ai_pipeline_tutorial.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# How to train a predicitve machine learning to predict Premier League Matches"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Table Of Contents\n",
    "1. [Abstract](#abstract)\n",
    "2. [Learning Objectives](#learning-objectives)\n",
    "3. [Import Libraries](#1-import-librarires)\n",
    "4. [Retrieve Dataset](#4-retrieve-dataset)\n",
    "5. [Data Preprocessing](#5-data-preprocessing)\n",
    "6. [Building Neural Network with Pytorch](#6-building-neural-network-with-pytorch)\n",
    "7. [Train Model](#7-train-model)\n",
    "8. [Plot the loss of model over training](#8-plot-the-loss-of-your-model-over-training)\n",
    "9. [Evaluate Model After Training](#9-evaluate-performance-of-the-model)\n",
    "10. [Hyperparameter Tuning](#10-hyperparameter-tuning)\n",
    "11. [Final Model](#11-final-model)\n",
    "12. [Display Predictions](#12-display-Predictions)\n",
    "13. [Conclusion](#13-conclusion)\n",
    "14. [Enhancements From Source Tutorials](#14-differences-and-similarities)\n",
    "15. [References](#15-references)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='abstract'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Abstract\n",
    "\n",
    "\n",
    "This tutorial presents a comprehensive guide to predicting Premier League football matches using a machine learning pipeline in PyTorch. The tutorial covers essential steps, including data retrieval, preprocessing, feature selection, and normalization. We delve into neural network architecture design and training, fine-tune model parameters through hyperparameter tuning, and evaluate model performance. By integrating both team and opponent data, the tutorial highlights the significance of context in predictive modeling. The final model, trained with optimal hyperparameters, demonstrates the practical application of machine learning in sports analytics."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Learning Objectives\n",
    "<a id='learning-objectives'></a>\n",
    "\n",
    "- Data preprocessing and normalisation\n",
    "\n",
    "- Feature selection: Find the best subset of features using filters\n",
    "\n",
    "- Model Building: Learn how to build a neural network using Pytorch\n",
    "\n",
    "- Model Training and Prediction: Learning how to train a machine learning model and plot loss function.\n",
    "\n",
    "- Model Evaluation: Gaining skills in evaluating the performance of machine learning models using metrics like accuracy and precision. As well as using visualisations such as confusion matrix\n",
    "\n",
    "- Hyperparameter tuning: Learn how to optimise our neural network by means of hyperparameter tuning\n",
    "\n",
    "- How to utilise pandas to output final predictions in readable format"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "### 3. Import librarires to be used throughout the tutorial\n",
    "<a id='1-import-librarires'></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2055,
   "metadata": {},
   "outputs": [],
   "source": [
    "# For NumPy\n",
    "import numpy as np\n",
    "# For PyTorch\n",
    "import torch\n",
    "# For matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "# For pandas\n",
    "import pandas as pd\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='4-retrieve-dataset'></a>\n",
    "### 4. Retrieve dataset\n",
    "\n",
    "We'll be diving into the heart of our predictive model: the dataset. For this tutorial, we're using a comprehensive dataset of Premier League matches. This dataset offers a rich compilation of match data, providing a solid foundation for our predictive analysis.\n",
    "\n",
    "We start by importing our dataset into our Python environment. This dataset, aptly named match_data.csv, encompasses a range of match-specific details, essential for our analysis. You can download this dataset directly from the provided Kaggle link [here](https://www.kaggle.com/datasets/mchinaka/epl-match-data-202021-202324/settings). Once downloaded, place it in an accessible directory for this tutorial.\n",
    "\n",
    "Let's load our dataset using Pandas, an indispensable tool in our data science arsenal, and glimpse the structure and breadth of our data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2056,
   "metadata": {},
   "outputs": [],
   "source": [
    "# eventually change this to fetch from my kaggle\n",
    "matches = pd.read_csv('match_data.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2057,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(2036, 55)"
      ]
     },
     "execution_count": 2057,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# print the shape of the dataframe\n",
    "matches.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2058,
   "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>Unnamed: 0</th>\n",
       "      <th>date</th>\n",
       "      <th>time</th>\n",
       "      <th>comp</th>\n",
       "      <th>round</th>\n",
       "      <th>day</th>\n",
       "      <th>venue</th>\n",
       "      <th>result</th>\n",
       "      <th>gf</th>\n",
       "      <th>ga</th>\n",
       "      <th>opponent</th>\n",
       "      <th>xg_x</th>\n",
       "      <th>xga</th>\n",
       "      <th>poss</th>\n",
       "      <th>attendance</th>\n",
       "      <th>captain</th>\n",
       "      <th>formation</th>\n",
       "      <th>referee</th>\n",
       "      <th>match report</th>\n",
       "      <th>notes</th>\n",
       "      <th>sh</th>\n",
       "      <th>sot</th>\n",
       "      <th>dist</th>\n",
       "      <th>xg_y</th>\n",
       "      <th>npxg</th>\n",
       "      <th>g-xg</th>\n",
       "      <th>sota</th>\n",
       "      <th>saves</th>\n",
       "      <th>save%</th>\n",
       "      <th>att (gk)</th>\n",
       "      <th>thr</th>\n",
       "      <th>psxg</th>\n",
       "      <th>psxg+/-</th>\n",
       "      <th>ast</th>\n",
       "      <th>xag</th>\n",
       "      <th>xa</th>\n",
       "      <th>kp</th>\n",
       "      <th>1/3</th>\n",
       "      <th>ppa</th>\n",
       "      <th>crspa</th>\n",
       "      <th>prgp</th>\n",
       "      <th>attempted passes</th>\n",
       "      <th>completed passes</th>\n",
       "      <th>corner kicks</th>\n",
       "      <th>shots blocked</th>\n",
       "      <th>tkl+int</th>\n",
       "      <th>err</th>\n",
       "      <th>mid 3rd</th>\n",
       "      <th>att pen</th>\n",
       "      <th>att 3rd</th>\n",
       "      <th>fls</th>\n",
       "      <th>fld</th>\n",
       "      <th>off</th>\n",
       "      <th>season</th>\n",
       "      <th>team</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1364</td>\n",
       "      <td>2020-09-12</td>\n",
       "      <td>12:30</td>\n",
       "      <td>Premier League</td>\n",
       "      <td>Matchweek 1</td>\n",
       "      <td>Sat</td>\n",
       "      <td>Away</td>\n",
       "      <td>W</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Fulham</td>\n",
       "      <td>1.9</td>\n",
       "      <td>0.1</td>\n",
       "      <td>54.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Pierre-Emerick Aubameyang</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>Chris Kavanagh</td>\n",
       "      <td>Match Report</td>\n",
       "      <td>NaN</td>\n",
       "      <td>13.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>14.1</td>\n",
       "      <td>1.9</td>\n",
       "      <td>1.9</td>\n",
       "      <td>1.1</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>29.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.1</td>\n",
       "      <td>2</td>\n",
       "      <td>0.8</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>38.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>47.0</td>\n",
       "      <td>668.0</td>\n",
       "      <td>589.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>30</td>\n",
       "      <td>0.0</td>\n",
       "      <td>343.0</td>\n",
       "      <td>26.0</td>\n",
       "      <td>145.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1365</td>\n",
       "      <td>2020-09-19</td>\n",
       "      <td>20:00</td>\n",
       "      <td>Premier League</td>\n",
       "      <td>Matchweek 2</td>\n",
       "      <td>Sat</td>\n",
       "      <td>Home</td>\n",
       "      <td>W</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>West Ham</td>\n",
       "      <td>1.1</td>\n",
       "      <td>2.0</td>\n",
       "      <td>62.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Pierre-Emerick Aubameyang</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>Michael Oliver</td>\n",
       "      <td>Match Report</td>\n",
       "      <td>NaN</td>\n",
       "      <td>7.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>14.8</td>\n",
       "      <td>1.1</td>\n",
       "      <td>1.1</td>\n",
       "      <td>0.9</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>50.0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.7</td>\n",
       "      <td>-0.3</td>\n",
       "      <td>2</td>\n",
       "      <td>1.1</td>\n",
       "      <td>1.6</td>\n",
       "      <td>6.0</td>\n",
       "      <td>38.0</td>\n",
       "      <td>16.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>52.0</td>\n",
       "      <td>710.0</td>\n",
       "      <td>599.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>19</td>\n",
       "      <td>1.0</td>\n",
       "      <td>423.0</td>\n",
       "      <td>28.0</td>\n",
       "      <td>166.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1366</td>\n",
       "      <td>2020-09-28</td>\n",
       "      <td>20:00</td>\n",
       "      <td>Premier League</td>\n",
       "      <td>Matchweek 3</td>\n",
       "      <td>Mon</td>\n",
       "      <td>Away</td>\n",
       "      <td>L</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>Liverpool</td>\n",
       "      <td>1.3</td>\n",
       "      <td>2.7</td>\n",
       "      <td>34.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Pierre-Emerick Aubameyang</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>Craig Pawson</td>\n",
       "      <td>Match Report</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>1.3</td>\n",
       "      <td>1.3</td>\n",
       "      <td>-0.3</td>\n",
       "      <td>8.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>62.5</td>\n",
       "      <td>34.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>2.6</td>\n",
       "      <td>-0.4</td>\n",
       "      <td>0</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.6</td>\n",
       "      <td>3.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>425.0</td>\n",
       "      <td>312.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>18</td>\n",
       "      <td>1.0</td>\n",
       "      <td>174.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>44.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1367</td>\n",
       "      <td>2020-10-04</td>\n",
       "      <td>14:00</td>\n",
       "      <td>Premier League</td>\n",
       "      <td>Matchweek 4</td>\n",
       "      <td>Sun</td>\n",
       "      <td>Home</td>\n",
       "      <td>W</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Sheffield Utd</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.2</td>\n",
       "      <td>64.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Pierre-Emerick Aubameyang</td>\n",
       "      <td>4-3-3</td>\n",
       "      <td>Lee Mason</td>\n",
       "      <td>Match Report</td>\n",
       "      <td>NaN</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>16.3</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.6</td>\n",
       "      <td>1.4</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>50.0</td>\n",
       "      <td>37.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.4</td>\n",
       "      <td>-0.6</td>\n",
       "      <td>2</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.6</td>\n",
       "      <td>6.0</td>\n",
       "      <td>41.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>49.0</td>\n",
       "      <td>741.0</td>\n",
       "      <td>646.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>18</td>\n",
       "      <td>1.0</td>\n",
       "      <td>410.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>169.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1368</td>\n",
       "      <td>2020-10-17</td>\n",
       "      <td>17:30</td>\n",
       "      <td>Premier League</td>\n",
       "      <td>Matchweek 5</td>\n",
       "      <td>Sat</td>\n",
       "      <td>Away</td>\n",
       "      <td>L</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Manchester City</td>\n",
       "      <td>0.9</td>\n",
       "      <td>1.3</td>\n",
       "      <td>42.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Pierre-Emerick Aubameyang</td>\n",
       "      <td>4-3-3</td>\n",
       "      <td>Chris Kavanagh</td>\n",
       "      <td>Match Report</td>\n",
       "      <td>NaN</td>\n",
       "      <td>11.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>18.4</td>\n",
       "      <td>0.9</td>\n",
       "      <td>0.9</td>\n",
       "      <td>-0.9</td>\n",
       "      <td>5.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>80.0</td>\n",
       "      <td>48.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0</td>\n",
       "      <td>0.7</td>\n",
       "      <td>0.6</td>\n",
       "      <td>7.0</td>\n",
       "      <td>17.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>22.0</td>\n",
       "      <td>496.0</td>\n",
       "      <td>411.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>19</td>\n",
       "      <td>0.0</td>\n",
       "      <td>239.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>92.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0        date   time            comp        round  day venue  \\\n",
       "0        1364  2020-09-12  12:30  Premier League  Matchweek 1  Sat  Away   \n",
       "1        1365  2020-09-19  20:00  Premier League  Matchweek 2  Sat  Home   \n",
       "2        1366  2020-09-28  20:00  Premier League  Matchweek 3  Mon  Away   \n",
       "3        1367  2020-10-04  14:00  Premier League  Matchweek 4  Sun  Home   \n",
       "4        1368  2020-10-17  17:30  Premier League  Matchweek 5  Sat  Away   \n",
       "\n",
       "  result   gf   ga         opponent  xg_x  xga  poss  attendance  \\\n",
       "0      W  3.0  0.0           Fulham   1.9  0.1  54.0         NaN   \n",
       "1      W  2.0  1.0         West Ham   1.1  2.0  62.0         NaN   \n",
       "2      L  1.0  3.0        Liverpool   1.3  2.7  34.0         NaN   \n",
       "3      W  2.0  1.0    Sheffield Utd   0.6  0.2  64.0         NaN   \n",
       "4      L  0.0  1.0  Manchester City   0.9  1.3  42.0         NaN   \n",
       "\n",
       "                     captain formation         referee  match report  notes  \\\n",
       "0  Pierre-Emerick Aubameyang     3-4-3  Chris Kavanagh  Match Report    NaN   \n",
       "1  Pierre-Emerick Aubameyang     3-4-3  Michael Oliver  Match Report    NaN   \n",
       "2  Pierre-Emerick Aubameyang     3-4-3    Craig Pawson  Match Report    NaN   \n",
       "3  Pierre-Emerick Aubameyang     4-3-3       Lee Mason  Match Report    NaN   \n",
       "4  Pierre-Emerick Aubameyang     4-3-3  Chris Kavanagh  Match Report    NaN   \n",
       "\n",
       "     sh  sot  dist  xg_y  npxg  g-xg  sota  saves  save%  att (gk)  thr  psxg  \\\n",
       "0  13.0  5.0  14.1   1.9   1.9   1.1   2.0    2.0  100.0      29.0  3.0   0.1   \n",
       "1   7.0  3.0  14.8   1.1   1.1   0.9   2.0    1.0   50.0      32.0  6.0   0.7   \n",
       "2   4.0  3.0  15.0   1.3   1.3  -0.3   8.0    5.0   62.5      34.0  6.0   2.6   \n",
       "3   6.0  5.0  16.3   0.6   0.6   1.4   2.0    1.0   50.0      37.0  2.0   0.4   \n",
       "4  11.0  3.0  18.4   0.9   0.9  -0.9   5.0    4.0   80.0      48.0  4.0   1.4   \n",
       "\n",
       "   psxg+/-  ast  xag   xa   kp   1/3   ppa  crspa  prgp  attempted passes  \\\n",
       "0      0.1    2  0.8  1.0  8.0  38.0  11.0    1.0  47.0             668.0   \n",
       "1     -0.3    2  1.1  1.6  6.0  38.0  16.0    3.0  52.0             710.0   \n",
       "2     -0.4    0  0.8  0.6  3.0  11.0   1.0    0.0  13.0             425.0   \n",
       "3     -0.6    2  0.6  0.6  6.0  41.0   4.0    1.0  49.0             741.0   \n",
       "4      0.4    0  0.7  0.6  7.0  17.0   5.0    3.0  22.0             496.0   \n",
       "\n",
       "   completed passes  corner kicks  shots blocked  tkl+int  err  mid 3rd  \\\n",
       "0             589.0           3.0            1.0       30  0.0    343.0   \n",
       "1             599.0           7.0            6.0       19  1.0    423.0   \n",
       "2             312.0           3.0            6.0       18  1.0    174.0   \n",
       "3             646.0           2.0            2.0       18  1.0    410.0   \n",
       "4             411.0           6.0            4.0       19  0.0    239.0   \n",
       "\n",
       "   att pen  att 3rd   fls   fld  off  season     team  \n",
       "0     26.0    145.0  12.0  12.0  2.0    2021  Arsenal  \n",
       "1     28.0    166.0  11.0  12.0  5.0    2021  Arsenal  \n",
       "2     10.0     44.0   7.0  11.0  5.0    2021  Arsenal  \n",
       "3     14.0    169.0   3.0   8.0  0.0    2021  Arsenal  \n",
       "4     13.0     92.0  10.0  15.0  2.0    2021  Arsenal  "
      ]
     },
     "execution_count": 2058,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Set Pandas display options to show all columns\n",
    "pd.set_option('display.max_columns', None)\n",
    "\n",
    "# Display the DataFrame\n",
    "matches.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='5-data-preprocessing'></a>\n",
    "\n",
