "      <th>event_id</th>\n",
       "      <th>elo_x</th>\n",
       "      <th>elo_y</th>\n",
       "      <th>tier1_x</th>\n",
       "      <th>tier1_y</th>\n",
       "      <th>tier2_x</th>\n",
       "      <th>tier2_y</th>\n",
       "      <th>tier3_x</th>\n",
       "      <th>tier3_y</th>\n",
       "      <th>tier5_x</th>\n",
       "      <th>tier5_y</th>\n",
       "      <th>days_between_events_x</th>\n",
       "      <th>days_between_events_y</th>\n",
       "      <th>result</th>\n",
       "      <th>all_comps_x</th>\n",
       "      <th>all_comps_y</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>73485780.0</td>\n",
       "      <td>1117616.0</td>\n",
       "      <td>422.432621</td>\n",
       "      <td>200.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>127.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>999.0</td>\n",
       "      <td>999.0</td>\n",
       "      <td>1</td>\n",
       "      <td>131.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>73485594.0</td>\n",
       "      <td>1117616.0</td>\n",
       "      <td>422.432621</td>\n",
       "      <td>200.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>127.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>999.0</td>\n",
       "      <td>999.0</td>\n",
       "      <td>1</td>\n",
       "      <td>131.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>72612176.0</td>\n",
       "      <td>1102572.0</td>\n",
       "      <td>394.198831</td>\n",
       "      <td>200.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>118.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>999.0</td>\n",
       "      <td>999.0</td>\n",
       "      <td>1</td>\n",
       "      <td>122.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>72612483.0</td>\n",
       "      <td>1102572.0</td>\n",
       "      <td>394.198831</td>\n",
       "      <td>200.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>118.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>999.0</td>\n",
       "      <td>999.0</td>\n",
       "      <td>1</td>\n",
       "      <td>122.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>72612162.0</td>\n",
       "      <td>1102572.0</td>\n",
       "      <td>394.198831</td>\n",
       "      <td>200.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>118.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>999.0</td>\n",
       "      <td>999.0</td>\n",
       "      <td>0</td>\n",
       "      <td>122.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       set_id   event_id       elo_x  elo_y  tier1_x  tier1_y  tier2_x  \\\n",
       "0  73485780.0  1117616.0  422.432621  200.0      0.0      0.0      1.0   \n",
       "1  73485594.0  1117616.0  422.432621  200.0      0.0      0.0      1.0   \n",
       "2  72612176.0  1102572.0  394.198831  200.0      0.0      0.0      1.0   \n",
       "3  72612483.0  1102572.0  394.198831  200.0      0.0      0.0      1.0   \n",
       "4  72612162.0  1102572.0  394.198831  200.0      0.0      0.0      1.0   \n",
       "\n",
       "   tier2_y  tier3_x  tier3_y  tier5_x  tier5_y  days_between_events_x  \\\n",
       "0      0.0      3.0      0.0    127.0      0.0                  999.0   \n",
       "1      0.0      3.0      0.0    127.0      0.0                  999.0   \n",
       "2      0.0      3.0      0.0    118.0      0.0                  999.0   \n",
       "3      0.0      3.0      0.0    118.0      0.0                  999.0   \n",
       "4      0.0      3.0      0.0    118.0      0.0                  999.0   \n",
       "\n",
       "   days_between_events_y  result  all_comps_x  all_comps_y  \n",
       "0                  999.0       1        131.0          0.0  \n",
       "1                  999.0       1        131.0          0.0  \n",
       "2                  999.0       1        122.0          0.0  \n",
       "3                  999.0       1        122.0          0.0  \n",
       "4                  999.0       0        122.0          0.0  "
      ]
     },
     "execution_count": 174,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "filter_cols = ['set_id','event_id', 'elo_x', 'elo_y', 'tier1_x', 'tier1_y', 'tier2_x', 'tier2_y',\n",
    "               'tier3_x', 'tier3_y', 'tier5_x', 'tier5_y', 'standing_x', 'days_between_events_x',\n",
    "               'days_between_events_y']\n",
    "\n",
    "df3 = df3[filter_cols]\n",
    "df3['result'] = 0\n",
    "df3.loc[df3['standing_x'] == 1, 'result'] = 0\n",
    "df3.loc[df3['standing_x'] == 2, 'result'] = 1\n",
    "\n",
    "df3['all_comps_x'] = df3['tier1_x'] + df3['tier2_x'] + df3['tier3_x'] + df3['tier5_x']\n",
    "df3['all_comps_y'] = df3['tier1_y'] + df3['tier2_y'] + df3['tier3_y'] + df3['tier5_y']\n",
    "\n",
    "df3['days_between_events_x'] = df3['days_between_events_x'].fillna(999)\n",
    "df3['days_between_events_y'] = df3['days_between_events_y'].fillna(999)\n",
    "\n",
    "df3.drop(columns=['standing_x'], inplace=True)\n",
    "df3.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Lastly, I also use a function to shuffle the order of winners and losers in the data since my initial data had a disproportionate amount of winners on the \"x\" side. This function splits the data more evenly."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "metadata": {},
   "outputs": [],
   "source": [
    "def shuffle_rows(df):\n",
    "    shuffle = ['set_id', 'event_id', 'elo_y', 'elo_x', 'tier1_y', 'tier1_x', \n",
    "                    'tier2_y', 'tier2_x', 'tier3_y', 'tier3_x', 'tier5_y', 'tier5_x',\n",
    "                    'days_between_events_y', 'days_between_events_x', 'result', 'all_comps_y', 'all_comps_x']\n",
    "\n",
    "    mask = np.random.rand(len(df)) < 0.5\n",
    "    df_shuffled = df.copy()\n",
    "    df_shuffled.loc[mask, :] = df_shuffled.loc[mask, shuffle].values\n",
    "    df_shuffled.loc[mask, 'result'] = 1 - df_shuffled.loc[mask, 'result'].values\n",