    "### 5. Data Preprocessing"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this crucial step, we transform our raw data into a format that will be more effective for training our machine learning model. Data preprocessing is vital in any data science project as it helps in reducing noise, handling missing values, and making the dataset more manageable for algorithms. We will take several actions to clean and prepare our dataset:\n",
    "\n",
    "1. Removing Irrelevant Columns: Certain columns in the dataset, such as referee names or attendance numbers, might not contribute significantly to our predictive model's performance. We will remove these to focus on more impactful features.\n",
    "\n",
    "2. One-Hot Encoding Categorical Variables: Machine learning models work best with numerical inputs. We will convert categorical variables like 'venue' and 'opponent' into numerical form using one-hot encoding. This process involves creating new binary columns for each category in the original column.\n",
    "\n",
    "3. Creating Target Variables: We will create a target variable 'target' to represent the result of the match as a binary outcome (win or not win). Similarly, we will encode 'venue' as a binary variable to represent home or away games.\n",
    "\n",
    "4. Handling Date and Time Data: The 'date' column will be converted to a DateTime object to facilitate any time-series analysis or feature engineering involving dates.\n",
    "\n",
    "Through these steps, we'll prepare our data to effectively feed into our neural network model. Let's dive into the code:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2059,
   "metadata": {},
   "outputs": [],
   "source": [
    "del matches[\"comp\"]\n",
    "del matches[\"notes\"]\n",
    "del matches[\"referee\"]\n",
    "del matches[\"Unnamed: 0\"] # remove the index column as pandas has its own index\n",
    "del matches[\"captain\"]\n",
    "del matches[\"match report\"]\n",
    "del matches[\"attendance\"] # lots of missing data and not relevant to the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2060,
   "metadata": {},
   "outputs": [],
   "source": [
    "# redundant field, has same values as xg_x \n",
    "del matches[\"xg_y\"] "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2061,
   "metadata": {},
   "outputs": [],
   "source": [
    "# source [3]\n",
    "# https://app.dataquest.io/m/99992/portfolio-project%3A-predicting-epl-football-match-winners-using-machine-learning/1/project-overview\n",
    "\n",
    "# create one hot encoding target variable representing win or not win.\n",
    "matches[\"target\"] = (matches[\"result\"] == \"W\").astype(\"int\")\n",
    "# create one hot encoding target variable representing Home or Away\n",
    "matches[\"venue_code\"] = matches[\"venue\"].astype(\"category\").cat.codes\n",
    "# create one hot encoding target variable giving a code representation of a team\n",
    "matches[\"opp_code\"] = matches[\"opponent\"].astype(\"category\").cat.codes\n",
    "# create one hot encoding target variable giving a code representation of a team\n",
    "matches[\"team_code\"] = matches[\"team\"].astype(\"category\").cat.codes\n",
    "\n",
    "#  end of source [3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2062,
   "metadata": {},
   "outputs": [],
   "source": [
    "# change the date to a datetime object\n",
    "matches[\"date\"] = pd.to_datetime(matches[\"date\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2063,
   "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>date</th>\n",
       "      <th>time</th>\n",
       "      <th>round</th>\n",
       "      <th>day</th>\n",
       "      <th>venue</th>\n",
       "      <th>result</th>\n",
       "      <th>gf</th>\n",
       "      <th>ga</th>\n",
       "      <th>opponent</th>\n",
       "      <th>xg_x</th>\n",
       "      <th>xga</th>\n",
       "      <th>poss</th>\n",
       "      <th>formation</th>\n",
       "      <th>sh</th>\n",
       "      <th>sot</th>\n",
       "      <th>dist</th>\n",
       "      <th>npxg</th>\n",
       "      <th>g-xg</th>\n",
       "      <th>sota</th>\n",
       "      <th>saves</th>\n",
       "      <th>save%</th>\n",
       "      <th>att (gk)</th>\n",
       "      <th>thr</th>\n",
       "      <th>psxg</th>\n",
       "      <th>psxg+/-</th>\n",
       "      <th>ast</th>\n",
       "      <th>xag</th>\n",
       "      <th>xa</th>\n",
       "      <th>kp</th>\n",
       "      <th>1/3</th>\n",
       "      <th>ppa</th>\n",
       "      <th>crspa</th>\n",
       "      <th>prgp</th>\n",
       "      <th>attempted passes</th>\n",
       "      <th>completed passes</th>\n",
       "      <th>corner kicks</th>\n",
       "      <th>shots blocked</th>\n",
       "      <th>tkl+int</th>\n",
       "      <th>err</th>\n",
       "      <th>mid 3rd</th>\n",
       "      <th>att pen</th>\n",
       "      <th>att 3rd</th>\n",
       "      <th>fls</th>\n",
       "      <th>fld</th>\n",
       "      <th>off</th>\n",
       "      <th>season</th>\n",
       "      <th>team</th>\n",
       "      <th>target</th>\n",
       "      <th>venue_code</th>\n",
       "      <th>opp_code</th>\n",
       "      <th>team_code</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2020-09-12</td>\n",
       "      <td>12:30</td>\n",
       "      <td>Matchweek 1</td>\n",
       "      <td>Sat</td>\n",
       "      <td>Away</td>\n",
       "      <td>W</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Fulham</td>\n",
       "      <td>1.9</td>\n",
       "      <td>0.1</td>\n",
       "      <td>54.0</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>13.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>14.1</td>\n",
       "      <td>1.9</td>\n",
       "      <td>1.1</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>29.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.1</td>\n",
       "      <td>2</td>\n",
       "      <td>0.8</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>38.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>47.0</td>\n",
       "      <td>668.0</td>\n",
       "      <td>589.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>30</td>\n",
       "      <td>0.0</td>\n",
       "      <td>343.0</td>\n",
       "      <td>26.0</td>\n",
       "      <td>145.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>9</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2020-09-19</td>\n",
       "      <td>20:00</td>\n",
       "      <td>Matchweek 2</td>\n",
       "      <td>Sat</td>\n",
       "      <td>Home</td>\n",
       "      <td>W</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>West Ham</td>\n",
       "      <td>1.1</td>\n",
       "      <td>2.0</td>\n",
       "      <td>62.0</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>7.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>14.8</td>\n",
       "      <td>1.1</td>\n",
       "      <td>0.9</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>50.0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.7</td>\n",
       "      <td>-0.3</td>\n",
       "      <td>2</td>\n",
       "      <td>1.1</td>\n",
       "      <td>1.6</td>\n",
       "      <td>6.0</td>\n",
       "      <td>38.0</td>\n",
       "      <td>16.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>52.0</td>\n",
       "      <td>710.0</td>\n",
       "      <td>599.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>19</td>\n",
       "      <td>1.0</td>\n",
       "      <td>423.0</td>\n",
       "      <td>28.0</td>\n",
       "      <td>166.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>22</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2020-09-28</td>\n",
       "      <td>20:00</td>\n",
       "      <td>Matchweek 3</td>\n",
       "      <td>Mon</td>\n",
       "      <td>Away</td>\n",
       "      <td>L</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>Liverpool</td>\n",
       "      <td>1.3</td>\n",
       "      <td>2.7</td>\n",
       "      <td>34.0</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>4.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>1.3</td>\n",
       "      <td>-0.3</td>\n",
       "      <td>8.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>62.5</td>\n",
       "      <td>34.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>2.6</td>\n",
       "      <td>-0.4</td>\n",
       "      <td>0</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.6</td>\n",
       "      <td>3.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>425.0</td>\n",
       "      <td>312.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>18</td>\n",
       "      <td>1.0</td>\n",
       "      <td>174.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>44.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>12</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2020-10-04</td>\n",
       "      <td>14:00</td>\n",
       "      <td>Matchweek 4</td>\n",
       "      <td>Sun</td>\n",
       "      <td>Home</td>\n",
       "      <td>W</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Sheffield Utd</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.2</td>\n",
       "      <td>64.0</td>\n",
       "      <td>4-3-3</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>16.3</td>\n",
       "      <td>0.6</td>\n",
       "      <td>1.4</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>50.0</td>\n",
       "      <td>37.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.4</td>\n",
       "      <td>-0.6</td>\n",
       "      <td>2</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.6</td>\n",
       "      <td>6.0</td>\n",
       "      <td>41.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>49.0</td>\n",
       "      <td>741.0</td>\n",
       "      <td>646.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>18</td>\n",
       "      <td>1.0</td>\n",
       "      <td>410.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>169.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>18</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2020-10-17</td>\n",
       "      <td>17:30</td>\n",
       "      <td>Matchweek 5</td>\n",
       "      <td>Sat</td>\n",
       "      <td>Away</td>\n",
       "      <td>L</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Manchester City</td>\n",
       "      <td>0.9</td>\n",
       "      <td>1.3</td>\n",
       "      <td>42.0</td>\n",
       "      <td>4-3-3</td>\n",
       "      <td>11.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>18.4</td>\n",
       "      <td>0.9</td>\n",
       "      <td>-0.9</td>\n",
       "      <td>5.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>80.0</td>\n",
       "      <td>48.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0</td>\n",
       "      <td>0.7</td>\n",
       "      <td>0.6</td>\n",
       "      <td>7.0</td>\n",
       "      <td>17.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>22.0</td>\n",
       "      <td>496.0</td>\n",
       "      <td>411.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>19</td>\n",
       "      <td>0.0</td>\n",
       "      <td>239.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>92.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>14</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2031</th>\n",
       "      <td>2024-02-01</td>\n",
       "      <td>20:15</td>\n",
       "      <td>Matchweek 22</td>\n",
       "      <td>Thu</td>\n",
       "      <td>Home</td>\n",
       "      <td>L</td>\n",
       "      <td>3.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>Manchester Utd</td>\n",
       "      <td>2.1</td>\n",
       "      <td>3.3</td>\n",
       "      <td>53.0</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>15.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>16.6</td>\n",
       "      <td>1.3</td>\n",
       "      <td>0.9</td>\n",
       "      <td>8.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>50.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3.4</td>\n",
       "      <td>-0.6</td>\n",
       "      <td>2</td>\n",
       "      <td>1.1</td>\n",
       "      <td>0.6</td>\n",
       "      <td>11.0</td>\n",
       "      <td>36.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>53.0</td>\n",
       "      <td>550.0</td>\n",
       "      <td>469.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>28</td>\n",
       "      <td>1.0</td>\n",
       "      <td>288.0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>196.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2023</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>15</td>\n",
       "      <td>23</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2032</th>\n",
       "      <td>2024-02-04</td>\n",
       "      <td>14:00</td>\n",
       "      <td>Matchweek 23</td>\n",
       "      <td>Sun</td>\n",
       "      <td>Away</td>\n",
       "      <td>W</td>\n",
       "      <td>4.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>Chelsea</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.7</td>\n",
       "      <td>36.0</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>13.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>16.6</td>\n",
       "      <td>1.2</td>\n",
       "      <td>1.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>50.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>2.1</td>\n",
       "      <td>0.1</td>\n",
       "      <td>2</td>\n",
       "      <td>0.7</td>\n",
       "      <td>0.6</td>\n",
       "      <td>8.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>386.0</td>\n",
       "      <td>305.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>26</td>\n",
       "      <td>0.0</td>\n",
       "      <td>193.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>93.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2023</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>23</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2033</th>\n",
       "      <td>2024-02-10</td>\n",
       "      <td>15:00</td>\n",
       "      <td>Matchweek 24</td>\n",
       "      <td>Sat</td>\n",
       "      <td>Home</td>\n",
       "      <td>L</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>Brentford</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.9</td>\n",
       "      <td>71.0</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>17.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>17.2</td>\n",
       "      <td>1.0</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>66.7</td>\n",
       "      <td>26.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.9</td>\n",
       "      <td>-0.1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.9</td>\n",
       "      <td>1.2</td>\n",
       "      <td>14.0</td>\n",
       "      <td>53.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>58.0</td>\n",
       "      <td>719.0</td>\n",
       "      <td>596.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>26</td>\n",
       "      <td>1.0</td>\n",
       "      <td>393.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>271.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2023</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>23</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2034</th>\n",
       "      <td>2024-02-17</td>\n",
       "      <td>15:00</td>\n",
       "      <td>Matchweek 25</td>\n",
       "      <td>Sat</td>\n",
       "      <td>Away</td>\n",
       "      <td>W</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Tottenham</td>\n",
       "      <td>1.8</td>\n",
       "      <td>1.6</td>\n",
       "      <td>29.0</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>12.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>15.9</td>\n",
       "      <td>1.8</td>\n",
       "      <td>0.2</td>\n",
       "      <td>4.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>75.0</td>\n",
       "      <td>19.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.4</td>\n",
       "      <td>2</td>\n",
       "      <td>1.1</td>\n",
       "      <td>0.6</td>\n",
       "      <td>10.0</td>\n",
       "      <td>18.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>26.0</td>\n",
       "      <td>329.0</td>\n",
       "      <td>232.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>27</td>\n",
       "      <td>0.0</td>\n",
       "      <td>141.0</td>\n",
       "      <td>22.0</td>\n",
       "      <td>110.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2023</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>20</td>\n",
       "      <td>23</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2035</th>\n",
       "      <td>2024-02-25</td>\n",
       "      <td>13:30</td>\n",
       "      <td>Matchweek 26</td>\n",
       "      <td>Sun</td>\n",
       "      <td>Home</td>\n",
       "      <td>W</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Sheffield Utd</td>\n",
       "      <td>1.2</td>\n",
       "      <td>0.6</td>\n",
       "      <td>63.0</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>13.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>14.4</td>\n",
       "      <td>1.2</td>\n",
       "      <td>-0.2</td>\n",
       "      <td>4.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>17.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.4</td>\n",
       "      <td>1</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.8</td>\n",
       "      <td>9.0</td>\n",
       "      <td>36.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>41.0</td>\n",
       "      <td>645.0</td>\n",
       "      <td>545.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>27</td>\n",
       "      <td>0.0</td>\n",
       "      <td>359.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>166.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2023</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>18</td>\n",
       "      <td>23</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2036 rows × 51 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           date   time         round  day venue result   gf   ga  \\\n",
       "0    2020-09-12  12:30   Matchweek 1  Sat  Away      W  3.0  0.0   \n",
       "1    2020-09-19  20:00   Matchweek 2  Sat  Home      W  2.0  1.0   \n",
       "2    2020-09-28  20:00   Matchweek 3  Mon  Away      L  1.0  3.0   \n",
       "3    2020-10-04  14:00   Matchweek 4  Sun  Home      W  2.0  1.0   \n",
       "4    2020-10-17  17:30   Matchweek 5  Sat  Away      L  0.0  1.0   \n",
       "...         ...    ...           ...  ...   ...    ...  ...  ...   \n",
       "2031 2024-02-01  20:15  Matchweek 22  Thu  Home      L  3.0  4.0   \n",
       "2032 2024-02-04  14:00  Matchweek 23  Sun  Away      W  4.0  2.0   \n",
       "2033 2024-02-10  15:00  Matchweek 24  Sat  Home      L  0.0  2.0   \n",
       "2034 2024-02-17  15:00  Matchweek 25  Sat  Away      W  2.0  1.0   \n",
       "2035 2024-02-25  13:30  Matchweek 26  Sun  Home      W  1.0  0.0   \n",
       "\n",
       "             opponent  xg_x  xga  poss formation    sh  sot  dist  npxg  g-xg  \\\n",
       "0              Fulham   1.9  0.1  54.0     3-4-3  13.0  5.0  14.1   1.9   1.1   \n",
       "1            West Ham   1.1  2.0  62.0     3-4-3   7.0  3.0  14.8   1.1   0.9   \n",
       "2           Liverpool   1.3  2.7  34.0     3-4-3   4.0  3.0  15.0   1.3  -0.3   \n",
       "3       Sheffield Utd   0.6  0.2  64.0     4-3-3   6.0  5.0  16.3   0.6   1.4   \n",
       "4     Manchester City   0.9  1.3  42.0     4-3-3  11.0  3.0  18.4   0.9  -0.9   \n",
       "...               ...   ...  ...   ...       ...   ...  ...   ...   ...   ...   \n",
       "2031   Manchester Utd   2.1  3.3  53.0     3-4-3  15.0  5.0  16.6   1.3   0.9   \n",
       "2032          Chelsea   2.0  1.7  36.0     3-4-3  13.0  6.0  16.6   1.2   1.0   \n",
       "2033        Brentford   1.0  1.9  71.0     3-4-3  17.0  5.0  17.2   1.0  -1.0   \n",
       "2034        Tottenham   1.8  1.6  29.0     3-4-3  12.0  7.0  15.9   1.8   0.2   \n",
       "2035    Sheffield Utd   1.2  0.6  63.0     3-4-3  13.0  2.0  14.4   1.2  -0.2   \n",
       "\n",
       "      sota  saves  save%  att (gk)  thr  psxg  psxg+/-  ast  xag   xa    kp  \\\n",
       "0      2.0    2.0  100.0      29.0  3.0   0.1      0.1    2  0.8  1.0   8.0   \n",
       "1      2.0    1.0   50.0      32.0  6.0   0.7     -0.3    2  1.1  1.6   6.0   \n",
       "2      8.0    5.0   62.5      34.0  6.0   2.6     -0.4    0  0.8  0.6   3.0   \n",
       "3      2.0    1.0   50.0      37.0  2.0   0.4     -0.6    2  0.6  0.6   6.0   \n",
       "4      5.0    4.0   80.0      48.0  4.0   1.4      0.4    0  0.7  0.6   7.0   \n",
       "...    ...    ...    ...       ...  ...   ...      ...  ...  ...  ...   ...   \n",
       "2031   8.0    4.0   50.0      15.0  2.0   3.4     -0.6    2  1.1  0.6  11.0   \n",
       "2032   4.0    2.0   50.0      21.0  6.0   2.1      0.1    2  0.7  0.6   8.0   \n",
       "2033   6.0    4.0   66.7      26.0  4.0   1.9     -0.1    0  0.9  1.2  14.0   \n",
       "2034   4.0    3.0   75.0      19.0  5.0   1.4      0.4    2  1.1  0.6  10.0   \n",
       "2035   4.0    4.0  100.0      17.0  7.0   0.4      0.4    1  0.8  0.8   9.0   \n",
       "\n",
       "       1/3   ppa  crspa  prgp  attempted passes  completed passes  \\\n",
       "0     38.0  11.0    1.0  47.0             668.0             589.0   \n",
       "1     38.0  16.0    3.0  52.0             710.0             599.0   \n",
       "2     11.0   1.0    0.0  13.0             425.0             312.0   \n",
       "3     41.0   4.0    1.0  49.0             741.0             646.0   \n",
       "4     17.0   5.0    3.0  22.0             496.0             411.0   \n",
       "...    ...   ...    ...   ...               ...               ...   \n",
       "2031  36.0  13.0    2.0  53.0             550.0             469.0   \n",
       "2032   8.0   3.0    0.0  13.0             386.0             305.0   \n",
       "2033  53.0   7.0    2.0  58.0             719.0             596.0   \n",
       "2034  18.0   5.0    2.0  26.0             329.0             232.0   \n",
       "2035  36.0   9.0    4.0  41.0             645.0             545.0   \n",
       "\n",
       "      corner kicks  shots blocked  tkl+int  err  mid 3rd  att pen  att 3rd  \\\n",
       "0              3.0            1.0       30  0.0    343.0     26.0    145.0   \n",
       "1              7.0            6.0       19  1.0    423.0     28.0    166.0   \n",
       "2              3.0            6.0       18  1.0    174.0     10.0     44.0   \n",
       "3              2.0            2.0       18  1.0    410.0     14.0    169.0   \n",
       "4              6.0            4.0       19  0.0    239.0     13.0     92.0   \n",
       "...            ...            ...      ...  ...      ...      ...      ...   \n",
       "2031           4.0            6.0       28  1.0    288.0     32.0    196.0   \n",
       "2032           1.0            4.0       26  0.0    193.0     24.0     93.0   \n",
       "2033           7.0            0.0       26  1.0    393.0     21.0    271.0   \n",
       "2034           4.0            4.0       27  0.0    141.0     22.0    110.0   \n",
       "2035           5.0            4.0       27  0.0    359.0     24.0    166.0   \n",
       "\n",
       "       fls   fld  off  season                     team  target  venue_code  \\\n",
       "0     12.0  12.0  2.0    2021                  Arsenal       1           0   \n",
       "1     11.0  12.0  5.0    2021                  Arsenal       1           1   \n",
       "2      7.0  11.0  5.0    2021                  Arsenal       0           0   \n",
       "3      3.0   8.0  0.0    2021                  Arsenal       1           1   \n",
       "4     10.0  15.0  2.0    2021                  Arsenal       0           0   \n",
       "...    ...   ...  ...     ...                      ...     ...         ...   \n",
       "2031  10.0   9.0  1.0    2023  Wolverhampton Wanderers       0           1   \n",
       "2032  11.0  15.0  0.0    2023  Wolverhampton Wanderers       1           0   \n",
       "2033  11.0   8.0  1.0    2023  Wolverhampton Wanderers       0           1   \n",
       "2034  13.0   8.0  2.0    2023  Wolverhampton Wanderers       1           0   \n",
       "2035   8.0   4.0  1.0    2023  Wolverhampton Wanderers       1           1   \n",
       "\n",
       "      opp_code  team_code  \n",
       "0            9          0  \n",
       "1           22          0  \n",
       "2           12          0  \n",
       "3           18          0  \n",
       "4           14          0  \n",
       "...        ...        ...  \n",
       "2031        15         23  \n",
       "2032         6         23  \n",
       "2033         3         23  \n",
       "2034        20         23  \n",
       "2035        18         23  \n",
       "\n",
       "[2036 rows x 51 columns]"
      ]
     },
     "execution_count": 2063,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# final dataset post cleaning \n",
    "matches"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.2 Calculate Rolling averages for match data to predict future matches "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As we are attepting to predict future matches with our approach, we will not be using current match data to predict the outcome since they have not occured yet. We will take match data for the past 5 games and obtain a rolling average to get a sense of team form and use that to predict the outcome of the upcoming fixture.\n",
    "\n",
    "Our rolling averages will focus on various match statistics such as goals for (gf), goals against (ga), expected goals (xg), and possession (poss), among others. The idea is to smooth out anomalies and provide a more consistent and reliable picture of each team's recent performance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2064,
   "metadata": {},
   "outputs": [],
   "source": [
    "# source [3]\n",
    "# https://app.dataquest.io/m/99992/portfolio-project%3A-predicting-epl-football-match-winners-using-machine-learning/1/project-overview\n",
    "\n",
    "def rolling_averages(group, cols, new_cols):\n",
    "    group = group.sort_values(\"date\")\n",
    "    rolling_stats = group[cols].rolling(5, closed='left').mean() # closed left current match for the average\n",
    "    group[new_cols] = rolling_stats\n",
    "    group = group.dropna(subset=new_cols)\n",
    "    return group\n",
    "\n",
    "# end of source [3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2065,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/var/folders/67/8z6tkktd6711zn_69ccfm1x80000gn/T/ipykernel_28547/3650290362.py:13: DeprecationWarning: DataFrameGroupBy.apply operated on the grouping columns. This behavior is deprecated, and in a future version of pandas the grouping columns will be excluded from the operation. Either pass `include_groups=False` to exclude the groupings or explicitly select the grouping columns after groupby to silence this warning.\n",
      "  matches_rolling = matches.groupby(\"team\").apply(lambda x: rolling_averages(x, cols, new_cols))\n"
     ]
    }
   ],
   "source": [
    "# source [3]\n",
    "# https://app.dataquest.io/m/99992/portfolio-project%3A-predicting-epl-football-match-winners-using-machine-learning/1/project-overview\n",
    "\n",
    "\n",
    "# match data columns to be used for rolling averages\n",
    "cols = [\"gf\", \"ga\", \"xg_x\", \"xga\", \"poss\", \n",
    "        \"sh\", \"sot\", \"dist\", \"npxg\", \"g-xg\", \"sota\", \"saves\", \"save%\", \"att (gk)\", \"thr\", \"psxg\", \"psxg+/-\",\n",
    "        \"ast\", \"xag\", \"xa\", \"kp\", \"1/3\", \"ppa\", \"crspa\", \"prgp\", \"attempted passes\", \"completed passes\", \"corner kicks\",\n",
    "        \"shots blocked\", \"tkl+int\", \"err\", \"mid 3rd\", \"att pen\", \"att 3rd\", \"fls\", \"fld\", \"off\"] \n",
    "\n",
    "new_cols = [f\"{c}_rolling\" for c in cols]\n",
    "\n",
    "matches_rolling = matches.groupby(\"team\").apply(lambda x: rolling_averages(x, cols, new_cols))\n",
    "\n",
    "# end of source [3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2066,
   "metadata": {},
   "outputs": [],
   "source": [
    "# remove the team index created by the groupby\n",
    "matches_rolling = matches_rolling.droplevel('team')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2067,
   "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>date</th>\n",
       "      <th>time</th>\n",
       "      <th>round</th>\n",
       "      <th>day</th>\n",
       "      <th>venue</th>\n",
       "      <th>result</th>\n",
       "      <th>gf</th>\n",
       "      <th>ga</th>\n",
       "      <th>opponent</th>\n",
       "      <th>xg_x</th>\n",
       "      <th>xga</th>\n",
       "      <th>poss</th>\n",
       "      <th>formation</th>\n",
       "      <th>sh</th>\n",
       "      <th>sot</th>\n",
       "      <th>dist</th>\n",
       "      <th>npxg</th>\n",
       "      <th>g-xg</th>\n",
       "      <th>sota</th>\n",
       "      <th>saves</th>\n",
       "      <th>save%</th>\n",
       "      <th>att (gk)</th>\n",
       "      <th>thr</th>\n",
       "      <th>psxg</th>\n",
       "      <th>psxg+/-</th>\n",
       "      <th>ast</th>\n",
       "      <th>xag</th>\n",
       "      <th>xa</th>\n",
       "      <th>kp</th>\n",
       "      <th>1/3</th>\n",
       "      <th>ppa</th>\n",
       "      <th>crspa</th>\n",
       "      <th>prgp</th>\n",
       "      <th>attempted passes</th>\n",
       "      <th>completed passes</th>\n",
       "      <th>corner kicks</th>\n",
       "      <th>shots blocked</th>\n",
       "      <th>tkl+int</th>\n",
       "      <th>err</th>\n",
       "      <th>mid 3rd</th>\n",
       "      <th>att pen</th>\n",
       "      <th>att 3rd</th>\n",
       "      <th>fls</th>\n",
       "      <th>fld</th>\n",
       "      <th>off</th>\n",
       "      <th>season</th>\n",
       "      <th>team</th>\n",
       "      <th>target</th>\n",
       "      <th>venue_code</th>\n",
       "      <th>opp_code</th>\n",
       "      <th>team_code</th>\n",
       "      <th>gf_rolling</th>\n",
       "      <th>ga_rolling</th>\n",
       "      <th>xg_x_rolling</th>\n",
       "      <th>xga_rolling</th>\n",
       "      <th>poss_rolling</th>\n",
       "      <th>sh_rolling</th>\n",
       "      <th>sot_rolling</th>\n",
       "      <th>dist_rolling</th>\n",
       "      <th>npxg_rolling</th>\n",
       "      <th>g-xg_rolling</th>\n",
       "      <th>sota_rolling</th>\n",
       "      <th>saves_rolling</th>\n",
       "      <th>save%_rolling</th>\n",
       "      <th>att (gk)_rolling</th>\n",
       "      <th>thr_rolling</th>\n",
       "      <th>psxg_rolling</th>\n",
       "      <th>psxg+/-_rolling</th>\n",
       "      <th>ast_rolling</th>\n",
       "      <th>xag_rolling</th>\n",
       "      <th>xa_rolling</th>\n",
       "      <th>kp_rolling</th>\n",
       "      <th>1/3_rolling</th>\n",
       "      <th>ppa_rolling</th>\n",
       "      <th>crspa_rolling</th>\n",
       "      <th>prgp_rolling</th>\n",
       "      <th>attempted passes_rolling</th>\n",
       "      <th>completed passes_rolling</th>\n",
       "      <th>corner kicks_rolling</th>\n",
       "      <th>shots blocked_rolling</th>\n",
       "      <th>tkl+int_rolling</th>\n",
       "      <th>err_rolling</th>\n",
       "      <th>mid 3rd_rolling</th>\n",
       "      <th>att pen_rolling</th>\n",
       "      <th>att 3rd_rolling</th>\n",
       "      <th>fls_rolling</th>\n",
       "      <th>fld_rolling</th>\n",
       "      <th>off_rolling</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>2020-10-25</td>\n",
       "      <td>19:15</td>\n",
       "      <td>Matchweek 6</td>\n",
       "      <td>Sun</td>\n",
       "      <td>Home</td>\n",
       "      <td>L</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Leicester City</td>\n",
       "      <td>0.7</td>\n",
       "      <td>0.7</td>\n",
       "      <td>56.0</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>12.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>0.7</td>\n",
       "      <td>-0.7</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>50.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.3</td>\n",
       "      <td>-0.7</td>\n",
       "      <td>0</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.7</td>\n",
       "      <td>10.0</td>\n",
       "      <td>41.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>50.0</td>\n",
       "      <td>621.0</td>\n",
       "      <td>525.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>27</td>\n",
       "      <td>0.0</td>\n",
       "      <td>437.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>153.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>11</td>\n",
       "      <td>0</td>\n",
       "      <td>1.6</td>\n",
       "      <td>1.2</td>\n",
       "      <td>1.16</td>\n",
       "      <td>1.26</td>\n",
       "      <td>51.2</td>\n",
       "      <td>8.2</td>\n",
       "      <td>3.8</td>\n",
       "      <td>15.72</td>\n",
       "      <td>1.16</td>\n",
       "      <td>4.400000e-01</td>\n",
       "      <td>3.8</td>\n",
       "      <td>2.6</td>\n",
       "      <td>68.50</td>\n",
       "      <td>36.0</td>\n",
       "      <td>4.2</td>\n",
       "      <td>1.04</td>\n",
       "      <td>-0.16</td>\n",
       "      <td>1.2</td>\n",
       "      <td>0.80</td>\n",
       "      <td>0.88</td>\n",
       "      <td>6.0</td>\n",
       "      <td>29.0</td>\n",
       "      <td>7.4</td>\n",
       "      <td>1.6</td>\n",
       "      <td>36.6</td>\n",
       "      <td>608.0</td>\n",
       "      <td>511.4</td>\n",
       "      <td>4.2</td>\n",
       "      <td>3.8</td>\n",
       "      <td>20.8</td>\n",
       "      <td>0.6</td>\n",
       "      <td>317.8</td>\n",
       "      <td>18.2</td>\n",
       "      <td>123.2</td>\n",
       "      <td>8.6</td>\n",
       "      <td>11.6</td>\n",
       "      <td>2.8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>2020-11-01</td>\n",
       "      <td>16:30</td>\n",
       "      <td>Matchweek 7</td>\n",
       "      <td>Sun</td>\n",
       "      <td>Away</td>\n",
       "      <td>W</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Manchester Utd</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.4</td>\n",
       "      <td>46.0</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>6.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>24.6</td>\n",
       "      <td>0.3</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>43.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>0.4</td>\n",
       "      <td>5.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>40.0</td>\n",
       "      <td>526.0</td>\n",
       "      <td>438.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>32</td>\n",
       "      <td>0.0</td>\n",
       "      <td>298.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>130.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>15</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.92</td>\n",
       "      <td>1.38</td>\n",
       "      <td>51.6</td>\n",
       "      <td>8.0</td>\n",
       "      <td>3.6</td>\n",
       "      <td>15.90</td>\n",
       "      <td>0.92</td>\n",
       "      <td>8.000000e-02</td>\n",
       "      <td>3.8</td>\n",
       "      <td>2.4</td>\n",
       "      <td>58.50</td>\n",
       "      <td>35.0</td>\n",
       "      <td>4.8</td>\n",
       "      <td>1.08</td>\n",
       "      <td>-0.32</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.76</td>\n",
       "      <td>0.82</td>\n",
       "      <td>6.4</td>\n",
       "      <td>29.6</td>\n",
       "      <td>7.4</td>\n",
       "      <td>2.4</td>\n",
       "      <td>37.2</td>\n",
       "      <td>598.6</td>\n",
       "      <td>498.6</td>\n",