    "    return df_shuffled\n",
    "    \n",
    "df3 = shuffle_rows(df3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 2: Initial Model Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Linear Regression Model\n",
    "from sklearn.linear_model import LinearRegression\n",
    "model = LinearRegression()\n",
    "\n",
    "X = df3.copy()[['all_comps_x', 'all_comps_y', 'elo_x', 'elo_y', 'tier1_x', 'tier1_y',\n",
    "                'tier2_x', 'tier2_y', 'tier3_x', 'tier3_y', 'tier5_x', 'tier5_y',\n",
    "                'days_between_events_x', 'days_between_events_y']]\n",
    "y = df3['result']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2067742\n",
      "4824732\n"
     ]
    }
   ],
   "source": [
    "from sklearn.preprocessing import normalize\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "X_norm = normalize(X, norm='l1')\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X_norm, y.astype(int), shuffle = True, train_size=0.3, random_state=24)\n",
    "print(len(X_train))\n",
    "print(len(X_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 178,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "result\n",
       "0    3446355\n",
       "1    3446119\n",
       "Name: count, dtype: int64"
      ]
     },
     "execution_count": 178,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 179,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The r2 is:  -5.788201125067616e-07\n",
      "The rmse is:  0.5000001352959642\n"
     ]
    }
   ],
   "source": [
    "# Linear Regression Model\n",
    "from sklearn.linear_model import Lasso\n",
    "model = Lasso(alpha=0.1)\n",
    "\n",
    "model.fit(X_train, y_train)\n",
    "\n",
    "from sklearn.metrics import mean_squared_error, r2_score\n",
    "predictions = model.predict(X_test)\n",
    "r2 = r2_score(y_test, predictions)\n",
    "rmse = mean_squared_error(y_test, predictions, squared=False)\n",
    "\n",
    "print('The r2 is: ', r2)\n",
    "print('The rmse is: ', rmse)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ 5.43819000e-03 -5.46065513e-03  1.69912882e-02 -1.69594632e-02\n",
      "  6.22567252e-05 -6.21376220e-05  4.48792177e-05 -4.47707248e-05\n",
      "  1.47068064e-04 -1.47338672e-04  5.18398599e-03 -5.20640811e-03\n",
      "  5.32176230e-06  5.32176230e-06]\n",
      "-0.6937141431072299\n",
      "0.499720    1412846\n",
      "0.498641        759\n",
      "0.498542        754\n",
      "0.500415        751\n",
      "0.500878        750\n",
      "             ...   \n",
      "0.499172          1\n",
      "0.500315          1\n",
      "0.499608          1\n",
      "0.500501          1\n",
      "0.500686          1\n",
      "Name: count, Length: 294056, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "from sklearn.linear_model import TweedieRegressor\n",
    "reg = TweedieRegressor(power=1, alpha=0.5, link='log')\n",
    "reg.fit(X_train, y_train)\n",
    "\n",
    "print(reg.coef_)\n",
    "print(reg.intercept_)\n",
    "\n",
    "y_preds = reg.predict(X_test)\n",
    "print(pd.Series(y_preds).sort_values().value_counts())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 181,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "      P1 win       0.61      0.97      0.75   2411898\n",
      "      P2 win       0.92      0.38      0.54   2412834\n",
      "\n",
      "    accuracy                           0.67   4824732\n",
      "   macro avg       0.77      0.67      0.64   4824732\n",
      "weighted avg       0.77      0.67      0.64   4824732\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import classification_report\n",
    "target_names = ['P1 win', 'P2 win']\n",
    "print(classification_report(y_test.astype(int), y_preds.round(), target_names=target_names))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Feature Selection"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 182,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "      P1 win       0.50      0.80      0.61   2413402\n",
      "      P2 win       0.50      0.20      0.29   2411330\n",
      "\n",
      "    accuracy                           0.50   4824732\n",
      "   macro avg       0.50      0.50      0.45   4824732\n",
      "weighted avg       0.50      0.50      0.45   4824732\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.feature_selection import SelectKBest, chi2\n",
    "X_new = SelectKBest(chi2, k=7).fit_transform(X_norm + 1, y)\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X_new, y.astype(int), shuffle = True, train_size=0.3)\n",
    "\n",
    "reg = TweedieRegressor(power=1, alpha=0.5, link='log')\n",
    "reg.fit(X_train, y_train.astype(int))\n",
    "\n",
    "target_names = ['P1 win', 'P2 win']\n",
    "print(classification_report(y_test, y_preds.round(), target_names=target_names))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 183,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1 1 1 ... 1 1 0]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      P1 win       1.00      0.71      0.83   2413402\n",
      "      P2 win       0.77      1.00      0.87   2411330\n",
      "\n",
      "    accuracy                           0.85   4824732\n",
      "   macro avg       0.89      0.85      0.85   4824732\n",
      "weighted avg       0.89      0.85      0.85   4824732\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.neural_network import MLPClassifier\n",
    "clf = MLPClassifier(solver='lbfgs', alpha=1e-5,\n",
    "                    hidden_layer_sizes=(15,), random_state=1)\n",
    "\n",
    "clf.fit(X_train, y_train)\n",
    "\n",
    "y_preds = clf.predict(X_test)\n",
    "print(y_preds)\n",
    "\n",
    "print(classification_report(y_test, y_preds, target_names=target_names))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "base",
   "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": "3.11.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}