       "      <td>5.4</td>\n",
       "      <td>3.8</td>\n",
       "      <td>20.2</td>\n",
       "      <td>0.6</td>\n",
       "      <td>336.6</td>\n",
       "      <td>17.2</td>\n",
       "      <td>124.8</td>\n",
       "      <td>8.8</td>\n",
       "      <td>11.0</td>\n",
       "      <td>2.6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>2020-11-08</td>\n",
       "      <td>19:15</td>\n",
       "      <td>Matchweek 8</td>\n",
       "      <td>Sun</td>\n",
       "      <td>Home</td>\n",
       "      <td>L</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>Aston Villa</td>\n",
       "      <td>1.5</td>\n",
       "      <td>1.8</td>\n",
       "      <td>59.0</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>13.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>16.4</td>\n",
       "      <td>1.5</td>\n",
       "      <td>-1.5</td>\n",
       "      <td>5.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>40.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.6</td>\n",
       "      <td>-0.4</td>\n",
       "      <td>0</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.6</td>\n",
       "      <td>8.0</td>\n",
       "      <td>30.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>43.0</td>\n",
       "      <td>588.0</td>\n",
       "      <td>481.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>15</td>\n",
       "      <td>0.0</td>\n",
       "      <td>288.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>153.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.8</td>\n",
       "      <td>1.2</td>\n",
       "      <td>0.90</td>\n",
       "      <td>1.06</td>\n",
       "      <td>48.4</td>\n",
       "      <td>7.8</td>\n",
       "      <td>3.2</td>\n",
       "      <td>17.86</td>\n",
       "      <td>0.76</td>\n",
       "      <td>-1.000000e-01</td>\n",
       "      <td>3.8</td>\n",
       "      <td>2.6</td>\n",
       "      <td>68.50</td>\n",
       "      <td>37.2</td>\n",
       "      <td>3.6</td>\n",
       "      <td>0.96</td>\n",
       "      <td>-0.24</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.58</td>\n",
       "      <td>0.58</td>\n",
       "      <td>6.2</td>\n",
       "      <td>26.8</td>\n",
       "      <td>5.6</td>\n",
       "      <td>2.2</td>\n",
       "      <td>34.8</td>\n",
       "      <td>561.8</td>\n",
       "      <td>466.4</td>\n",
       "      <td>4.6</td>\n",
       "      <td>3.0</td>\n",
       "      <td>22.8</td>\n",
       "      <td>0.4</td>\n",
       "      <td>311.6</td>\n",
       "      <td>14.4</td>\n",
       "      <td>117.6</td>\n",
       "      <td>9.0</td>\n",
       "      <td>10.8</td>\n",
       "      <td>1.6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>2020-11-22</td>\n",
       "      <td>16:30</td>\n",
       "      <td>Matchweek 9</td>\n",
       "      <td>Sun</td>\n",
       "      <td>Away</td>\n",
       "      <td>D</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Leeds United</td>\n",
       "      <td>0.9</td>\n",
       "      <td>1.8</td>\n",
       "      <td>34.0</td>\n",
       "      <td>4-2-3-1</td>\n",
       "      <td>9.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>19.6</td>\n",
       "      <td>0.9</td>\n",
       "      <td>-0.9</td>\n",
       "      <td>4.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0</td>\n",
       "      <td>0.7</td>\n",
       "      <td>0.5</td>\n",
       "      <td>6.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>29.0</td>\n",
       "      <td>343.0</td>\n",
       "      <td>236.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>21</td>\n",
       "      <td>0.0</td>\n",
       "      <td>184.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>69.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>0.6</td>\n",
       "      <td>1.2</td>\n",
       "      <td>0.94</td>\n",
       "      <td>0.88</td>\n",
       "      <td>53.4</td>\n",
       "      <td>9.6</td>\n",
       "      <td>3.0</td>\n",
       "      <td>18.14</td>\n",
       "      <td>0.80</td>\n",
       "      <td>-3.400000e-01</td>\n",
       "      <td>3.2</td>\n",
       "      <td>2.2</td>\n",
       "      <td>64.00</td>\n",
       "      <td>35.2</td>\n",
       "      <td>2.6</td>\n",
       "      <td>0.76</td>\n",
       "      <td>-0.24</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.54</td>\n",
       "      <td>0.58</td>\n",
       "      <td>7.2</td>\n",
       "      <td>30.6</td>\n",
       "      <td>6.2</td>\n",
       "      <td>2.4</td>\n",
       "      <td>40.8</td>\n",
       "      <td>594.4</td>\n",
       "      <td>500.2</td>\n",
       "      <td>4.6</td>\n",
       "      <td>3.8</td>\n",
       "      <td>22.2</td>\n",
       "      <td>0.2</td>\n",
       "      <td>334.4</td>\n",
       "      <td>17.4</td>\n",
       "      <td>139.4</td>\n",
       "      <td>9.8</td>\n",
       "      <td>11.2</td>\n",
       "      <td>0.8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>2020-11-29</td>\n",
       "      <td>19:15</td>\n",
       "      <td>Matchweek 10</td>\n",
       "      <td>Sun</td>\n",
       "      <td>Home</td>\n",
       "      <td>L</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>1.5</td>\n",
       "      <td>1.8</td>\n",
       "      <td>56.0</td>\n",
       "      <td>4-2-3-1</td>\n",
       "      <td>13.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>15.6</td>\n",
       "      <td>1.5</td>\n",
       "      <td>-0.5</td>\n",
       "      <td>5.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>60.0</td>\n",
       "      <td>17.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1.8</td>\n",
       "      <td>-0.2</td>\n",
       "      <td>1</td>\n",
       "      <td>0.8</td>\n",
       "      <td>1.6</td>\n",
       "      <td>6.0</td>\n",
       "      <td>49.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>63.0</td>\n",
       "      <td>557.0</td>\n",
       "      <td>447.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>22</td>\n",
       "      <td>0.0</td>\n",
       "      <td>307.0</td>\n",
       "      <td>35.0</td>\n",
       "      <td>219.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>23</td>\n",
       "      <td>0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.00</td>\n",
       "      <td>1.20</td>\n",
       "      <td>47.4</td>\n",
       "      <td>10.2</td>\n",
       "      <td>2.4</td>\n",
       "      <td>18.80</td>\n",
       "      <td>0.86</td>\n",
       "      <td>-8.000000e-01</td>\n",
       "      <td>3.6</td>\n",
       "      <td>2.8</td>\n",
       "      <td>74.00</td>\n",
       "      <td>32.8</td>\n",
       "      <td>2.6</td>\n",
       "      <td>0.84</td>\n",
       "      <td>0.04</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.56</td>\n",
       "      <td>0.56</td>\n",
       "      <td>7.2</td>\n",
       "      <td>25.4</td>\n",
       "      <td>6.0</td>\n",
       "      <td>2.2</td>\n",
       "      <td>36.8</td>\n",
       "      <td>514.8</td>\n",
       "      <td>418.2</td>\n",
       "      <td>4.8</td>\n",
       "      <td>4.4</td>\n",
       "      <td>22.8</td>\n",
       "      <td>0.0</td>\n",
       "      <td>289.2</td>\n",
       "      <td>17.2</td>\n",
       "      <td>119.4</td>\n",
       "      <td>10.8</td>\n",
       "      <td>11.4</td>\n",
       "      <td>1.2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2026</th>\n",
       "      <td>2023-12-17</td>\n",
       "      <td>14:00</td>\n",
       "      <td>Matchweek 17</td>\n",
       "      <td>Sun</td>\n",
       "      <td>Away</td>\n",
       "      <td>L</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>West Ham</td>\n",
       "      <td>0.7</td>\n",
       "      <td>1.1</td>\n",
       "      <td>55.0</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>14.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>20.5</td>\n",
       "      <td>0.7</td>\n",
       "      <td>-0.7</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>22.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.5</td>\n",
       "      <td>-1.5</td>\n",
       "      <td>0</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.8</td>\n",
       "      <td>12.0</td>\n",
       "      <td>33.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>49.0</td>\n",
       "      <td>579.0</td>\n",
       "      <td>474.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>27</td>\n",
       "      <td>0.0</td>\n",
       "      <td>300.0</td>\n",
       "      <td>26.0</td>\n",
       "      <td>186.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2023</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>22</td>\n",
       "      <td>23</td>\n",
       "      <td>1.4</td>\n",
       "      <td>1.4</td>\n",
       "      <td>1.24</td>\n",
       "      <td>1.72</td>\n",
       "      <td>48.8</td>\n",
       "      <td>9.8</td>\n",
       "      <td>4.0</td>\n",
       "      <td>15.38</td>\n",
       "      <td>1.08</td>\n",
       "      <td>1.600000e-01</td>\n",
       "      <td>3.8</td>\n",
       "      <td>2.4</td>\n",
       "      <td>70.00</td>\n",
       "      <td>22.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.40</td>\n",
       "      <td>0.20</td>\n",
       "      <td>1.2</td>\n",
       "      <td>0.88</td>\n",
       "      <td>0.72</td>\n",
       "      <td>7.2</td>\n",
       "      <td>26.6</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1.2</td>\n",
       "      <td>28.6</td>\n",
       "      <td>535.2</td>\n",
       "      <td>426.8</td>\n",
       "      <td>4.4</td>\n",
       "      <td>3.2</td>\n",
       "      <td>22.4</td>\n",
       "      <td>0.8</td>\n",
       "      <td>293.0</td>\n",
       "      <td>17.0</td>\n",
       "      <td>122.2</td>\n",
       "      <td>12.4</td>\n",
       "      <td>12.4</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2027</th>\n",
       "      <td>2023-12-24</td>\n",
       "      <td>13:00</td>\n",
       "      <td>Matchweek 18</td>\n",
       "      <td>Sun</td>\n",
       "      <td>Home</td>\n",
       "      <td>W</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Chelsea</td>\n",
       "      <td>1.3</td>\n",
       "      <td>2.0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>14.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>16.7</td>\n",
       "      <td>1.3</td>\n",
       "      <td>0.7</td>\n",
       "      <td>4.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>75.0</td>\n",
       "      <td>20.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.4</td>\n",
       "      <td>1</td>\n",
       "      <td>0.7</td>\n",
       "      <td>0.5</td>\n",
       "      <td>8.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>309.0</td>\n",
       "      <td>200.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>34</td>\n",
       "      <td>2.0</td>\n",
       "      <td>174.0</td>\n",
       "      <td>18.0</td>\n",
       "      <td>108.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>2023</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>23</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.8</td>\n",
       "      <td>0.96</td>\n",
       "      <td>1.80</td>\n",
       "      <td>51.4</td>\n",
       "      <td>9.2</td>\n",
       "      <td>3.8</td>\n",
       "      <td>15.78</td>\n",
       "      <td>0.80</td>\n",
       "      <td>4.000000e-02</td>\n",
       "      <td>4.2</td>\n",
       "      <td>2.4</td>\n",
       "      <td>65.00</td>\n",
       "      <td>22.4</td>\n",
       "      <td>4.2</td>\n",
       "      <td>1.42</td>\n",
       "      <td>-0.18</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.68</td>\n",
       "      <td>0.70</td>\n",
       "      <td>7.4</td>\n",
       "      <td>27.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1.6</td>\n",
       "      <td>31.0</td>\n",
       "      <td>554.4</td>\n",
       "      <td>443.6</td>\n",
       "      <td>3.2</td>\n",
       "      <td>3.6</td>\n",
       "      <td>21.2</td>\n",
       "      <td>0.8</td>\n",
       "      <td>304.4</td>\n",
       "      <td>17.8</td>\n",
       "      <td>125.0</td>\n",
       "      <td>11.6</td>\n",
       "      <td>11.6</td>\n",
       "      <td>1.6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2028</th>\n",
       "      <td>2023-12-27</td>\n",
       "      <td>19:30</td>\n",
       "      <td>Matchweek 19</td>\n",
       "      <td>Wed</td>\n",
       "      <td>Away</td>\n",
       "      <td>W</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Brentford</td>\n",
       "      <td>2.4</td>\n",
       "      <td>2.2</td>\n",
       "      <td>40.0</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>11.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>17.8</td>\n",
       "      <td>2.4</td>\n",
       "      <td>1.6</td>\n",
       "      <td>4.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>75.0</td>\n",
       "      <td>29.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>2.2</td>\n",
       "      <td>1.2</td>\n",
       "      <td>3</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.8</td>\n",
       "      <td>9.0</td>\n",
       "      <td>29.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>26.0</td>\n",
       "      <td>436.0</td>\n",
       "      <td>319.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>34</td>\n",
       "      <td>0.0</td>\n",
       "      <td>271.0</td>\n",
       "      <td>19.0</td>\n",
       "      <td>103.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>2023</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>23</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.94</td>\n",
       "      <td>1.62</td>\n",
       "      <td>49.8</td>\n",
       "      <td>10.2</td>\n",
       "      <td>4.0</td>\n",
       "      <td>15.80</td>\n",
       "      <td>0.94</td>\n",
       "      <td>6.000000e-02</td>\n",
       "      <td>3.8</td>\n",
       "      <td>2.4</td>\n",
       "      <td>63.34</td>\n",
       "      <td>21.0</td>\n",
       "      <td>3.8</td>\n",
       "      <td>1.02</td>\n",
       "      <td>-0.18</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.72</td>\n",
       "      <td>0.68</td>\n",
       "      <td>7.4</td>\n",
       "      <td>26.6</td>\n",
       "      <td>5.6</td>\n",
       "      <td>1.8</td>\n",
       "      <td>32.6</td>\n",
       "      <td>523.8</td>\n",
       "      <td>414.0</td>\n",
       "      <td>4.8</td>\n",
       "      <td>4.6</td>\n",
       "      <td>25.0</td>\n",
       "      <td>1.2</td>\n",
       "      <td>294.8</td>\n",
       "      <td>17.0</td>\n",
       "      <td>131.0</td>\n",
       "      <td>11.2</td>\n",
       "      <td>10.8</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2029</th>\n",
       "      <td>2023-12-30</td>\n",
       "      <td>15:00</td>\n",
       "      <td>Matchweek 20</td>\n",
       "      <td>Sat</td>\n",
       "      <td>Home</td>\n",
       "      <td>W</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Everton</td>\n",
       "      <td>1.9</td>\n",
       "      <td>0.4</td>\n",
       "      <td>53.0</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>12.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>13.4</td>\n",
       "      <td>1.9</td>\n",
       "      <td>1.1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>16.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2</td>\n",
       "      <td>1.5</td>\n",
       "      <td>1.4</td>\n",
       "      <td>9.0</td>\n",
       "      <td>34.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>44.0</td>\n",
       "      <td>556.0</td>\n",
       "      <td>441.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>28</td>\n",
       "      <td>0.0</td>\n",
       "      <td>283.0</td>\n",
       "      <td>27.0</td>\n",
       "      <td>139.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>2023</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "      <td>23</td>\n",
       "      <td>1.6</td>\n",
       "      <td>1.2</td>\n",
       "      <td>1.28</td>\n",
       "      <td>1.50</td>\n",
       "      <td>49.2</td>\n",
       "      <td>11.2</td>\n",
       "      <td>4.4</td>\n",
       "      <td>17.20</td>\n",
       "      <td>1.28</td>\n",
       "      <td>3.200000e-01</td>\n",
       "      <td>3.4</td>\n",
       "      <td>2.2</td>\n",
       "      <td>65.00</td>\n",
       "      <td>22.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.14</td>\n",
       "      <td>-0.06</td>\n",
       "      <td>1.2</td>\n",
       "      <td>0.88</td>\n",
       "      <td>0.72</td>\n",
       "      <td>8.4</td>\n",
       "      <td>27.0</td>\n",
       "      <td>6.4</td>\n",
       "      <td>1.8</td>\n",
       "      <td>34.4</td>\n",
       "      <td>511.0</td>\n",
       "      <td>398.2</td>\n",
       "      <td>5.2</td>\n",
       "      <td>4.4</td>\n",
       "      <td>26.6</td>\n",
       "      <td>0.8</td>\n",
       "      <td>300.6</td>\n",
       "      <td>18.4</td>\n",
       "      <td>133.8</td>\n",
       "      <td>12.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>2.6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2035</th>\n",
       "      <td>2024-02-25</td>\n",
       "      <td>13:30</td>\n",
       "      <td>Matchweek 26</td>\n",
       "      <td>Sun</td>\n",
       "      <td>Home</td>\n",
       "      <td>W</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Sheffield Utd</td>\n",
       "      <td>1.2</td>\n",
       "      <td>0.6</td>\n",
       "      <td>63.0</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>13.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>14.4</td>\n",
       "      <td>1.2</td>\n",
       "      <td>-0.2</td>\n",
       "      <td>4.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>17.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.4</td>\n",
       "      <td>1</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.8</td>\n",
       "      <td>9.0</td>\n",
       "      <td>36.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>41.0</td>\n",
       "      <td>645.0</td>\n",
       "      <td>545.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>27</td>\n",
       "      <td>0.0</td>\n",
       "      <td>359.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>166.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2023</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>18</td>\n",
       "      <td>23</td>\n",
       "      <td>1.8</td>\n",
       "      <td>1.8</td>\n",
       "      <td>1.60</td>\n",
       "      <td>1.90</td>\n",
       "      <td>43.4</td>\n",
       "      <td>13.0</td>\n",
       "      <td>5.2</td>\n",
       "      <td>16.98</td>\n",
       "      <td>1.28</td>\n",
       "      <td>1.110223e-17</td>\n",
       "      <td>5.0</td>\n",
       "      <td>3.2</td>\n",
       "      <td>68.34</td>\n",
       "      <td>21.0</td>\n",
       "      <td>4.8</td>\n",
       "      <td>1.96</td>\n",
       "      <td>0.16</td>\n",
       "      <td>1.2</td>\n",
       "      <td>0.94</td>\n",
       "      <td>0.68</td>\n",
       "      <td>9.6</td>\n",
       "      <td>24.8</td>\n",
       "      <td>6.4</td>\n",
       "      <td>1.2</td>\n",
       "      <td>33.8</td>\n",
       "      <td>464.4</td>\n",
       "      <td>371.4</td>\n",
       "      <td>3.6</td>\n",
       "      <td>4.2</td>\n",
       "      <td>28.0</td>\n",
       "      <td>0.4</td>\n",
       "      <td>238.2</td>\n",
       "      <td>22.0</td>\n",
       "      <td>147.8</td>\n",
       "      <td>11.6</td>\n",
       "      <td>10.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1694 rows × 88 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           date   time         round  day venue result   gf   ga  \\\n",
       "5    2020-10-25  19:15   Matchweek 6  Sun  Home      L  0.0  1.0   \n",
       "6    2020-11-01  16:30   Matchweek 7  Sun  Away      W  1.0  0.0   \n",
       "7    2020-11-08  19:15   Matchweek 8  Sun  Home      L  0.0  3.0   \n",
       "8    2020-11-22  16:30   Matchweek 9  Sun  Away      D  0.0  0.0   \n",
       "9    2020-11-29  19:15  Matchweek 10  Sun  Home      L  1.0  2.0   \n",
       "...         ...    ...           ...  ...   ...    ...  ...  ...   \n",
       "2026 2023-12-17  14:00  Matchweek 17  Sun  Away      L  0.0  3.0   \n",
       "2027 2023-12-24  13:00  Matchweek 18  Sun  Home      W  2.0  1.0   \n",
       "2028 2023-12-27  19:30  Matchweek 19  Wed  Away      W  4.0  1.0   \n",
       "2029 2023-12-30  15:00  Matchweek 20  Sat  Home      W  3.0  0.0   \n",
       "2035 2024-02-25  13:30  Matchweek 26  Sun  Home      W  1.0  0.0   \n",
       "\n",
       "            opponent  xg_x  xga  poss formation    sh  sot  dist  npxg  g-xg  \\\n",
       "5     Leicester City   0.7  0.7  56.0     3-4-3  12.0  4.0  15.0   0.7  -0.7   \n",
       "6     Manchester Utd   1.0  0.4  46.0     3-4-3   6.0  1.0  24.6   0.3   0.0   \n",
       "7        Aston Villa   1.5  1.8  59.0     3-4-3  13.0  2.0  16.4   1.5  -1.5   \n",
       "8       Leeds United   0.9  1.8  34.0   4-2-3-1   9.0  2.0  19.6   0.9  -0.9   \n",
       "9             Wolves   1.5  1.8  56.0   4-2-3-1  13.0  2.0  15.6   1.5  -0.5   \n",
       "...              ...   ...  ...   ...       ...   ...  ...   ...   ...   ...   \n",
       "2026        West Ham   0.7  1.1  55.0     3-4-3  14.0  3.0  20.5   0.7  -0.7   \n",
       "2027         Chelsea   1.3  2.0  32.0     3-4-3  14.0  6.0  16.7   1.3   0.7   \n",
       "2028       Brentford   2.4  2.2  40.0     3-4-3  11.0  5.0  17.8   2.4   1.6   \n",
       "2029         Everton   1.9  0.4  53.0     3-4-3  12.0  6.0  13.4   1.9   1.1   \n",
       "2035   Sheffield Utd   1.2  0.6  63.0     3-4-3  13.0  2.0  14.4   1.2  -0.2   \n",
       "\n",
       "      sota  saves  save%  att (gk)  thr  psxg  psxg+/-  ast  xag   xa    kp  \\\n",
       "5      2.0    1.0   50.0      24.0  6.0   0.3     -0.7    0  0.6  0.7  10.0   \n",
       "6      2.0    2.0  100.0      43.0  0.0   0.1      0.1    0  0.2  0.4   5.0   \n",
       "7      5.0    3.0   40.0      24.0  1.0   1.6     -0.4    0  0.6  0.6   8.0   \n",
       "8      4.0    4.0  100.0      25.0  2.0   0.8      0.8    0  0.7  0.5   6.0   \n",
       "9      5.0    3.0   60.0      17.0  5.0   1.8     -0.2    1  0.8  1.6   6.0   \n",
       "...    ...    ...    ...       ...  ...   ...      ...  ...  ...  ...   ...   \n",
       "2026   4.0    1.0   25.0      22.0  4.0   1.5     -1.5    0  0.6  0.8  12.0   \n",
       "2027   4.0    3.0   75.0      20.0  5.0   1.4      0.4    1  0.7  0.5   8.0   \n",
       "2028   4.0    3.0   75.0      29.0  3.0   2.2      1.2    3  1.4  0.8   9.0   \n",
       "2029   0.0    0.0    NaN      16.0  4.0   0.0      0.0    2  1.5  1.4   9.0   \n",
       "2035   4.0    4.0  100.0      17.0  7.0   0.4      0.4    1  0.8  0.8   9.0   \n",
       "\n",
       "       1/3   ppa  crspa  prgp  attempted passes  completed passes  \\\n",
       "5     41.0  11.0    5.0  50.0             621.0             525.0   \n",
       "6     24.0   7.0    2.0  40.0             526.0             438.0   \n",
       "7     30.0   4.0    1.0  43.0             588.0             481.0   \n",
       "8     15.0   3.0    0.0  29.0             343.0             236.0   \n",
       "9     49.0  15.0    2.0  63.0             557.0             447.0   \n",
       "...    ...   ...    ...   ...               ...               ...   \n",
       "2026  33.0   7.0    3.0  49.0             579.0             474.0   \n",
       "2027   9.0   6.0    1.0  25.0             309.0             200.0   \n",
       "2028  29.0   8.0    1.0  26.0             436.0             319.0   \n",
       "2029  34.0  14.0    2.0  44.0             556.0             441.0   \n",
       "2035  36.0   9.0    4.0  41.0             645.0             545.0   \n",
       "\n",
       "      corner kicks  shots blocked  tkl+int  err  mid 3rd  att pen  att 3rd  \\\n",
       "5              9.0            1.0       27  0.0    437.0     21.0    153.0   \n",
       "6              3.0            2.0       32  0.0    298.0     14.0    130.0   \n",
       "7              3.0           10.0       15  0.0    288.0     25.0    153.0   \n",
       "8              3.0            5.0       21  0.0    184.0     13.0     69.0   \n",
       "9              8.0            4.0       22  0.0    307.0     35.0    219.0   \n",
       "...            ...            ...      ...  ...      ...      ...      ...   \n",
       "2026           5.0            4.0       27  0.0    300.0     26.0    186.0   \n",
       "2027          10.0            8.0       34  2.0    174.0     18.0    108.0   \n",
       "2028           2.0            4.0       34  0.0    271.0     19.0    103.0   \n",
       "2029           5.0            4.0       28  0.0    283.0     27.0    139.0   \n",
       "2035           5.0            4.0       27  0.0    359.0     24.0    166.0   \n",
       "\n",
       "       fls   fld  off  season                     team  target  venue_code  \\\n",
       "5     13.0   9.0  1.0    2021                  Arsenal       0           1   \n",
       "6     12.0  11.0  0.0    2021                  Arsenal       1           0   \n",
       "7     11.0  13.0  1.0    2021                  Arsenal       0           1   \n",
       "8      8.0   9.0  2.0    2021                  Arsenal       0           0   \n",
       "9     13.0  12.0  2.0    2021                  Arsenal       0           1   \n",
       "...    ...   ...  ...     ...                      ...     ...         ...   \n",
       "2026  15.0  11.0  2.0    2023  Wolverhampton Wanderers       0           0   \n",
       "2027  11.0   8.0  3.0    2023  Wolverhampton Wanderers       1           1   \n",
       "2028  14.0   5.0  3.0    2023  Wolverhampton Wanderers       1           0   \n",
       "2029   9.0  14.0  3.0    2023  Wolverhampton Wanderers       1           1   \n",
       "2035   8.0   4.0  1.0    2023  Wolverhampton Wanderers       1           1   \n",
       "\n",
       "      opp_code  team_code  gf_rolling  ga_rolling  xg_x_rolling  xga_rolling  \\\n",
       "5           11          0         1.6         1.2          1.16         1.26   \n",
       "6           15          0         1.0         1.4          0.92         1.38   \n",
       "7            1          0         0.8         1.2          0.90         1.06   \n",
       "8           10          0         0.6         1.2          0.94         0.88   \n",
       "9           23          0         0.2         1.0          1.00         1.20   \n",
       "...        ...        ...         ...         ...           ...          ...   \n",
       "2026        22         23         1.4         1.4          1.24         1.72   \n",
       "2027         6         23         1.0         1.8          0.96         1.80   \n",
       "2028         3         23         1.0         1.4          0.94         1.62   \n",
       "2029         8         23         1.6         1.2          1.28         1.50   \n",
       "2035        18         23         1.8         1.8          1.60         1.90   \n",
       "\n",
       "      poss_rolling  sh_rolling  sot_rolling  dist_rolling  npxg_rolling  \\\n",
       "5             51.2         8.2          3.8         15.72          1.16   \n",
       "6             51.6         8.0          3.6         15.90          0.92   \n",
       "7             48.4         7.8          3.2         17.86          0.76   \n",
       "8             53.4         9.6          3.0         18.14          0.80   \n",
       "9             47.4        10.2          2.4         18.80          0.86   \n",
       "...            ...         ...          ...           ...           ...   \n",
       "2026          48.8         9.8          4.0         15.38          1.08   \n",
       "2027          51.4         9.2          3.8         15.78          0.80   \n",
       "2028          49.8        10.2          4.0         15.80          0.94   \n",
       "2029          49.2        11.2          4.4         17.20          1.28   \n",
       "2035          43.4        13.0          5.2         16.98          1.28   \n",
       "\n",
       "      g-xg_rolling  sota_rolling  saves_rolling  save%_rolling  \\\n",
       "5     4.400000e-01           3.8            2.6          68.50   \n",
       "6     8.000000e-02           3.8            2.4          58.50   \n",
       "7    -1.000000e-01           3.8            2.6          68.50   \n",
       "8    -3.400000e-01           3.2            2.2          64.00   \n",
       "9    -8.000000e-01           3.6            2.8          74.00   \n",
       "...            ...           ...            ...            ...   \n",
       "2026  1.600000e-01           3.8            2.4          70.00   \n",
       "2027  4.000000e-02           4.2            2.4          65.00   \n",
       "2028  6.000000e-02           3.8            2.4          63.34   \n",
       "2029  3.200000e-01           3.4            2.2          65.00   \n",
       "2035  1.110223e-17           5.0            3.2          68.34   \n",
       "\n",
       "      att (gk)_rolling  thr_rolling  psxg_rolling  psxg+/-_rolling  \\\n",
       "5                 36.0          4.2          1.04            -0.16   \n",
       "6                 35.0          4.8          1.08            -0.32   \n",
       "7                 37.2          3.6          0.96            -0.24   \n",
       "8                 35.2          2.6          0.76            -0.24   \n",
       "9                 32.8          2.6          0.84             0.04   \n",
       "...                ...          ...           ...              ...   \n",
       "2026              22.0          4.0          1.40             0.20   \n",
       "2027              22.4          4.2          1.42            -0.18   \n",
       "2028              21.0          3.8          1.02            -0.18   \n",
       "2029              22.0          4.0          1.14            -0.06   \n",
       "2035              21.0          4.8          1.96             0.16   \n",
       "\n",
       "      ast_rolling  xag_rolling  xa_rolling  kp_rolling  1/3_rolling  \\\n",
       "5             1.2         0.80        0.88         6.0         29.0   \n",
       "6             0.8         0.76        0.82         6.4         29.6   \n",
       "7             0.4         0.58        0.58         6.2         26.8   \n",
       "8             0.4         0.54        0.58         7.2         30.6   \n",
       "9             0.0         0.56        0.56         7.2         25.4   \n",
       "...           ...          ...         ...         ...          ...   \n",
       "2026          1.2         0.88        0.72         7.2         26.6   \n",
       "2027          0.8         0.68        0.70         7.4         27.0   \n",
       "2028          0.8         0.72        0.68         7.4         26.6   \n",
       "2029          1.2         0.88        0.72         8.4         27.0   \n",
       "2035          1.2         0.94        0.68         9.6         24.8   \n",
       "\n",
       "      ppa_rolling  crspa_rolling  prgp_rolling  attempted passes_rolling  \\\n",
       "5             7.4            1.6          36.6                     608.0   \n",
       "6             7.4            2.4          37.2                     598.6   \n",
       "7             5.6            2.2          34.8                     561.8   \n",
       "8             6.2            2.4          40.8                     594.4   \n",
       "9             6.0            2.2          36.8                     514.8   \n",
       "...           ...            ...           ...                       ...   \n",
       "2026          5.0            1.2          28.6                     535.2   \n",
       "2027          5.0            1.6          31.0                     554.4   \n",
       "2028          5.6            1.8          32.6                     523.8   \n",
       "2029          6.4            1.8          34.4                     511.0   \n",
       "2035          6.4            1.2          33.8                     464.4   \n",
       "\n",
       "      completed passes_rolling  corner kicks_rolling  shots blocked_rolling  \\\n",
       "5                        511.4                   4.2                    3.8   \n",
       "6                        498.6                   5.4                    3.8   \n",
       "7                        466.4                   4.6                    3.0   \n",
       "8                        500.2                   4.6                    3.8   \n",
       "9                        418.2                   4.8                    4.4   \n",
       "...                        ...                   ...                    ...   \n",
       "2026                     426.8                   4.4                    3.2   \n",
       "2027                     443.6                   3.2                    3.6   \n",
       "2028                     414.0                   4.8                    4.6   \n",
       "2029                     398.2                   5.2                    4.4   \n",
       "2035                     371.4                   3.6                    4.2   \n",
       "\n",
       "      tkl+int_rolling  err_rolling  mid 3rd_rolling  att pen_rolling  \\\n",
       "5                20.8          0.6            317.8             18.2   \n",
       "6                20.2          0.6            336.6             17.2   \n",
       "7                22.8          0.4            311.6             14.4   \n",
       "8                22.2          0.2            334.4             17.4   \n",
       "9                22.8          0.0            289.2             17.2   \n",
       "...               ...          ...              ...              ...   \n",
       "2026             22.4          0.8            293.0             17.0   \n",
       "2027             21.2          0.8            304.4             17.8   \n",
       "2028             25.0          1.2            294.8             17.0   \n",
       "2029             26.6          0.8            300.6             18.4   \n",
       "2035             28.0          0.4            238.2             22.0   \n",
       "\n",
       "      att 3rd_rolling  fls_rolling  fld_rolling  off_rolling  \n",
       "5               123.2          8.6         11.6          2.8  \n",
       "6               124.8          8.8         11.0          2.6  \n",
       "7               117.6          9.0         10.8          1.6  \n",
       "8               139.4          9.8         11.2          0.8  \n",
       "9               119.4         10.8         11.4          1.2  \n",
       "...               ...          ...          ...          ...  \n",
       "2026            122.2         12.4         12.4          2.0  \n",
       "2027            125.0         11.6         11.6          1.6  \n",
       "2028            131.0         11.2         10.8          2.0  \n",
       "2029            133.8         12.0         10.0          2.6  \n",
       "2035            147.8         11.6         10.0          1.0  \n",
       "\n",
       "[1694 rows x 88 columns]"
      ]
     },
     "execution_count": 2067,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "matches_rolling"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.3 Feature selection"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Feature selection is an important step in the machine learning pipeline. It involves identifying the most relevant variables to use in model construction. This process helps in enhancing the model's performance by reducing overfitting, improving accuracy, and decreasing training time.\n",
    "\n",
    "Filter methods assess the relevance of features by looking at their statistical properties. In this section, we will focus on selecting features based on their correlation with the target variable (in our case, the binary outcome of a match). By examining these correlations, we can identify which features have the most significant influence on match outcomes and should therefore be included in our model.\n",
    "\n",
    "We'll begin by visualizing the distribution of correlation values between each feature and the target variable using a cumulative distribution plot. This approach will help us to determine an appropriate threshold for selecting features. Features with a correlation above this threshold are likely to be more predictive and will be chosen for our model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2068,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA04AAAIjCAYAAAA0vUuxAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/SrBM8AAAACXBIWXMAAA9hAAAPYQGoP6dpAABoxElEQVR4nO3deXwT1frH8W9Ik7YBSstaQCyrgmwiFazIolQLIoiKInJlUcAFRAQFUaSAV0BRFr0qXhdw4wIiFxeURTYRKsjmCgjIosim7LS0aXp+f/Aj15i2aUrSdPm8X6++JGfOzDzzZBrzdM6csRhjjAAAAAAAOSoV6gAAAAAAoLCjcAIAAAAAHyicAAAAAMAHCicAAAAA8IHCCQAAAAB8oHACAAAAAB8onAAAAADABwonAAAAAPCBwgkAAAAAfKBwAlCo9OnTRzVr1gzoNmfOnCmLxaI9e/YEdLvBMGbMGFkslgLZV7t27dSuXTv365UrV8pisWjevHkFsv9gvNeBdvr0afXr10+xsbGyWCwaMmRIqEMKqJo1a6pPnz4B3WZReF//yp/fufN9//jjjyBHBaAwonACiqFdu3bpvvvuU+3atRUREaGoqCi1atVK06ZNU1paWqjDC5rx48drwYIFoQ7D7XzBdv4nIiJC1apVU1JSkl588UWdOnUqIPv5/fffNWbMGG3ZsiUg2wukwhxbXowfP14zZ87UAw88oHfffVd33313rv1dLpdmzJihdu3aqXz58goPD1fNmjXVt29fbdiwoYCiDr6i/r76EujPkvN/lMjLT2Eza9YsTZ06NdRhAIWCxRhjQh0EgMBZuHChbr/9doWHh6tXr15q1KiRMjIy9NVXX+nDDz9Unz599O9//zvUYeaoT58+WrlyZb6uDpUpU0bdunXTzJkzPdpdLpecTqfCw8ML9IvJzJkz1bdvX40bN061atWS0+nUwYMHtXLlSi1dulQXX3yxPv74YzVp0sS9TmZmpjIzMxUREZHn/WzYsEFXXnmlZsyY4dfVg4yMDEmS3W6XdO7L3bXXXqsPPvhA3bp1y/N28hub0+lUVlaWwsPDA7KvYLjqqqsUFhamr776ymfftLQ03XrrrVq0aJHatGmjzp07q3z58tqzZ4/mzp2rn3/+Wfv27dNFF11UAJHnTc2aNdWuXTuv3xlfivr7+lfZ/c7l9FkyZswYjR07VkeOHFHFihXzvI9Dhw5p6dKlHm0jR45UmTJl9OSTT3q0/+Mf//D/IILopptu0g8//FAkrtgDwRYW6gAABM7u3bt15513Ki4uTsuXL1fVqlXdywYOHKidO3dq4cKFIYwwNKxWq6xWa8j237FjR8XHx7tfjxw5UsuXL9dNN92kLl26aOvWrYqMjJQkhYWFKSwsuB/Nqampcjgc7oIpVGw2W0j3nxeHDx/WZZddlqe+jz32mBYtWqQpU6Z4DelLTk7WlClTAhLTmTNnVLp06WyXnX9vQ6kovK9/VRC/c1WqVPEqiCZOnKiKFSsGpFDKzMxUVlZWyH+ngeKOoXpAMfLcc8/p9OnTevPNNz2KpvPq1q2rhx9+WJK0Z88eWSyWbP/SbLFYNGbMGPfr8+P6f/75Z/3jH/9QuXLlVKlSJT311FMyxujXX3/VzTffrKioKMXGxuqFF17w2F5O9xidH76ycuXKXI/r+eef19VXX60KFSooMjJSzZs397oPx2Kx6MyZM3r77bfdQ17O/yX87/u/6aabVLt27Wz3lZCQ4FHkSNJ7772n5s2bKzIyUuXLl9edd96pX3/9NdeYfbnuuuv01FNPae/evXrvvffc7dndb7F06VJdc801io6OVpkyZXTppZfqiSeekHQuh1deeaUkqW/fvu5jP/++tmvXTo0aNdLGjRvVpk0bORwO97p/v8fpPJfLpSeeeEKxsbEqXbq0unTp4nW8Od0b89dt+ootu3thzpw5o2HDhqlGjRoKDw/XpZdequeff15/HxxhsVg0aNAgLViwQI0aNVJ4eLgaNmyoRYsWZZ/wvzl8+LDuvfdeValSRREREWratKnefvtt9/Lz5+bu3bu1cOFCd+w5/dX9t99+02uvvabrr78+2/ugrFarHn30UY+rTZs3b1bHjh0VFRWlMmXKqH379vr666891jt/7q5atUoPPvigKleu7N5Gbu9tenq6kpOTVbduXYWHh6tGjRoaPny40tPTc83L0aNH9eijj6px48YqU6aMoqKi1LFjR3377bceuSls76sxRhUrVtTQoUPdbVlZWYqOjpbVatXx48fd7c8++6zCwsJ0+vRpSd6/c7l9lpx3/Phx9enTR9HR0SpXrpz69u2r1NTUXGP0JSMjQ6NHj1bz5s1Vrlw5lS5dWq1bt9aKFSs8+p3/7H7++ec1depU1alTR+Hh4frpp58knXt/4uPjFRERoTp16ui1117L8T4uX59t7dq108KFC7V37153LorS/WtAoHHFCShGPvnkE9WuXVtXX311ULbfvXt3NWjQQBMnTtTChQv1z3/+U+XLl9drr72m6667Ts8++6zef/99Pfroo7ryyivVpk2bgOx32rRp6tKli3r27KmMjAzNnj1bt99+uz799FN16tRJkvTuu++qX79+atGihQYMGCBJqlOnTo7H0atXL33zzTfuL4CStHfvXn399deaNGmSu+2ZZ57RU089pTvuuEP9+vXTkSNH9NJLL6lNmzbavHmzoqOj831cd999t5544gktWbJE/fv3z7bPjz/+qJtuuklNmjTRuHHjFB4erp07d2rNmjWSpAYNGmjcuHEaPXq0BgwYoNatW0uSxznw559/qmPHjrrzzjv1j3/8Q1WqVMk1rmeeeUYWi0UjRozQ4cOHNXXqVCUmJmrLli3uK2N5kZfY/soYoy5dumjFihW69957dfnll2vx4sV67LHHtH//fq8rNl999ZXmz5+vBx98UGXLltWLL76o2267Tfv27VOFChVyjCstLU3t2rXTzp07NWjQINWqVUsffPCB+vTpo+PHj+vhhx9WgwYN9O677+qRRx7RRRddpGHDhkmSKlWqlO02P//8c2VmZvq8B+q8H3/8Ua1bt1ZUVJSGDx8um82m1157Te3atdOqVavUsmVLj/4PPvigKlWqpNGjR+vMmTPu9uze26ysLHXp0kVfffWVBgwYoAYNGuj777/XlClT9PPPP+d6784vv/yiBQsW6Pbbb1etWrV06NAhvfbaa2rbtq1++uknVatWrVC+rxaLRa1atdKXX37pbvvuu+904sQJlSpVSmvWrHF/VqxevVrNmjVTmTJlst1WXj5L7rjjDtWqVUsTJkzQpk2b9MYbb6hy5cp69tlnc8ytLydPntQbb7yhHj16qH///jp16pTefPNNJSUlaf369br88ss9+s+YMUNnz57VgAEDFB4ervLly2vz5s3q0KGDqlatqrFjx8rlcmncuHHZnrd5+Wx78skndeLECf3222/u9ymnvAElggFQLJw4ccJIMjfffHOe+u/evdtIMjNmzPBaJskkJye7XycnJxtJZsCAAe62zMxMc9FFFxmLxWImTpzobj927JiJjIw0vXv3drfNmDHDSDK7d+/22M+KFSuMJLNixQp3W+/evU1cXJxHv9TUVI/XGRkZplGjRua6667zaC9durTHfnPa/4kTJ0x4eLgZNmyYR7/nnnvOWCwWs3fvXmOMMXv27DFWq9U888wzHv2+//57ExYW5tWe036/+eabHPuUK1fONGvWzP36fK7PmzJlipFkjhw5kuM2vvnmmxzfy7Zt2xpJZvr06dkua9u2rfv1+fejevXq5uTJk+72uXPnGklm2rRp7ra4uLhsc/33beYW29/f6wULFhhJ5p///KdHv27duhmLxWJ27tzpbpNk7Ha7R9u3335rJJmXXnrJa19/NXXqVCPJvPfee+62jIwMk5CQYMqUKeNx7HFxcaZTp065bs8YYx555BEjyWzevNlnX2OM6dq1q7Hb7WbXrl3utt9//92ULVvWtGnTxt12/hy65pprTGZmpsc2cnpv3333XVOqVCmzevVqj/bp06cbSWbNmjUex/fX9/Hs2bPG5XJ5rLd7924THh5uxo0b524rjO/rpEmTjNVqdb9/L774oomLizMtWrQwI0aMMMYY43K5THR0tHnkkUfc6/39d86YnD9Lzve95557PNpvueUWU6FChVzj+7uGDRt6/K5kZmaa9PR0jz7Hjh0zVapU8djf+c/uqKgoc/jwYY/+nTt3Ng6Hw+zfv9/dtmPHDhMWFuZxjP58tnXq1MnrMxkoqRiqBxQTJ0+elCSVLVs2aPvo16+f+99Wq1Xx8fEyxujee+91t0dHR+vSSy/VL7/8ErD9/vUqx7Fjx3TixAm1bt1amzZtytf2zg8/mjt3rsdQoTlz5uiqq67SxRdfLEmaP3++srKydMcdd+iPP/5w/8TGxqpevXpeQ2jyo0yZMrnOrnf+itZHH32krKysfO0jPDxcffv2zXP/Xr16eZxH3bp1U9WqVfXZZ5/la/959dlnn8lqtWrw4MEe7cOGDZMxRp9//rlHe2JioseVgCZNmigqKsrnuffZZ58pNjZWPXr0cLfZbDYNHjxYp0+f1qpVq/yO3Z/fP5fLpSVLlqhr164eQ0arVq2qu+66S1999ZV7e+f1798/2/v0sntvP/jgAzVo0ED169f3OG+vu+46Scr1vA0PD1epUqXccf7555/u4aH5/X0rqPe1devWcrlcWrt2raRzV5Zat26t1q1ba/Xq1ZKkH374QcePH3dfJcuv+++/32vff/75p9f75g+r1eq+RykrK0tHjx5VZmam4uPjs839bbfd5nElyeVy6YsvvlDXrl1VrVo1d3vdunXVsWNHj3UL4rMNKI4onIBiIioqSpICNsV1ds4XFOeVK1dOERERXrNLlStXTseOHQvYfj/99FNdddVVioiIUPny5VWpUiW9+uqrOnHiRL632b17d/36669KSUmRdG4K940bN6p79+7uPjt27JAxRvXq1VOlSpU8frZu3arDhw9f8LGdPn061y/b3bt3V6tWrdSvXz9VqVJFd955p+bOnetXEVW9enW/bhqvV6+ex2uLxaK6desGfVatvXv3qlq1al75aNCggXv5X/39fJSkmJgYn+fe3r17Va9ePXeB4Gs/eeHP79+RI0eUmpqqSy+91GtZgwYNlJWV5XVPWa1atbLdVnbv7Y4dO/Tjjz96nbOXXHKJJOV63mZlZWnKlCmqV6+ewsPDVbFiRVWqVMk97C0/Cup9veKKK+RwONxF0vnCqU2bNtqwYYPOnj3rXnbNNdfk61hyijEmJkaSLvhz7+2331aTJk0UERGhChUqqFKlSlq4cGG2uf/7OXH48GGlpaWpbt26Xn3/3lYQn21AccQ9TkAxERUVpWrVqumHH37IU/+cpuV2uVw5rpPdX7xzmq3ur1dy8rOv81avXq0uXbqoTZs2euWVV1S1alXZbDbNmDFDs2bN8rl+Tjp37iyHw6G5c+fq6quv1ty5c1WqVCndfvvt7j5ZWVmyWCz6/PPPsz3OCx3r/9tvv+nEiRPZftE5LzIyUl9++aVWrFihhQsXatGiRZozZ46uu+46LVmyJE+zBfpzX1Je5faeFtQMhnk59wpK/fr1JUnff/+9170ogZDTe5hde1ZWlho3bqzJkydnu06NGjVy3M/48eP11FNP6Z577tHTTz+t8uXLq1SpUhoyZEi+r3j6K7/vq81mU8uWLfXll19q586dOnjwoFq3bq0qVarI6XRq3bp1Wr16terXr5/jvWrBjjE37733nvr06aOuXbvqscceU+XKlWW1WjVhwgTt2rXLq/+F/F4H+7MNKK4onIBi5KabbtK///1vpaSkKCEhIde+5/9C+tfZpqT8/bXdlwvZ14cffqiIiAgtXrzY47kwM2bM8OrrzzOaSpcurZtuukkffPCBJk+erDlz5qh169YeQ1zq1KkjY4xq1arl/mt9IL377ruSpKSkpFz7lSpVSu3bt1f79u01efJkjR8/Xk8++aRWrFihxMTEgD+baseOHR6vjTHauXOnx/OmYmJivN5P6dx7+tfhZ/7EFhcXpy+++EKnTp3yuDqxbds29/JAiIuL03fffaesrCyPq04Xsp+OHTvKarXqvffe8zlBRKVKleRwOLR9+3avZdu2bVOpUqVyLW58qVOnjr799lu1b9/e73Nj3rx5uvbaa/Xmm296tB8/ftzjynJhfF+lc0Pmnn32WX3xxReqWLGi6tevL4vFooYNG2r16tVavXq1brrpJp/bCcWDaOfNm6fatWtr/vz5HvtPTk7O0/qVK1dWRESEdu7c6bXs723+fLYVxofyAqHCUD2gGBk+fLhKly6tfv366dChQ17Ld+3apWnTpkk6d4WqYsWKHrNQSdIrr7wS8LjO36/w1325XK48PYjXarXKYrF4XJ3as2dPtjODlS5dOtsv8znp3r27fv/9d73xxhv69ttvPYbpSdKtt94qq9WqsWPHev0l2RijP//8M8/7+rvly5fr6aefVq1atdSzZ88c+x09etSr7fwVjfNTS59/po8/x56bd955x2PI2bx583TgwAGP+yTq1Kmjr7/+2v0QXenckMq/DzHzJ7Ybb7xRLpdL//rXvzzap0yZIovF4nWfRn7deOONOnjwoObMmeNuy8zM1EsvvaQyZcqobdu2fm+zRo0a6t+/v5YsWaKXXnrJa3lWVpZeeOEF/fbbb7Jarbrhhhv00UcfeQx/PHTokGbNmqVrrrnGPfQvP+644w7t379fr7/+uteytLQ0j1n5/s5qtXqd6x988IH279/v0VYY31fpXOGUnp6uqVOn6pprrnF/6W/durXeffdd/f7773m6v8nfz5JAOH/l56/5X7dunXs4cV7WT0xM1IIFC/T777+723fu3Ol1H5k/n22lS5e+oGHRQHHCFSegGKlTp45mzZrlnja8V69eatSokTIyMrR27Vr3lMvn9evXTxMnTlS/fv0UHx+vL7/8Uj///HPA42rYsKGuuuoqjRw5UkePHlX58uU1e/ZsZWZm+ly3U6dOmjx5sjp06KC77rpLhw8f1ssvv6y6devqu+++8+jbvHlzffHFF5o8ebKqVaumWrVqeU3r/Fc33nijypYtq0cffVRWq1W33Xabx/I6deron//8p0aOHKk9e/aoa9euKlu2rHbv3q3//ve/GjBggB599FGfx/D5559r27ZtyszM1KFDh7R8+XItXbpUcXFx+vjjjxUREZHjuuPGjdOXX36pTp06KS4uTocPH9Yrr7yiiy66yH2fRp06dRQdHa3p06erbNmyKl26tFq2bJnjfTG+lC9fXtdcc4369u2rQ4cOaerUqapbt67HlOn9+vXTvHnz1KFDB91xxx3atWuX3nvvPa9pm/2JrXPnzrr22mv15JNPas+ePWratKmWLFmijz76SEOGDMlxenl/DRgwQK+99pr69OmjjRs3qmbNmpo3b57WrFmjqVOn5nuClRdeeEG7du3S4MGDNX/+fN10002KiYnRvn379MEHH2jbtm268847JUn//Oc/3c/nevDBBxUWFqbXXntN6enpeu655y7o+O6++27NnTtX999/v1asWKFWrVrJ5XJp27Ztmjt3rhYvXuz1rLLzbrrpJo0bN059+/bV1Vdfre+//17vv/++13PPCuP7Kp17DltYWJi2b9/unkpcktq0aaNXX31VkvJUOPn7WRIIN910k+bPn69bbrlFnTp10u7duzV9+nRddtll7mdO+TJmzBgtWbJErVq10gMPPOAuWBs1aqQtW7a4+/nz2da8eXPNmTNHQ4cO1ZVXXqkyZcqoc+fOwUgBUPgV8Cx+AArAzz//bPr3729q1qxp7Ha7KVu2rGnVqpV56aWXzNmzZ939UlNTzb333mvKlStnypYta+644w5z+PDhHKcj//uU2L179zalS5f22n/btm1Nw4YNPdp27dplEhMTTXh4uKlSpYp54oknzNKlS/M0Hfmbb75p6tWrZ8LDw039+vXNjBkzsp1CeNu2baZNmzYmMjLSSHJPJ5zTdOjGGNOzZ08jySQmJuaYzw8//NBcc801pnTp0qZ06dKmfv36ZuDAgWb79u05rvPX/Z7/sdvtJjY21lx//fVm2rRpHtNen/f341q2bJm5+eabTbVq1YzdbjfVqlUzPXr0MD///LPHeh999JG57LLL3NMOn58mOrv34rycpiP/z3/+Y0aOHGkqV65sIiMjTadOndxTtP/VCy+8YKpXr27Cw8NNq1atzIYNG7y2mVts2b3Xp06dMo888oipVq2asdlspl69embSpEkmKyvLo58kM3DgQK+Ycpom/e8OHTpk+vbtaypWrGjsdrtp3LhxtlNr53U68vMyMzPNG2+8YVq3bm3KlStnbDabiYuLM3379vWaqnzTpk0mKSnJlClTxjgcDnPttdeatWvXevTJbUr73N7bjIwM8+yzz5qGDRua8PBwExMTY5o3b27Gjh1rTpw44XF8f5+OfNiwYaZq1aomMjLStGrVyqSkpBSZ99UYY6688kojyaxbt87d9ttvvxlJpkaNGl79/fksyemzMLfPmJz8fTryrKwsM378eBMXF2fCw8NNs2bNzKeffuqVz/PTkU+aNCnb7S5btsw0a9bM2O12U6dOHfPGG2+YYcOGmYiICK++eflsO336tLnrrrtMdHS0kcTU5CjRLMaE4C5aAAAAFIiuXbvqxx9/9Lp/EYB/uMcJAACgmEhLS/N4vWPHDn322Wdq165daAICihGuOAEAABQTVatWVZ8+fVS7dm3t3btXr776qtLT07V582avZ7QB8A+TQwAAABQTHTp00H/+8x8dPHhQ4eHhSkhI0Pjx4ymagAAI6VC9L7/8Up07d1a1atVksViynV7471auXKkrrrhC4eHhqlu3rmbOnBn0OAEAAIqCGTNmaM+ePTp79qxOnDihRYsW6Yorrgh1WECxENLC6cyZM2ratKlefvnlPPXfvXu3OnXqpGuvvVZbtmzRkCFD1K9fPy1evDjIkQIAAAAoyQrNPU4Wi0X//e9/1bVr1xz7jBgxQgsXLtQPP/zgbrvzzjt1/PhxLVq0qACiBAAAAFASFal7nFJSUpSYmOjRlpSUpCFDhuS4Tnp6utLT092vs7KydPToUVWoUMH9RHEAAAAAJY8xRqdOnVK1atVUqlTug/GKVOF08OBBValSxaOtSpUqOnnypNLS0hQZGem1zoQJEzR27NiCChEAAABAEfPrr7/qoosuyrVPkSqc8mPkyJEaOnSo+/WJEyd08cUXa/fu3Spbtmy+t+t0OrVixQpde+21stlsgQgV/4/cBg+5DR5yGxzkNXjIbfCQ2+Aht8FRkHlNzchUq+e+lCStGd5GDntoy5FTp06pVq1aeaoLilThFBsbq0OHDnm0HTp0SFFRUdlebZKk8PBwhYeHe7WXL19eUVFR+Y7F6XTK4XCoQoUK/OIGGLkNHnIbPOQ2OMhr8JDb4CG3wUNug6Mg8xqZkalS4Q5JUoUKFUJeOJ0/3rzcwhPSWfX8lZCQoGXLlnm0LV26VAkJCSGKCAAAAEBJENLC6fTp09qyZYu2bNki6dx041u2bNG+ffsknRtm16tXL3f/+++/X7/88ouGDx+ubdu26ZVXXtHcuXP1yCOPhCJ8AAAAACVESAunDRs2qFmzZmrWrJkkaejQoWrWrJlGjx4tSTpw4IC7iJKkWrVqaeHChVq6dKmaNm2qF154QW+88YaSkpJCEj8AAACAkiGkgwrbtWun3B4jNXPmzGzX2bx5cxCjAgAAAABPReoeJwAAAAAIBQonAAAAAPCBwgkAAAAAfChSz3ECAAAAEFrGGKU5XflaNzUjf+sVBhROAAAAAPLEGKNu01O0ce+xUIdS4BiqBwAAACBP0pyugBRN8XExirRZAxBRweGKEwAAAAC/bRiVKIc9f8VPpM0qi8US4IiCi8IJAAAAgN8cdqsc9pJTTjBUDwAAAAB8oHACAAAAAB8onAAAAADABwonAAAAAPCBwgkAAAAAfKBwAgAAAAAfKJwAAAAAwAcKJwAAAADwoeQ8sQoAAACAX4wxSnO63K9TM1y59C7eKJwAAAAAeDHGqNv0FG3ceyzUoRQKDNUDAAAA4CXN6cqxaIqPi1GkzVrAEYUWV5wAAAAA5GrDqEQ57P8rlCJtVlkslhBGVPAonAAAAADkymG3ymEv2aUDQ/UAAAAAwAcKJwAAAADwgcIJAAAAAHygcAIAAAAAHyicAAAAAMAHCicAAAAA8IHCCQAAAAB8KNmTsQMAAAAlgDFGaU5Xtsuczkylu6TUjEzZzP8eapuakX3/korCCQAAACjGjDHqNj1FG/cey6VXmIavX15gMRVFDNUDAAAAirE0p8tH0ZS7+LgYRdqsAYyoaOKKEwAAAFBCbBiVKIfdswhyOp1avHiJkpJukM1m81on0maVxWLxai9pKJwAAACAEsJht8ph9ywBnBajcKvksIfJZqM8yAlD9QAAAADABwonAAAAAPCBwgkAAAAAfKBwAgAAAAAfKJwAAAAAwAcKJwAAAADwgcIJAAAAAHygcAIAAAAAH3jCFQAAAFDIGWOU5nTla93UjPytB08UTgAAAEAhZoxRt+kp2rj3WKhDKdEYqgcAAAAUYmlOV0CKpvi4GEXarAGIqGTiihMAAABQRGwYlSiHPX/FT6TNKovFEuCISg4KJwAAAKCIcNitctj5Ch8KDNUDAAAAAB8onAAAAADABwonAAAAAPCBwgkAAAAAfKBwAgAAAAAfKJwAAAAAwAcKJwAAAADwgcIJAAAAKKSMMUrNcIU6DIgH4AIAAACFkjFG3aanaOPeY6EOBeKKEwAAAFAopTldHkVTfFyMIm3WEEZUsnHFCQAAACjkNoxKVIXSdlksllCHUmJxxQkAAAAo5Bx2K0VTiFE4AQAAAIAPFE4AAAAA4AOFEwAAAAD4QOEEAAAAAD5QOAEAAACADxROAAAAAOADhRMAAAAA+EDhBAAAAAA+hIU6AAAAAKCkMMYozenKU9/UjLz1Q8GgcAIAAAAKgDFG3aanaOPeY6EOBfnAUD0AAACgAKQ5XfkqmuLjYhRpswYhIviDK04AAABAAdswKlEOe96KoUibVRaLJcgRwRcKJwAAAKCAOexWOex8FS9KGKoHAAAAAD5QOAEAAACADxROAAAAAOADhRMAAAAA+EDhBAAAAAA+UDgBAAAAgA8UTgAAAADgA4UTAAAAUACMCXUEuBAUTgAAAECQGWN0+/SUUIeBC0DhBAAAAARZmtOlnw6clCRdVjVKkTZriCOCv0JeOL388suqWbOmIiIi1LJlS61fvz7X/lOnTtWll16qyMhI1ahRQ4888ojOnj1bQNECAAAAF+aD+xNksVhCHQb8FNLCac6cORo6dKiSk5O1adMmNW3aVElJSTp8+HC2/WfNmqXHH39cycnJ2rp1q958803NmTNHTzzxRAFHDgAAAOQPNVPRFNLCafLkyerfv7/69u2ryy67TNOnT5fD4dBbb72Vbf+1a9eqVatWuuuuu1SzZk3dcMMN6tGjh8+rVAAAAABwIcJCteOMjAxt3LhRI0eOdLeVKlVKiYmJSknJ/sa5q6++Wu+9957Wr1+vFi1a6JdfftFnn32mu+++O8f9pKenKz093f365MlzY0udTqecTme+4z+/7oVsA9kjt8FDboOH3AYHeQ0echs85DZ4inJunc7Mv/zbKael8EyxV5TzeqH8OWaLMaGZGPH3339X9erVtXbtWiUkJLjbhw8frlWrVmndunXZrvfiiy/q0UcflTFGmZmZuv/++/Xqq6/muJ8xY8Zo7NixXu2zZs2Sw+G48AMBAAAAfEh3ScPXn7tm8VyLTIUzN0ShkJqaqrvuuksnTpxQVFRUrn1DdsUpP1auXKnx48frlVdeUcuWLbVz5049/PDDevrpp/XUU09lu87IkSM1dOhQ9+uTJ0+qRo0auuGGG3wmJzdOp1NLly7V9ddfL5vNlu/twBu5DR5yGzzkNjjIa/CQ2+Aht8FTlHObmpGp4euXS5KSkm6Qw154voYX5bxeqPOj0fIiZO9YxYoVZbVadejQIY/2Q4cOKTY2Ntt1nnrqKd19993q16+fJKlx48Y6c+aMBgwYoCeffFKlSnnfshUeHq7w8HCvdpvNFpATI1DbgTdyGzzkNnjIbXCQ1+Aht8FDboOnKObWZv43I8S5+AtP4XReUczrhfLneEM2OYTdblfz5s21bNkyd1tWVpaWLVvmMXTvr1JTU72KI6v13HXOEI04BAAAAFAChLTUHTp0qHr37q34+Hi1aNFCU6dO1ZkzZ9S3b19JUq9evVS9enVNmDBBktS5c2dNnjxZzZo1cw/Ve+qpp9S5c2d3AQUAAAAAgRbSwql79+46cuSIRo8erYMHD+ryyy/XokWLVKVKFUnSvn37PK4wjRo1ShaLRaNGjdL+/ftVqVIlde7cWc8880yoDgEAAABACRDywZWDBg3SoEGDsl22cuVKj9dhYWFKTk5WcnJyAUQGAAAAXBhjjNKcLqVmuEIdCi5QyAsnAAAAoDgyxqjb9BRt3Hss1KEgAEI2OQQAAABQnKU5XV5FU3xcjCJt3JtfFHHFCQAAAAiyDaMS5bBbFWmzymKx+F4BhQ6FEwAAABBkDru1UD30Fv5jqB4AAAAA+EDhBAAAAAA+UDgBAAAAgA8UTgAAAADgA4UTAAAAAPhA4QQAAAAAPlA4AQAAAIAPTCYPAACAEs8YozSnK6DbTM0I7PYQWhROAAAAKNGMMeo2PUUb9x4LdSgoxBiqBwAAgBItzekKatEUHxejSJs1aNtHweCKEwAAAPD/NoxKlMMe2CIn0maVxWIJ6DZR8CicAAAAgP/nsFvlsPMVGd4YqgcAAAAAPlA4AQAAAIAPFE4AAAAA4AOFEwAAAAD4QOEEAAAAAD5QOAEAAACADxROAAAAAOADhRMAAAAA+MDTvQAAAFAiGWOU5nQpNcMV6lBQBFA4AQAAoMQxxqjb9BRt3Hss1KGgiGCoHgAAAEqcNKfLq2iKj4tRpM0aoohQ2HHFCQAAACXahlGJctitirRZZbFYQh0OCikKJwAAAJRoDrtVDjtfi5E7huoBAAAAgA8UTgAAAADgA4UTAAAAAPhA4QQAAAAAPlA4AQAAAIAPFE4AAAAA4AOFEwAAAAD4wIT1AAAAKDGMMUpzupSa4Qp1KChiKJwAAABQIhhj1G16ijbuPRbqUFAEMVQPAAAAJUKa0+VVNMXHxSjSZg1RRChKuOIEAACAEmfDqEQ57FZF2qyyWCyhDgdFAIUTAAAAShyH3SqHna/CyDuG6gEAAACADxROAAAAAOADhRMAAAAA+EDhBAAAAAA+UDgBAAAAgA8UTgAAAADgA4UTAAAAAPhA4QQAAAAAPvDULwAAABQLxhilOV05Lk/NyHkZ4AuFEwAAAIo8Y4y6TU/Rxr3HQh0KiimG6gEAAKDIS3O68lw0xcfFKNJmDXJEKG644gQAAIBiZcOoRDnsORdGkTarLBZLAUaE4oDCCQAAAMWKw26Vw87XXAQWQ/UAAAAAwAcKJwAAAADw4YILJ5fLpS1btujYMWYwAQAAAFA8+V04DRkyRG+++aakc0VT27ZtdcUVV6hGjRpauXJloOMDAAAAgJDzu3CaN2+emjZtKkn65JNPtHv3bm3btk2PPPKInnzyyYAHCAAAAACh5nfh9Mcffyg2NlaS9Nlnn+n222/XJZdconvuuUfff/99wAMEAAAAgFDzu3CqUqWKfvrpJ7lcLi1atEjXX3+9JCk1NVVWKw8SAwAAAFD8+D3Bfd++fXXHHXeoatWqslgsSkxMlCStW7dO9evXD3iAAAAAQE6MMUpzupSa4Qp1KCjm/C6cxowZo0aNGunXX3/V7bffrvDwcEmS1WrV448/HvAAAQAAgOwYY9Rteoo27mV2ZwRfvh6p3K1bN0nS2bNn3W29e/cOTEQAAABAHqQ5XV5FU3xcjCJt3D6CwPP7HieXy6Wnn35a1atXV5kyZfTLL79Ikp566in3NOUAAABAQdowKlE/jUvSB/cnyGKxhDocFEN+F07PPPOMZs6cqeeee052u93d3qhRI73xxhsBDQ4AAADIC4fdKoc9jKIJQeN34fTOO+/o3//+t3r27Okxi17Tpk21bdu2gAYHAAAAAIWB34XT/v37VbduXa/2rKwsOZ3OgAQFAAAAAIWJ34XTZZddptWrV3u1z5s3T82aNQtIUAAAAABQmPg9q97o0aPVu3dv7d+/X1lZWZo/f762b9+ud955R59++mkwYgQAAACAkPL7itPNN9+sTz75RF988YVKly6t0aNHa+vWrfrkk090/fXXByNGAAAAAAgpv644ZWZmavz48brnnnu0dOnSYMUEAAAAAIWKX1ecwsLC9NxzzykzMzNY8QAAAABAoeP3UL327dtr1apVwYgFAAAAAAolvyeH6Nixox5//HF9//33at68uUqXLu2xvEuXLgELDgAAAEWXMUZpTlfAtud0ZirdJaVmZMpmLErNCNy2AV/8LpwefPBBSdLkyZO9llksFrlcnMAAAAAlnTFG3aanaOPeYwHecpiGr18e4G0CvvldOGVlZQUjDgAAABQjaU5XEIqm7MXHxSjSZi2QfaHk8rtwAgAAAPyxYVSiHPYLL2ycTqcWL16ipKQbZLPZ3O2RNqssFssFbx/Ijd+F07hx43JdPnr06HwHAwAAgOLHYbfKYb/wv9c7LUbhVslhD5PNxt//UbD8PuP++9//erx2Op3avXu3wsLCVKdOHQonAAAAAMWO34XT5s2bvdpOnjypPn366JZbbglIUAAAAABQmPj9HKfsREVFaezYsXrqqacCsTkAAAAAKFQCUjhJ0okTJ3TixAm/13v55ZdVs2ZNRUREqGXLllq/fn2u/Y8fP66BAweqatWqCg8P1yWXXKLPPvssv2EDAAAAgE9+D9V78cUXPV4bY3TgwAG9++676tixo1/bmjNnjoYOHarp06erZcuWmjp1qpKSkrR9+3ZVrlzZq39GRoauv/56Va5cWfPmzVP16tW1d+9eRUdH+3sYAAAAAJBnfhdOU6ZM8XhdqlQpVapUSb1799bIkSP92tbkyZPVv39/9e3bV5I0ffp0LVy4UG+99ZYef/xxr/5vvfWWjh49qrVr17qnoKxZs6a/hwAAAAAAfvG7cNq9e3dAdpyRkaGNGzd6FFulSpVSYmKiUlJSsl3n448/VkJCggYOHKiPPvpIlSpV0l133aURI0bIas3+2QDp6elKT093vz558qSkc7MBOp3OfMd/ft0L2QayR26Dh9wGD7kNDvIaPOQ2eMjtOU5n5l/+7ZTTYgKwTXIbDCU5r/4cs8UY49dZfM8992jatGkqW7asR/uZM2f00EMP6a233srTdn7//XdVr15da9euVUJCgrt9+PDhWrVqldatW+e1Tv369bVnzx717NlTDz74oHbu3KkHH3xQgwcPVnJycrb7GTNmjMaOHevVPmvWLDkcjjzFCgAAUNIYI2Vk5X/9jCxp1IZzf6N/rkWmwi/8+bdAwKWmpuquu+7SiRMnFBUVlWtfvwsnq9WqAwcOeN2D9Mcffyg2NlaZmZk5rOkpP4XTJZdcorNnz2r37t3uK0yTJ0/WpEmTdODAgWz3k90Vpxo1auiPP/7wmZzcOJ1OLV26VNdff73Hk6tx4cht8JDb4CG3wUFeg4fcBk9xyK0xRne+8Y027TsekO19+9R1gXkAbjHIbWFUkvN68uRJVaxYMU+FU57P4JMnT8oYI2OMTp06pYiICPcyl8ulzz77LNsJHXJSsWJFWa1WHTp0yKP90KFDio2NzXadqlWrymazeQzLa9CggQ4ePKiMjAzZ7XavdcLDwxUeHu7VbrPZAnJiBGo78EZug4fcBg+5DQ7yGjzkNniKcm5TMzIDVjTFx8UoyhEhi8USkO1JRTu3hVlJzKs/x5vnwik6OloWi0UWi0WXXHKJ13KLxZLtkLic2O12NW/eXMuWLVPXrl0lSVlZWVq2bJkGDRqU7TqtWrXSrFmzlJWVpVKlzs2k/vPPP6tq1arZFk0AAAC4MBtGJcphz/84u0ibNaBFExAqeS6cVqxYIWOMrrvuOn344YcqX768e5ndbldcXJyqVavm186HDh2q3r17Kz4+Xi1atNDUqVN15swZ9yx7vXr1UvXq1TVhwgRJ0gMPPKB//etfevjhh/XQQw9px44dGj9+vAYPHuzXfgEAAJA3Drs1IMPsgKIuz78Fbdu2lXRuVr0aNWq4r/hciO7du+vIkSMaPXq0Dh48qMsvv1yLFi1SlSpVJEn79u3z2E+NGjW0ePFiPfLII2rSpImqV6+uhx9+WCNGjLjgWAAAAAAgJ37/+SAuLk7SuRko9u3bp4yMDI/lTZo08Wt7gwYNynFo3sqVK73aEhIS9PXXX/u1DwAAAAC4EH4XTkeOHFHfvn31+eefZ7vc5XJdcFAAAAAAUJj4Pd5uyJAhOn78uNatW6fIyEgtWrRIb7/9turVq6ePP/44GDECAAAAQEj5fcVp+fLl+uijjxQfH69SpUopLi5O119/vaKiojRhwgR16tQpGHECAAAAQMj4fcXpzJkz7uc1xcTE6MiRI5Kkxo0ba9OmTYGNDgAAAAAKAb8Lp0svvVTbt2+XJDVt2lSvvfaa9u/fr+nTp6tq1aoBDxAAAAAAQs3voXoPP/ywDhw4IElKTk5Whw4d9P7778tut2vmzJmBjg8AAAAAQs7vwukf//iH+9/NmzfX3r17tW3bNl188cWqWLFiQIMDAAAo6owxSnMWrVmHUzOKVrxAQcj3Y6AzMjK0e/du1alTR1dccUUgYwIAACgWjDHqNj1FG/ceC3UoAC6Q3/c4paam6t5775XD4VDDhg21b98+SdJDDz2kiRMnBjxAAACAoirN6SrSRVN8XIwibdZQhwEUCn5fcRo5cqS+/fZbrVy5Uh06dHC3JyYmasyYMXr88ccDGiAAAEBxsGFUohz2olWERNqsslgsoQ4DKBT8LpwWLFigOXPm6KqrrvL4RWrYsKF27doV0OAAAACKC4fdKoc933dJAAgxv4fqHTlyxP0cp786c+YMf5EAAAAAUCz5XTjFx8dr4cKF7tfni6U33nhDCQkJgYsMAAAAAAoJv68Xjx8/Xh07dtRPP/2kzMxMTZs2TT/99JPWrl2rVatWBSNGAAAAAAgpv684XXPNNdqyZYsyMzPVuHFjLVmyRJUrV1ZKSoqaN28ejBgBAAAAIKTyfMVp+fLlatOmjcLCwlSnTh29/vrrwYwLAAAAAAqNPF9xuv7663X06FH366uuukr79+8PSlAAAAAAUJjkuXAyxni8/vHHH5Wenh7wgAAAAACgsOFhAgAAAH4wxijN6cpxudOZqXSXlJqRKafx+3ZyAIVUngsni8Xi8Zymv78GAAAo7owx6jY9RRv3HvPRM0zD1y8vkJgAFIw8F07GGLVv315hYedWSU1NVefOnWW32z36bdq0KbARAgAAFBJpTlceiiZv8XExirRZgxARgIKS58IpOTnZ4/XNN98c8GAAAACKig2jEuWwexdDTqdTixcvUVLSDbLZbJKkSJuVkTpAEZfvwgkAAKAkc9itcti9v0o5LUbhVslhD5PNxu3kQHHBHYsAAAAA4AOFEwAAAAD4QOEEAAAAAD5QOAEAAACAD3kqnMqXL68//vhDknTPPffo1KlTQQ0KAAAAAAqTPBVOGRkZOnnypCTp7bff1tmzZ4MaFAAAAAAUJnmaIzMhIUFdu3ZV8+bNZYzR4MGDFRkZmW3ft956K6ABAgAAAECo5alweu+99zRlyhTt2rVLFotFJ06c4KoTAAC4YMYYpTldoQ4jz1Izik6sAAIrT4VTlSpVNHHiRElSrVq19O6776pChQpBDQwAABRvxhh1m56ijXuPhToUAPDJ78dZ7969OxhxAACAEibN6SqyRVN8XIwibdZQhwGgAPldOEnSqlWr9Pzzz2vr1q2SpMsuu0yPPfaYWrduHdDgAABAybBhVKIc9qJTiETarLJYLKEOA0AB8rtweu+999S3b1/deuutGjx4sCRpzZo1at++vWbOnKm77ror4EECAIDizWG3ymHP199zAaBA+P0J9cwzz+i5557TI4884m4bPHiwJk+erKeffprCCQAAAECxk6fnOP3VL7/8os6dO3u1d+nShfufAAAAABRLfhdONWrU0LJly7zav/jiC9WoUSMgQQEAAABAYeL3UL1hw4Zp8ODB2rJli66++mpJ5+5xmjlzpqZNmxbwAAEAAAAg1PwunB544AHFxsbqhRde0Ny5cyVJDRo00Jw5c3TzzTcHPEAAAAAACLV8TV9zyy236JZbbgl0LAAAAABQKPl9jxMAAAAAlDQUTgAAAADgA0+aAwAAPhljlOZ0BXSbqRmB3R4ABBOFEwAAyJUxRt2mp2jj3mOhDgUAQoahegAAIFdpTldQi6b4uBhF2qxB2z4ABILfV5xcLpdmzpypZcuW6fDhw8rKyvJYvnz58oAFBwAACpcNoxLlsAe2yIm0WWWxWAK6TQAINL8Lp4cfflgzZ85Up06d1KhRIz7oAAAoQRx2qxx2RvoDKHn8/uSbPXu25s6dqxtvvDEY8QAAAABAoeP3PU52u11169YNRiwAAAAAUCj5XTgNGzZM06ZNkzEmGPEAAAAAQKHj91C9r776SitWrNDnn3+uhg0bymazeSyfP39+wIIDAAAAgMLA78IpOjpat9xySzBiAQAAAIBCye/CacaMGcGIAwAAAAAKrXzPJ3rkyBFt375dknTppZeqUqVKAQsKAAAAAAoTvwunM2fO6KGHHtI777zjfvit1WpVr1699NJLL8nhcAQ8SAAA4D9jjNKcrmyXOZ2ZSndJqRmZspncn8mYmpH9NgCgJPG7cBo6dKhWrVqlTz75RK1atZJ0bsKIwYMHa9iwYXr11VcDHiQAAPCPMUbdpqdo495jufQK0/D1ywssJgAoyvwunD788EPNmzdP7dq1c7fdeOONioyM1B133EHhBABAIZDmdPkomvwXHxejSJs1oNsEgKLC78IpNTVVVapU8WqvXLmyUlNTAxIUAAAInA2jEuWwexY8TqdTixcvUVLSDV6PFslJpM0qiyX3YX0AUFz5/QDchIQEJScn6+zZs+62tLQ0jR07VgkJCQENDgAAXDiH3SqHPczrJ9yqbNtz+qFoAlCS+X3Fadq0aUpKStJFF12kpk2bSpK+/fZbRUREaPHixQEPEAAAAABCze/CqVGjRtqxY4fef/99bdu2TZLUo0cP9ezZU5GRkQEPEAAAAABCLV/PcXI4HOrfv3+gYwEAAACAQilPhdPHH3+sjh07ymaz6eOPP861b5cuXQISGAAAAAAUFnkqnLp27aqDBw+qcuXK6tq1a479LBaLXC4ekgcAAACgeMlT4ZSVlZXtvwEAAACgJPB7OvJ33nlH6enpXu0ZGRl65513AhIUAAAAABQmfhdOffv21YkTJ7zaT506pb59+wYkKAAAAAAoTPyeVc8Yk+0D8H777TeVK1cuIEEBAIC8M8Yozel5j3FqBvccA0Ag5blwatasmSwWiywWi9q3b6+wsP+t6nK5tHv3bnXo0CEoQQIAgOwZY9Rteoo27j0W6lAAoFjLc+F0fja9LVu2KCkpSWXKlHEvs9vtqlmzpm677baABwgAAHKW5nTlWjTFx8Uo0mYtwIgAoHjKc+GUnJwsl8ulmjVr6oYbblDVqlWDGRcAAPDThlGJctg9i6RImzXbIfYAAP/4NTmE1WrVfffdp7NnzwYrHgAAkE8Ou1UOe5jHD0UTAASG37PqNWrUSL/88kswYgEAAACAQsnvwumf//ynHn30UX366ac6cOCATp486fEDAAAAAMWN39OR33jjjZKkLl26eFz+Pz9NucvF9KcAAAAAihe/C6cVK1YEIw4AAAAAKLT8Lpzatm0bjDgAAAAAoNDyu3CSpOPHj+vNN9/U1q1bJUkNGzbUPffco3LlygU0OAAAAAAoDPyeHGLDhg2qU6eOpkyZoqNHj+ro0aOaPHmy6tSpo02bNgUjRgAAAAAIKb+vOD3yyCPq0qWLXn/9dYWFnVs9MzNT/fr105AhQ/Tll18GPEgAAPA/xhilOc9NxpSawaRMAFAQ8nXFacSIEe6iSZLCwsI0fPhwbdiwIV9BvPzyy6pZs6YiIiLUsmVLrV+/Pk/rzZ49WxaLRV27ds3XfgEAKGqMMeo2PUWXjV6sy0YvVvw/vwh1SABQIvhdOEVFRWnfvn1e7b/++qvKli3rdwBz5szR0KFDlZycrE2bNqlp06ZKSkrS4cOHc11vz549evTRR9W6dWu/9wkAQFGV5nRp495jXu3xcTGKtFlDEBEAlAx+D9Xr3r277r33Xj3//PO6+uqrJUlr1qzRY489ph49evgdwOTJk9W/f3/17dtXkjR9+nQtXLhQb731lh5//PFs13G5XOrZs6fGjh2r1atX6/jx437vFwCAom7DqEQ57OeKpUib1eP5igCAwPK7cHr++edlsVjUq1cvZWZmSpJsNpseeOABTZw40a9tZWRkaOPGjRo5cqS7rVSpUkpMTFRKSkqO640bN06VK1fWvffeq9WrV+e6j/T0dKWnp7tfnzx5UpLkdDrldDr9ivevzq97IdtA9sht8JDb4CG3wUFevTmdme5/2yxZslnODR45///kvG+H3AYLuQ0echscJTmv/hyzxRhj8rOT1NRU7dq1S5JUp04dORwOv7fx+++/q3r16lq7dq0SEhLc7cOHD9eqVau0bt06r3W++uor3XnnndqyZYsqVqyoPn366Pjx41qwYEG2+xgzZozGjh3r1T5r1qx8xQwAQCilu6Th68/93fO5FpkKZ3QeAORbamqq7rrrLp04cUJRUVG59s3Xc5wkyeFwKDo62v3vgnDq1Cndfffdev3111WxYsU8rTNy5EgNHTrU/frkyZOqUaOGbrjhBp/JyY3T6dTSpUt1/fXXy2az5Xs78EZug4fcBg+5DQ7y6i01I1PD1y+XJCUl3SCHPX//Kye3wUNug4fcBkdJzuv50Wh54fenbWZmpsaOHasXX3xRp0+fliSVKVNGDz30kJKTk/1KdsWKFWW1WnXo0CGP9kOHDik2Ntar/65du7Rnzx517tzZ3ZaVlXXuQMLCtH37dtWpU8djnfDwcIWHh3tty2azBeTECNR24I3cBg+5DR5yGxzk9X9s5n/3MZ3LS77/BvqXbZDbYCC3wUNug6Mk5tWf4/X70/ahhx7S/Pnz9dxzz7mH16WkpGjMmDH6888/9eqrr+Z5W3a7Xc2bN9eyZcvcU4pnZWVp2bJlGjRokFf/+vXr6/vvv/doGzVqlE6dOqVp06apRo0a/h4OAAAAAPjkd+E0a9YszZ49Wx07dnS3NWnSRDVq1FCPHj38KpwkaejQoerdu7fi4+PVokULTZ06VWfOnHHPsterVy9Vr15dEyZMUEREhBo1auSx/vnhgn9vBwAAAIBA8btwCg8PV82aNb3aa9WqJbvd7ncA3bt315EjRzR69GgdPHhQl19+uRYtWqQqVapIkvbt26dSpfx+3BQAAAAABIzfhdOgQYP09NNPa8aMGe57h9LT0/XMM89kO7wur9vMad2VK1fmuu7MmTPztU8AAAAAyCu/C6fNmzdr2bJluuiii9S0aVNJ0rfffquMjAy1b99et956q7vv/PnzAxcpAAAAAISI34VTdHS0brvtNo82JmUAACC4jDFKc7qUmuEKdSgAUCL5XTjNmDEjGHEAAIAcGGPUbXqKNu49FupQAKDEyvfDH44cOaLt27dLki699FJVqlQpYEEBAID/SXO6vIqm+LgYRdqsIYoIAEoevwunM2fO6KGHHtI777zjfvis1WpVr1699NJLL8nhcAQ8SAAAcM6GUYly2K2KtFllsVh8rwAACAi/5/keOnSoVq1apU8++UTHjx/X8ePH9dFHH2nVqlUaNmxYMGIEAAD/z2G3ymEPo2gCgALm9xWnDz/8UPPmzVO7du3cbTfeeKMiIyN1xx13+P0AXAAAAAAo7Py+4pSamup+OO1fVa5cWampqQEJCgAAAAAKE78Lp4SEBCUnJ+vs2bPutrS0NI0dO1YJCQkBDQ4AAAAACgO/h+pNnTpVHTp08HoAbkREhBYvXhzwAAEAAAAg1PwunBo3bqwdO3bo/fff17Zt2yRJPXr0UM+ePRUZGRnwAAEAAAAg1PwqnJxOp+rXr69PP/1U/fv3D1ZMAAAAAFCo+HWPk81m87i3CQAAAABKAr+H6g0cOFDPPvus3njjDYWF+b06AACFljFGaU5XqMPwkppR+GICgJLG78rnm2++0bJly7RkyRI1btxYpUuX9lg+f/78gAUHAEBBMcao2/QUbdx7LNShAAAKIb8Lp+joaN12223BiAUAgJBJc7oKfdEUHxejSJs11GEAQInkd+E0Y8aMYMQBAEChsWFUohz2wlegRNqsslgsoQ4DAEqkPBdOWVlZmjRpkj7++GNlZGSoffv2Sk5OZgpyAECx47Bb5bBzHy8A4H/yPKveM888oyeeeEJlypRR9erVNW3aNA0cODCYsQEAAABAoZDnwumdd97RK6+8osWLF2vBggX65JNP9P777ysrKyuY8QEAAABAyOW5cNq3b59uvPFG9+vExERZLBb9/vvvQQkMAAAAAAqLPBdOmZmZioiI8Giz2WxyOp0BDwoAAAAACpM83/lqjFGfPn0UHh7ubjt79qzuv/9+j2c58RwnAAAAAMVNngun3r17e7X94x//CGgwAAAAAFAY5blw4vlNAAAAAEqqPN/jBAAAAAAlFU/3AwAUGGOM0pyuPPd3OjOV7pJSMzJlM5YgRialZuQ9LgBAyUPhBAAoEMYYdZueoo17j/m5ZpiGr18elJgAAMgrhuoBAApEmtOVj6Kp4MXHxSjSZg11GACAQoYrTgCAArdhVKIcdt/FidPp1OLFS5SUdINsNlsBRCZF2qyyWII7LBAAUPRQOAEACpzDbpXD7vt/QU6LUbhVctjDZLPxvywAQOgwVA8AAAAAfKBwAgAAAAAfKJwAAAAAwAcKJwAAAADwgcIJAAAAAHygcAIAAAAAHyicAAAAAMAHHooBoNgwxijN6QrJvp3OTKW7pNSMTNkMD0/NTmpGaN4bAAACgcIJQLFgjFG36SnauPdYCKMI0/D1y0O4fwAAECwM1QNQLKQ5XSEumpBX8XExirRZQx0GAAB+4YoTgGJnw6hEOewF+8Xc6XRq8eIlSkq6QTabrUD3XdRE2qyyWBjOCAAoWiicABQ7DrtVDnvBfrw5LUbhVslhD5PNxkcrAADFDUP1AAAAAMAHCicAAAAA8IHCCQAAAAB8oHACAAAAAB8onAAAAADABwonAAAAAPCBwgkAAAAAfKBwAgAAAAAfeEojgDwzxijN6Qp1GNlKzSiccQEAgOKBwglAnhhj1G16ijbuPRbqUAAAAAocQ/UA5Ema01Ukiqb4uBhF2qyhDgMAABQzXHEC4LcNoxLlsBfO4iTSZpXFYgl1GAAAoJihcALgN4fdKoedjw8AAFByMFQPAAAAAHygcAIAAAAAHyicAAAAAMAHCicAAAAA8IHCCQAAAAB8oHACAAAAAB8onAAAAADABx7EAhQjxhilOV05Lnc6M5XuklIzMmUz/j0kNjUj5+0CAAAUdxROQDFhjFG36SnauPeYj55hGr5+eYHEBAAAUFwwVA8oJtKcrjwUTRcuPi5GkTZr0PcDAABQmHDFCSiGNoxKlMPuXdw4nU4tXrxESUk3yGaz5WvbkTarLBb/hvkBAAAUdRROQDHksFvlsHv/ejstRuFWyWEPk83Grz8AAEBeMVQPAAAAAHygcAIAAAAAHyicAAAAAMAHCicAAAAA8IHCCQAAAAB8oHACAAAAAB8onAAAAADABwonAAAAAPCBJ2ACIWaMUZrTdcHbSc248G0AAAAgexROQAgZY9Rteoo27j0W6lAAAACQC4bqASGU5nQFvGiKj4tRpM0a0G0CAACUdFxxAgqJDaMS5bBfeMETabPKYrEEICIAAACcR+EEFBIOu1UOO7+SAAAAhVGhGKr38ssvq2bNmoqIiFDLli21fv36HPu+/vrrat26tWJiYhQTE6PExMRc+wMAAADAhQp54TRnzhwNHTpUycnJ2rRpk5o2baqkpCQdPnw42/4rV65Ujx49tGLFCqWkpKhGjRq64YYbtH///gKOHAAAAEBJEfLCafLkyerfv7/69u2ryy67TNOnT5fD4dBbb72Vbf/3339fDz74oC6//HLVr19fb7zxhrKysrRs2bICjhwAAABASRHSGyoyMjK0ceNGjRw50t1WqlQpJSYmKiUlJU/bSE1NldPpVPny5bNdnp6ervT0dPfrkydPSpKcTqecTme+Yz+/7oVsA9krSbl1OjP/8m+nnBYT5P2VnNwWNHIbHOQ1eMht8JDb4CG3wVGS8+rPMVuMMcH9ppaL33//XdWrV9fatWuVkJDgbh8+fLhWrVqldevW+dzGgw8+qMWLF+vHH39URESE1/IxY8Zo7NixXu2zZs2Sw+G4sAMALlC6Sxq+/tzfL55rkalwZhEHAAAoMKmpqbrrrrt04sQJRUVF5dq3SE/hNXHiRM2ePVsrV67MtmiSpJEjR2ro0KHu1ydPnnTfF+UrOblxOp1aunSprr/+etlstnxvB95KUm5TMzI1fP1ySVJS0g1Bn1WvJOW2oJHb4CCvwUNug4fcBg+5DY6SnNfzo9HyIqSFU8WKFWW1WnXo0CGP9kOHDik2NjbXdZ9//nlNnDhRX3zxhZo0aZJjv/DwcIWHh3u122y2gJwYgdoOvJWE3NrM/563dO54C+ZXsiTkNlTIbXCQ1+Aht8FDboOH3AZHScyrP8cb0skh7Ha7mjdv7jGxw/mJHv46dO/vnnvuOT399NNatGiR4uPjCyJUwG/GGKVmZPr4cYU6TAAAAORByIfqDR06VL1791Z8fLxatGihqVOn6syZM+rbt68kqVevXqpevbomTJggSXr22Wc1evRozZo1SzVr1tTBgwclSWXKlFGZMmVCdhzAXxlj1G16ijbuPRbqUAAAABAAIS+cunfvriNHjmj06NE6ePCgLr/8ci1atEhVqlSRJO3bt0+lSv3vwtirr76qjIwMdevWzWM7ycnJGjNmTEGGDuQozenyq2iKj4tRpI2ZIQAAAAqrkBdOkjRo0CANGjQo22UrV670eL1nz57gBwQE0IZRiXLYcy+KIm1WWSyWXPsAAAAgdApF4QQUZw67Neiz5QEAACC4Qjo5BAAAAAAUBRROAAAAAOADhRMAAAAA+EDhBAAAAAA+UDgBAAAAgA8UTgAAAADgA4UTAAAAAPhA4QQAAAAAPvBUTsAHY4zSnC6/1knN8K8/AAAACjcKJyAXxhh1m56ijXuPhToUAAAAhBBD9YBcpDldF1Q0xcfFKNJmDWBEAAAACAWuOAF5tGFUohx2/4qgSJtVFoslSBEBAACgoFA4AXnksFvlsPMrAwAAUBIxVA8AAAAAfKBwAgAAAAAfKJwAAAAAwAcKJwAAAADwgcIJAAAAAHygcAIAAAAAHyicAAAAAMAHHkqDbBljlOZ0hWTfTmem0l1SakambCa0D49NzQhNDgAAAFC4UDjBizFG3aanaOPeYyGMIkzD1y8P4f4BAACA/2GoHrykOV0hLpoKn/i4GEXarKEOAwAAACHCFSfkasOoRDnsBVswOJ1OLV68RElJN8hmsxXovnMSabPKYgntsEEAAACEDoUTcuWwW+WwF+xp4rQYhVslhz1MNhunKAAAAEKPoXoAAAAA4AOFEwAAAAD4QOEEAAAAAD5QOAEAAACADxROAAAAAOADhRMAAAAA+EDhBAAAAAA+UDgBAAAAgA88XRRuxhilOV1KzXCFOhQAAACgUKFwgqRzRVO36SnauPdYqEMBAAAACh2G6kGSlOZ0eRVN8XExirRZQxQRAAAAUHhwxQleNoxKlMNuVaTNKovFEupwAAAAgJCjcIIXh90qh51TAwAAADiPoXoAAAAA4AOFEwAAAAD4QOEEAAAAAD5QOAEAAACADxROAAAAAOADhRMAAAAA+EDhBAAAAAA+8LCeIsYYozSnK+DbTc0I/DYBAACA4oLCqQgxxqjb9BRt3Hss1KEAAAAAJQpD9YqQNKcr6EVTfFyMIm3WoO4DAAAAKGq44lREbRiVKIc98AVOpM0qi8US8O0CAAAARRmFUxHlsFvlsPP2AQAAAAWBoXoAAAAA4AOFEwAAAAD4QOEEAAAAAD5QOAEAAACADxROAAAAAOADhRMAAAAA+EDhBAAAAAA+UDgBAAAAgA8UTkWEMUapGa5QhwEAAACUSGGhDgC+GWPUbXqKNu49FupQAAAAgBKJK05FQJrT5VE0xcfFKNJmDWFEAAAAQMnCFaciZsOoRFUobZfFYgl1KAAAAECJwRWnIsZht1I0AQAAAAWMwgkAAAAAfKBwAgAAAAAfKJwAAAAAwAcKJwAAAADwgcIJAAAAAHygcAIAAAAAHyicAAAAAMAHHoAbYsYYpTldufZJzch9OQAAAIDgonAKIWOMuk1P0ca9x0IdCgAAAIBcMFQvhNKcLr+Kpvi4GEXarEGMCAAAAEB2uOJUSGwYlSiHPfeiKNJmlcViKaCIAAAAAJxH4VRIOOxWOey8HQAAAEBhxFA9AAAAAPCBwgkAAAAAfKBwAgAAAAAfKJwAAAAAwAcKJwAAAADwgcIJAAAAAHwoFIXTyy+/rJo1ayoiIkItW7bU+vXrc+3/wQcfqH79+oqIiFDjxo312WefFVCkAAAAAEqikBdOc+bM0dChQ5WcnKxNmzapadOmSkpK0uHDh7Ptv3btWvXo0UP33nuvNm/erK5du6pr16764YcfCjhyAAAAACVFyAunyZMnq3///urbt68uu+wyTZ8+XQ6HQ2+99Va2/adNm6YOHTroscceU4MGDfT000/riiuu0L/+9a8CjhwAAABASREWyp1nZGRo48aNGjlypLutVKlSSkxMVEpKSrbrpKSkaOjQoR5tSUlJWrBgQbb909PTlZ6e7n594sQJSdLRo0fldDrzHbvT6VRqaqr+/PNP2Wy2fG0jNSNTWempkqQ///xTafaQvh2FRiByi+yR2+Aht8FBXoOH3AYPuQ0echscJTmvp06dkiQZY3z2Dek39T/++EMul0tVqlTxaK9SpYq2bduW7ToHDx7Mtv/Bgwez7T9hwgSNHTvWq71WrVr5jDo4Lp4a6ggAAACAkunUqVMqV65crn2K/SWOkSNHelyhysrK0tGjR1WhQgVZLJZ8b/fkyZOqUaOGfv31V0VFRQUiVPw/chs85DZ4yG1wkNfgIbfBQ26Dh9wGR0nOqzFGp06dUrVq1Xz2DWnhVLFiRVmtVh06dMij/dChQ4qNjc12ndjYWL/6h4eHKzw83KMtOjo6/0H/TVRUVIk7wQoKuQ0echs85DY4yGvwkNvgIbfBQ26Do6Tm1deVpvNCOjmE3W5X8+bNtWzZMndbVlaWli1bpoSEhGzXSUhI8OgvSUuXLs2xPwAAAABcqJAP1Rs6dKh69+6t+Ph4tWjRQlOnTtWZM2fUt29fSVKvXr1UvXp1TZgwQZL08MMPq23btnrhhRfUqVMnzZ49Wxs2bNC///3vUB4GAAAAgGIs5IVT9+7ddeTIEY0ePVoHDx7U5ZdfrkWLFrkngNi3b59KlfrfhbGrr75as2bN0qhRo/TEE0+oXr16WrBggRo1alSgcYeHhys5OdlrGCAuHLkNHnIbPOQ2OMhr8JDb4CG3wUNug4O85o3F5GXuPQAAAAAowUL+AFwAAAAAKOwonAAAAADABwonAAAAAPCBwgkAAAAAfCjRhdPLL7+smjVrKiIiQi1bttT69etz7f/BBx+ofv36ioiIUOPGjfXZZ595LDfGaPTo0apataoiIyOVmJioHTt2ePQ5evSoevbsqaioKEVHR+vee+/V6dOnA35soRTIvDqdTo0YMUKNGzdW6dKlVa1aNfXq1Uu///67xzZq1qwpi8Xi8TNx4sSgHF8oBfqc7dOnj1feOnTo4NGnJJyzUuBz+/e8nv+ZNGmSu09JOG/9yeuPP/6o2267zZ2XqVOn5mubZ8+e1cCBA1WhQgWVKVNGt912m9eD04uDQOd2woQJuvLKK1W2bFlVrlxZXbt21fbt2z36tGvXzuucvf/++wN9aCEX6NyOGTPGK2/169f36MN56y0vuc3uc9RisWjgwIHuPiXhvPUnr6+//rpat26tmJgYxcTEKDEx0as/32lzYEqo2bNnG7vdbt566y3z448/mv79+5vo6Ghz6NChbPuvWbPGWK1W89xzz5mffvrJjBo1ythsNvP999+7+0ycONGUK1fOLFiwwHz77bemS5cuplatWiYtLc3dp0OHDqZp06bm66+/NqtXrzZ169Y1PXr0CPrxFpRA5/X48eMmMTHRzJkzx2zbts2kpKSYFi1amObNm3tsJy4uzowbN84cOHDA/XP69OmgH29BCsY527t3b9OhQwePvB09etRjO8X9nDUmOLn9a04PHDhg3nrrLWOxWMyuXbvcfYr7eetvXtevX28effRR85///MfExsaaKVOm5Gub999/v6lRo4ZZtmyZ2bBhg7nqqqvM1VdfHazDDIlg5DYpKcnMmDHD/PDDD2bLli3mxhtvNBdffLHHOdm2bVvTv39/j3P2xIkTwTrMkAhGbpOTk03Dhg098nbkyBGPPpy33vKS28OHD3vkdenSpUaSWbFihbtPcT9v/c3rXXfdZV5++WWzefNms3XrVtOnTx9Trlw589tvv7n78J02eyW2cGrRooUZOHCg+7XL5TLVqlUzEyZMyLb/HXfcYTp16uTR1rJlS3PfffcZY4zJysoysbGxZtKkSe7lx48fN+Hh4eY///mPMcaYn376yUgy33zzjbvP559/biwWi9m/f3/Aji2UAp3X7Kxfv95IMnv37nW3xcXFZfuBWpwEI7e9e/c2N998c477LAnnrDEFc97efPPN5rrrrvNoK+7nrb95/auccuNrm8ePHzc2m8188MEH7j5bt241kkxKSsoFHE3hEozc/t3hw4eNJLNq1Sp3W9u2bc3DDz+cn5CLjGDkNjk52TRt2jTH9ThvA3fePvzww6ZOnTomKyvL3Vbcz9sLyasxxmRmZpqyZcuat99+2xjDd9rclMihehkZGdq4caMSExPdbaVKlVJiYqJSUlKyXSclJcWjvyQlJSW5++/evVsHDx706FOuXDm1bNnS3SclJUXR0dGKj49390lMTFSpUqW0bt26gB1fqAQjr9k5ceKELBaLoqOjPdonTpyoChUqqFmzZpo0aZIyMzPzfzCFTDBzu3LlSlWuXFmXXnqpHnjgAf35558e2yjO56xUMOftoUOHtHDhQt17771ey4rreZufvAZimxs3bpTT6fToU79+fV188cX53m9hE4zcZufEiROSpPLly3u0v//++6pYsaIaNWqkkSNHKjU1NWD7DLVg5nbHjh2qVq2aateurZ49e2rfvn3uZZy3gTnGjIwMvffee7rnnntksVg8lhXX8zYQeU1NTZXT6XT/rvOdNmdhoQ4gFP744w+5XC5VqVLFo71KlSratm1btuscPHgw2/4HDx50Lz/fllufypUreywPCwtT+fLl3X2KsmDk9e/Onj2rESNGqEePHoqKinK3Dx48WFdccYXKly+vtWvXauTIkTpw4IAmT558gUdVOAQrtx06dNCtt96qWrVqadeuXXriiSfUsWNHpaSkyGq1FvtzViqY8/btt99W2bJldeutt3q0F+fzNj95DcQ2Dx48KLvd7vWHldzen6ImGLn9u6ysLA0ZMkStWrVSo0aN3O133XWX4uLiVK1aNX333XcaMWKEtm/frvnz5wdkv6EWrNy2bNlSM2fO1KWXXqoDBw5o7Nixat26tX744QeVLVuW8zZA5+2CBQt0/Phx9enTx6O9OJ+3gcjriBEjVK1aNXehxHfanJXIwglFk9Pp1B133CFjjF599VWPZUOHDnX/u0mTJrLb7brvvvs0YcIEhYeHF3SoRcadd97p/nfjxo3VpEkT1alTRytXrlT79u1DGFnx8tZbb6lnz56KiIjwaOe8RWE1cOBA/fDDD/rqq6882gcMGOD+d+PGjVW1alW1b99eu3btUp06dQo6zCKjY8eO7n83adJELVu2VFxcnObOnZvtlWjkz5tvvqmOHTuqWrVqHu2ctzmbOHGiZs+erZUrV3r9PwreSuRQvYoVK8pqtXrNVnPo0CHFxsZmu05sbGyu/c//11efw4cPeyzPzMzU0aNHc9xvURKMvJ53vmjau3evli5d6nG1KTstW7ZUZmam9uzZ4/+BFELBzO1f1a5dWxUrVtTOnTvd2yjO56wU/NyuXr1a27dvV79+/XzGUpzO2/zkNRDbjI2NVUZGho4fPx6w/RY2wcjtXw0aNEiffvqpVqxYoYsuuijXvi1btpQk92dGURfs3J4XHR2tSy65xOOzlvP2wuzdu1dffPFFnj9rpeJx3l5IXp9//nlNnDhRS5YsUZMmTdztfKfNWYksnOx2u5o3b65ly5a527KysrRs2TIlJCRku05CQoJHf0launSpu3+tWrUUGxvr0efkyZNat26du09CQoKOHz+ujRs3uvssX75cWVlZ7l/ioiwYeZX+VzTt2LFDX3zxhSpUqOAzli1btqhUqVJel5GLqmDl9u9+++03/fnnn6patap7G8X5nJWCn9s333xTzZs3V9OmTX3GUpzO2/zkNRDbbN68uWw2m0ef7du3a9++ffneb2ETjNxK56YfHjRokP773/9q+fLlqlWrls91tmzZIknuz4yiLli5/bvTp09r165d7rxx3l74Mc6YMUOVK1dWp06dfPYtTudtfvP63HPP6emnn9aiRYs87lOS+E6bq1DPThEqs2fPNuHh4WbmzJnmp59+MgMGDDDR0dHm4MGDxhhj7r77bvP444+7+69Zs8aEhYWZ559/3mzdutUkJydnOx15dHS0+eijj8x3331nbr755mynbmzWrJlZt26d+eqrr0y9evWK1dSNgc5rRkaG6dKli7nooovMli1bPKYSTU9PN8YYs3btWjNlyhSzZcsWs2vXLvPee++ZSpUqmV69ehV8AoIo0Lk9deqUefTRR01KSorZvXu3+eKLL8wVV1xh6tWrZ86ePeveTnE/Z40JzueBMcacOHHCOBwO8+qrr3rtsySct/7mNT093WzevNls3rzZVK1a1Tz66KNm8+bNZseOHXnepjHnpnW++OKLzfLly82GDRtMQkKCSUhIKLgDLwDByO0DDzxgypUrZ1auXOnxWZuammqMMWbnzp1m3LhxZsOGDWb37t3mo48+MrVr1zZt2rQp2IMPsmDkdtiwYWblypVm9+7dZs2aNSYxMdFUrFjRHD582N2H8zZ/uTXm3CxyF198sRkxYoTXPkvCeetvXidOnGjsdruZN2+ex+/6qVOnPPqU9O+02SmxhZMxxrz00kvm4osvNna73bRo0cJ8/fXX7mVt27Y1vXv39ug/d+5cc8kllxi73W4aNmxoFi5c6LE8KyvLPPXUU6ZKlSomPDzctG/f3mzfvt2jz59//ml69OhhypQpY6Kiokzfvn09TtTiIJB53b17t5GU7c/5ZzRs3LjRtGzZ0pQrV85ERESYBg0amPHjx3t8+S8uApnb1NRUc8MNN5hKlSoZm81m4uLiTP/+/T2+gBpTMs5ZYwL/eWCMMa+99pqJjIw0x48f91pWUs5bf/Ka0+9727Zt87xNY4xJS0szDz74oImJiTEOh8Pccsst5sCBA8E8zJAIdG5z+qydMWOGMcaYffv2mTZt2pjy5cub8PBwU7duXfPYY48Vq+fhnBfo3Hbv3t1UrVrV2O12U716ddO9e3ezc+dOj31y3ub/M2Hx4sVGktd3LmNKznnrT17j4uKyzWtycrK7D99ps2cxxphgXtECAAAAgKKuRN7jBAAAAAD+oHACAAAAAB8onAAAAADABwonAAAAAPCBwgkAAAAAfKBwAgAAAAAfKJwAAAAAwAcKJwAAAADwgcIJABA0FotFCxYsKDTbCZQ9e/bIYrFoy5YtufZr166dhgwZUiAxAQCCi8IJAIqJgwcP6qGHHlLt2rUVHh6uGjVqqHPnzlq2bFmoQ8uzMWPG6PLLL/dqP3DggDp27FjwAeWgRo0aOnDggBo1aiRJWrlypSwWi44fP35B261Zs6YsFkuOP3369Lnw4C8gtqlTp4Zs/wAQamGhDgAAcOH27NmjVq1aKTo6WpMmTVLjxo3ldDq1ePFiDRw4UNu2bcvXdjMyMmS3273anU6nbDbbhYadZ7GxsQW2r7ywWq1Biembb76Ry+WSJK1du1a33Xabtm/frqioKElSZGSkX9vL6f0DAPiPK04AUAw8+OCDslgsWr9+vW677TZdcsklatiwoYYOHaqvv/7a3W/fvn26+eabVaZMGUVFRemOO+7QoUOH3MvPX/F54403VKtWLUVEREg6N1Tu1VdfVZcuXVS6dGk988wzkqSPPvpIV1xxhSIiIlS7dm2NHTtWmZmZOcY5YsQIXXLJJXI4HKpdu7aeeuopOZ1OSdLMmTM1duxYffvtt+4rLDNnznTv/69D9b7//ntdd911ioyMVIUKFTRgwACdPn3avbxPnz7q2rWrnn/+eVWtWlUVKlTQwIED3fv6uxMnTshqtWrDhg2SpKysLJUvX15XXXWVu897772nGjVqSPIcqrdnzx5de+21kqSYmBivK0NZWVkaPny4ypcvr9jYWI0ZMybH/FSqVEmxsbGKjY1V+fLlJUmVK1dWbGysbDab7r//flWvXl0Oh0ONGzfWf/7zH4/127Vrp0GDBmnIkCGqWLGikpKSJEkff/yx6tWrp4iICF177bV6++23va6QffXVV2rdurUiIyNVo0YNDR48WGfOnHFvd+/evXrkkUfc7w0AlDQUTgBQxB09elSLFi3SwIEDVbp0aa/l0dHRks59gb/55pt19OhRrVq1SkuXLtUvv/yi7t27e/TfuXOnPvzwQ82fP9/jHp4xY8bolltu0ffff6977rlHq1evVq9evfTwww/rp59+0muvvaaZM2e6i6rslC1bVjNnztRPP/2kadOm6fXXX9eUKVMkSd27d9ewYcPUsGFDHThwQAcOHPCKTZLOnDmjpKQkxcTE6JtvvtEHH3ygL774QoMGDfLot2LFCu3atUsrVqzQ22+/rZkzZ7oLsb8rV66cLr/8cq1cuVLSucLMYrFo8+bN7oJs1apVatu2rde6NWrU0IcffihJ2r59uw4cOKBp06a5l7/99tsqXbq01q1bp+eee07jxo3T0qVLc8xRTs6ePavmzZtr4cKF+uGHHzRgwADdfffdWr9+vUe/t99+W3a7XWvWrNH06dO1e/dudevWTV27dtW3336r++67T08++aTHOrt27VKHDh1022236bvvvtOcOXP01VdfuXM6f/58XXTRRRo3bpz7vQGAEscAAIq0devWGUlm/vz5ufZbsmSJsVqtZt++fe62H3/80Ugy69evN8YYk5ycbGw2mzl8+LDHupLMkCFDPNrat29vxo8f79H27rvvmqpVq3qs99///jfHmCZNmmSaN2/ufp2cnGyaNm3q1e+v2/n3v/9tYmJizOnTp93LFy5caEqVKmUOHjxojDGmd+/eJi4uzmRmZrr73H777aZ79+45xjJ06FDTqVMnY4wxU6dONd27dzdNmzY1n3/+uTHGmLp165p///vfxhhjdu/ebSSZzZs3G2OMWbFihZFkjh075rHNtm3bmmuuucaj7corrzQjRozIMY7zctrmX3Xq1MkMGzbMY3/NmjXz6DNixAjTqFEjj7Ynn3zSY9v33nuvGTBggEef1atXm1KlSpm0tDRjjDFxcXFmypQpPuMGgOKKe5wAoIgzxuSp39atW1WjRg33cDNJuuyyyxQdHa2tW7fqyiuvlCTFxcWpUqVKXuvHx8d7vP7222+1Zs0ajytMLpdLZ8+eVWpqqhwOh9c25syZoxdffFG7du3S6dOnlZmZ6b5/J6+2bt2qpk2belxda9WqlbKysrR9+3ZVqVJFktSwYUNZrVZ3n6pVq+r777/Pcbtt27bVm2++KZfLpVWrVumGG25QbGysVq5cqSZNmmjnzp1q166dX7FKUpMmTTxeV61aVYcPH/Z7Oy6XS+PHj9fcuXO1f/9+ZWRkKD093SvPzZs393i9fft293t7XosWLTxef/vtt/ruu+/0/vvvu9uMMcrKytLu3bvVoEEDv+MFgOKGwgkAirh69erJYrHkewKIv8tuuF927adPn9bYsWN16623evU9f2/UX6WkpKhnz54aO3askpKSVK5cOc2ePVsvvPBCQOL+u79PXmGxWJSVlZVj/zZt2ujUqVPatGmTvvzyS40fP16xsbGaOHGimjZtqmrVqqlevXpBjyMnkyZN0rRp0zR16lQ1btxYpUuX1pAhQ5SRkeHRL6f3LzenT5/Wfffdp8GDB3stu/jii/3eHgAURxROAFDElS9fXklJSXr55Zc1ePBgry/Ox48fV3R0tBo0aKBff/1Vv/76q/uq008//aTjx4/rsssu83u/V1xxhbZv3666devmqf/atWsVFxfncX/N3r17PfrY7Xb3rHI5adCggWbOnKkzZ864j3XNmjUqVaqULr30Uj+P4n+io6PVpEkT/etf/5LNZlP9+vVVuXJlde/eXZ9++mm29zf9NW5JPmO/EGvWrNHNN9+sf/zjH5LO3bP2888/+3zvLr30Un322Wcebd98843H6yuuuEI//fRTru9lXt4bACjOmBwCAIqBl19+WS6XSy1atNCHH36oHTt2aOvWrXrxxReVkJAgSUpMTFTjxo3Vs2dPbdq0SevXr1evXr3Utm1br2F4eTF69Gi98847Gjt2rH788Udt3bpVs2fP1qhRo7LtX69ePe3bt0+zZ8/Wrl279OKLL+q///2vR5+aNWtq9+7d2rJli/744w+lp6d7badnz56KiIhQ79699cMPP2jFihV66KGHdPfdd7uH6eVXu3bt9P7777uLpPLly6tBgwaaM2dOroVTXFycLBaLPv30Ux05csRjhr9AqVevnpYuXaq1a9dq69atuu+++zxmRMzJfffdp23btmnEiBH6+eefNXfuXI/ZCqVzsx2uXbtWgwYN0pYtW7Rjxw599NFHHhNu1KxZU19++aX279+vP/74I+DHBwCFHYUTABQDtWvX1qZNm3Tttddq2LBhatSoka6//notW7ZMr776qqRzX5I/+ugjxcTEqE2bNkpMTFTt2rU1Z86cfO0zKSlJn376qZYsWaIrr7xSV111laZMmaK4uLhs+3fp0kWPPPKIBg0apMsvv1xr167VU0895dHntttuU4cOHXTttdeqUqVKXtNtS5LD4dDixYt19OhRXXnllerWrZvat2+vf/3rX/k6jr9q27atXC6Xx71M7dq182r7u+rVq2vs2LF6/PHHVaVKFa8Z/gJh1KhRuuKKK5SUlKR27dopNjZWXbt29blerVq1NG/ePM2fP19NmjTRq6++6r7qFx4eLuncfVirVq3Szz//rNatW6tZs2YaPXq0qlWr5t7OuHHjtGfPHtWpUyfbe+AAoLizmLzeVQwAAIqFZ555RtOnT9evv/4a6lAAoMjgHicAAIq5V155RVdeeaUqVKigNWvWaNKkSUG5KgYAxRmFEwAAxdyOHTv0z3/+U0ePHtXFF1+sYcOGaeTIkaEOCwCKFIbqAQAAAIAPTA4BAAAAAD5QOAEAAACADxROAAAAAOADhRMAAAAA+EDhBAAAAAA+UDgBAAAAgA8UTgAAAADgA4UTAAAAAPjwf23wf8o5l/yXAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 1000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Assuming that rolling data columns end with '_rolling'\n",
    "\n",
    "rolling_columns = [col for col in matches_rolling.columns if col.endswith('_rolling')]\n",
    "\n",
    "# Assuming 'merged_data' is your DataFrame, 'features' is a list of your feature names, and 'target' is your target column name\n",
    "features_target = matches_rolling[rolling_columns + ['target']]\n",
    "correlation_matrix = features_target.corr()\n",
    "\n",
    "# Extract the absolute values of the correlations with the target variable, and drop the target-target correlation\n",
    "target_correlation = correlation_matrix['target'].abs().drop('target')\n",
    "\n",
    "# Create a cumulative distribution plot for the correlation values\n",
    "plt.figure(figsize=(10, 6))  # Adjust the figure size as needed\n",
    "sns.ecdfplot(target_correlation)\n",
    "plt.title(\"Cumulative Distribution of Correlation with Target\")\n",
    "plt.xlabel(\"Correlation with Target\")\n",
    "plt.ylabel(\"Proportion of Features\")\n",
    "plt.grid(True)\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The above plot provides a visual representation of how the features correlate with the target. We will select the top 50% of features, which in our case corresponds to a threshold of approximately 0.13. Feel free to adjust this threshold based on your analysis and understanding of the dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2069,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Selected Features based on Correlation Threshold: ['gf_rolling', 'xg_x_rolling', 'poss_rolling', 'sh_rolling', 'sot_rolling', 'npxg_rolling', 'sota_rolling', 'xag_rolling', 'xa_rolling', 'kp_rolling', '1/3_rolling', 'ppa_rolling', 'prgp_rolling', 'attempted passes_rolling', 'completed passes_rolling', 'tkl+int_rolling', 'mid 3rd_rolling', 'att pen_rolling', 'att 3rd_rolling']\n",
      "Number of Selected Features: 19\n"
     ]
    }
   ],
   "source": [
    "\n",
    "# Set a threshold for selecting features\n",
    "threshold = 0.13  # This is an example value; you should choose your own threshold\n",
    "\n",
    "selected_features = target_correlation[target_correlation > threshold].index.tolist()\n",
    "\n",
    "print(\"Selected Features based on Correlation Threshold:\", selected_features)\n",
    "print(\"Number of Selected Features:\", len(selected_features))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.4 Merge Data of opponent onto single row"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A crucial aspect of analysing football matches is considering the performance of both teams involved. Dataquest's implementation [1] looks at a football match from the perspective of one team. That only paints part of the picture. One team's numbers might not be able to tell our model much without context of how our opponent is fairing as well. To give us both perspectives of the match, we will merge the opponents rolling data onto our match row data. This encompasses a more comprehensive dataset that reflects both sides of each match."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2070,
   "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>date</th>\n",
       "      <th>time</th>\n",
       "      <th>round</th>\n",
       "      <th>day</th>\n",
       "      <th>venue</th>\n",
       "      <th>result</th>\n",
       "      <th>gf</th>\n",
       "      <th>ga</th>\n",
       "      <th>opponent</th>\n",
       "      <th>xg_x</th>\n",
       "      <th>xga</th>\n",
       "      <th>poss</th>\n",
       "      <th>formation</th>\n",
       "      <th>sh</th>\n",
       "      <th>sot</th>\n",
       "      <th>dist</th>\n",
       "      <th>npxg</th>\n",
       "      <th>g-xg</th>\n",
       "      <th>sota</th>\n",
       "      <th>saves</th>\n",
       "      <th>save%</th>\n",
       "      <th>att (gk)</th>\n",
       "      <th>thr</th>\n",
       "      <th>psxg</th>\n",
       "      <th>psxg+/-</th>\n",
       "      <th>ast</th>\n",
       "      <th>xag</th>\n",
       "      <th>xa</th>\n",
       "      <th>kp</th>\n",
       "      <th>1/3</th>\n",
       "      <th>ppa</th>\n",
       "      <th>crspa</th>\n",
       "      <th>prgp</th>\n",
       "      <th>attempted passes</th>\n",
       "      <th>completed passes</th>\n",
       "      <th>corner kicks</th>\n",
       "      <th>shots blocked</th>\n",
       "      <th>tkl+int</th>\n",
       "      <th>err</th>\n",
       "      <th>mid 3rd</th>\n",
       "      <th>att pen</th>\n",
       "      <th>att 3rd</th>\n",
       "      <th>fls</th>\n",
       "      <th>fld</th>\n",
       "      <th>off</th>\n",
       "      <th>season</th>\n",
       "      <th>team</th>\n",
       "      <th>target</th>\n",
       "      <th>venue_code</th>\n",
       "      <th>opp_code</th>\n",
       "      <th>team_code</th>\n",
       "      <th>gf_rolling</th>\n",
       "      <th>ga_rolling</th>\n",
       "      <th>xg_x_rolling</th>\n",
       "      <th>xga_rolling</th>\n",
       "      <th>poss_rolling</th>\n",
       "      <th>sh_rolling</th>\n",
       "      <th>sot_rolling</th>\n",
       "      <th>dist_rolling</th>\n",
       "      <th>npxg_rolling</th>\n",
       "      <th>g-xg_rolling</th>\n",
       "      <th>sota_rolling</th>\n",
       "      <th>saves_rolling</th>\n",
       "      <th>save%_rolling</th>\n",
       "      <th>att (gk)_rolling</th>\n",
       "      <th>thr_rolling</th>\n",
       "      <th>psxg_rolling</th>\n",
       "      <th>psxg+/-_rolling</th>\n",
       "      <th>ast_rolling</th>\n",
       "      <th>xag_rolling</th>\n",
       "      <th>xa_rolling</th>\n",
       "      <th>kp_rolling</th>\n",
       "      <th>1/3_rolling</th>\n",
       "      <th>ppa_rolling</th>\n",
       "      <th>crspa_rolling</th>\n",
       "      <th>prgp_rolling</th>\n",
       "      <th>attempted passes_rolling</th>\n",
       "      <th>completed passes_rolling</th>\n",
       "      <th>corner kicks_rolling</th>\n",
       "      <th>shots blocked_rolling</th>\n",
       "      <th>tkl+int_rolling</th>\n",
       "      <th>err_rolling</th>\n",
       "      <th>mid 3rd_rolling</th>\n",
       "      <th>att pen_rolling</th>\n",
       "      <th>att 3rd_rolling</th>\n",
       "      <th>fls_rolling</th>\n",
       "      <th>fld_rolling</th>\n",
       "      <th>off_rolling</th>\n",
       "      <th>opponent_normalized</th>\n",
       "      <th>gf_rolling_opponent</th>\n",
       "      <th>xg_x_rolling_opponent</th>\n",
       "      <th>poss_rolling_opponent</th>\n",
       "      <th>sh_rolling_opponent</th>\n",
       "      <th>sot_rolling_opponent</th>\n",
       "      <th>npxg_rolling_opponent</th>\n",
       "      <th>sota_rolling_opponent</th>\n",
       "      <th>xag_rolling_opponent</th>\n",
       "      <th>xa_rolling_opponent</th>\n",
       "      <th>kp_rolling_opponent</th>\n",
       "      <th>1/3_rolling_opponent</th>\n",
       "      <th>ppa_rolling_opponent</th>\n",
       "      <th>prgp_rolling_opponent</th>\n",
       "      <th>attempted passes_rolling_opponent</th>\n",
       "      <th>completed passes_rolling_opponent</th>\n",
       "      <th>tkl+int_rolling_opponent</th>\n",
       "      <th>mid 3rd_rolling_opponent</th>\n",
       "      <th>att pen_rolling_opponent</th>\n",
       "      <th>att 3rd_rolling_opponent</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2020-10-25</td>\n",
       "      <td>19:15</td>\n",
       "      <td>Matchweek 6</td>\n",
       "      <td>Sun</td>\n",
       "      <td>Home</td>\n",
       "      <td>L</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Leicester City</td>\n",
       "      <td>0.7</td>\n",
       "      <td>0.7</td>\n",
       "      <td>56.0</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>12.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>0.7</td>\n",
       "      <td>-0.7</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>50.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.3</td>\n",
       "      <td>-0.7</td>\n",
       "      <td>0</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.7</td>\n",
       "      <td>10.0</td>\n",
       "      <td>41.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>50.0</td>\n",
       "      <td>621.0</td>\n",
       "      <td>525.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>27</td>\n",
       "      <td>0.0</td>\n",
       "      <td>437.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>153.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>11</td>\n",
       "      <td>0</td>\n",
       "      <td>1.6</td>\n",
       "      <td>1.2</td>\n",
       "      <td>1.16</td>\n",
       "      <td>1.26</td>\n",
       "      <td>51.2</td>\n",
       "      <td>8.2</td>\n",
       "      <td>3.8</td>\n",
       "      <td>15.72</td>\n",
       "      <td>1.16</td>\n",
       "      <td>0.44</td>\n",
       "      <td>3.8</td>\n",
       "      <td>2.6</td>\n",
       "      <td>68.5</td>\n",
       "      <td>36.0</td>\n",
       "      <td>4.2</td>\n",
       "      <td>1.04</td>\n",
       "      <td>-0.16</td>\n",
       "      <td>1.2</td>\n",
       "      <td>0.80</td>\n",
       "      <td>0.88</td>\n",
       "      <td>6.0</td>\n",
       "      <td>29.0</td>\n",
       "      <td>7.4</td>\n",
       "      <td>1.6</td>\n",
       "      <td>36.6</td>\n",
       "      <td>608.0</td>\n",
       "      <td>511.4</td>\n",
       "      <td>4.2</td>\n",
       "      <td>3.8</td>\n",
       "      <td>20.8</td>\n",
       "      <td>0.6</td>\n",
       "      <td>317.8</td>\n",
       "      <td>18.2</td>\n",
       "      <td>123.2</td>\n",
       "      <td>8.6</td>\n",
       "      <td>11.6</td>\n",
       "      <td>2.8</td>\n",
       "      <td>Leicester City</td>\n",
       "      <td>2.4</td>\n",
       "      <td>1.62</td>\n",
       "      <td>55.6</td>\n",
       "      <td>8.8</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.84</td>\n",
       "      <td>4.2</td>\n",
       "      <td>0.54</td>\n",
       "      <td>0.60</td>\n",
       "      <td>6.0</td>\n",
       "      <td>29.6</td>\n",
       "      <td>9.2</td>\n",
       "      <td>41.6</td>\n",
       "      <td>561.6</td>\n",
       "      <td>448.6</td>\n",
       "      <td>30.0</td>\n",
       "      <td>306.2</td>\n",
       "      <td>21.8</td>\n",
       "      <td>144.4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2020-11-01</td>\n",
       "      <td>16:30</td>\n",
       "      <td>Matchweek 7</td>\n",
       "      <td>Sun</td>\n",
       "      <td>Away</td>\n",
       "      <td>W</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Manchester Utd</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.4</td>\n",
       "      <td>46.0</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>6.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>24.6</td>\n",
       "      <td>0.3</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>43.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>0.4</td>\n",
       "      <td>5.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>40.0</td>\n",
       "      <td>526.0</td>\n",
       "      <td>438.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>32</td>\n",
       "      <td>0.0</td>\n",
       "      <td>298.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>130.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>15</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.92</td>\n",
       "      <td>1.38</td>\n",
       "      <td>51.6</td>\n",
       "      <td>8.0</td>\n",
       "      <td>3.6</td>\n",
       "      <td>15.90</td>\n",
       "      <td>0.92</td>\n",
       "      <td>0.08</td>\n",
       "      <td>3.8</td>\n",
       "      <td>2.4</td>\n",
       "      <td>58.5</td>\n",
       "      <td>35.0</td>\n",
       "      <td>4.8</td>\n",
       "      <td>1.08</td>\n",
       "      <td>-0.32</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.76</td>\n",
       "      <td>0.82</td>\n",
       "      <td>6.4</td>\n",
       "      <td>29.6</td>\n",
       "      <td>7.4</td>\n",
       "      <td>2.4</td>\n",
       "      <td>37.2</td>\n",
       "      <td>598.6</td>\n",
       "      <td>498.6</td>\n",
       "      <td>5.4</td>\n",
       "      <td>3.8</td>\n",
       "      <td>20.2</td>\n",
       "      <td>0.6</td>\n",
       "      <td>336.6</td>\n",
       "      <td>17.2</td>\n",
       "      <td>124.8</td>\n",
       "      <td>8.8</td>\n",
       "      <td>11.0</td>\n",
       "      <td>2.6</td>\n",
       "      <td>Manchester United</td>\n",
       "      <td>1.8</td>\n",
       "      <td>1.30</td>\n",
       "      <td>55.0</td>\n",
       "      <td>13.6</td>\n",
       "      <td>4.4</td>\n",
       "      <td>0.86</td>\n",
       "      <td>4.6</td>\n",
       "      <td>0.70</td>\n",
       "      <td>0.66</td>\n",
       "      <td>11.2</td>\n",
       "      <td>37.4</td>\n",
       "      <td>8.8</td>\n",
       "      <td>46.2</td>\n",
       "      <td>597.2</td>\n",
       "      <td>494.2</td>\n",
       "      <td>26.0</td>\n",
       "      <td>333.0</td>\n",
       "      <td>24.4</td>\n",
       "      <td>199.2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2020-11-08</td>\n",
       "      <td>19:15</td>\n",
       "      <td>Matchweek 8</td>\n",
       "      <td>Sun</td>\n",
       "      <td>Home</td>\n",
       "      <td>L</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>Aston Villa</td>\n",
       "      <td>1.5</td>\n",
       "      <td>1.8</td>\n",
       "      <td>59.0</td>\n",
       "      <td>3-4-3</td>\n",
       "      <td>13.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>16.4</td>\n",
       "      <td>1.5</td>\n",
       "      <td>-1.5</td>\n",
       "      <td>5.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>40.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.6</td>\n",
       "      <td>-0.4</td>\n",
       "      <td>0</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.6</td>\n",
       "      <td>8.0</td>\n",
       "      <td>30.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>43.0</td>\n",
       "      <td>588.0</td>\n",
       "      <td>481.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>15</td>\n",
       "      <td>0.0</td>\n",
       "      <td>288.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>153.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.8</td>\n",
       "      <td>1.2</td>\n",
       "      <td>0.90</td>\n",
       "      <td>1.06</td>\n",
       "      <td>48.4</td>\n",
       "      <td>7.8</td>\n",
       "      <td>3.2</td>\n",
       "      <td>17.86</td>\n",
       "      <td>0.76</td>\n",
       "      <td>-0.10</td>\n",
       "      <td>3.8</td>\n",
       "      <td>2.6</td>\n",
       "      <td>68.5</td>\n",
       "      <td>37.2</td>\n",
       "      <td>3.6</td>\n",
       "      <td>0.96</td>\n",
       "      <td>-0.24</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.58</td>\n",
       "      <td>0.58</td>\n",
       "      <td>6.2</td>\n",
       "      <td>26.8</td>\n",
       "      <td>5.6</td>\n",
       "      <td>2.2</td>\n",
       "      <td>34.8</td>\n",
       "      <td>561.8</td>\n",
       "      <td>466.4</td>\n",
       "      <td>4.6</td>\n",
       "      <td>3.0</td>\n",
       "      <td>22.8</td>\n",
       "      <td>0.4</td>\n",
       "      <td>311.6</td>\n",
       "      <td>14.4</td>\n",
       "      <td>117.6</td>\n",
       "      <td>9.0</td>\n",
       "      <td>10.8</td>\n",
       "      <td>1.6</td>\n",
       "      <td>Aston Villa</td>\n",
       "      <td>2.8</td>\n",
       "      <td>1.62</td>\n",
       "      <td>42.8</td>\n",
       "      <td>14.0</td>\n",
       "      <td>6.4</td>\n",
       "      <td>1.48</td>\n",
       "      <td>5.6</td>\n",
       "      <td>1.28</td>\n",
       "      <td>1.04</td>\n",
       "      <td>11.2</td>\n",
       "      <td>21.2</td>\n",
       "      <td>8.2</td>\n",
       "      <td>30.0</td>\n",
       "      <td>374.6</td>\n",
       "      <td>264.4</td>\n",
       "      <td>29.2</td>\n",
       "      <td>192.0</td>\n",
       "      <td>28.8</td>\n",
       "      <td>131.6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2020-11-22</td>\n",
       "      <td>16:30</td>\n",
       "      <td>Matchweek 9</td>\n",
       "      <td>Sun</td>\n",
       "      <td>Away</td>\n",
       "      <td>D</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Leeds United</td>\n",
       "      <td>0.9</td>\n",
       "      <td>1.8</td>\n",
       "      <td>34.0</td>\n",
       "      <td>4-2-3-1</td>\n",
       "      <td>9.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>19.6</td>\n",
       "      <td>0.9</td>\n",
       "      <td>-0.9</td>\n",
       "      <td>4.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0</td>\n",
       "      <td>0.7</td>\n",
       "      <td>0.5</td>\n",
       "      <td>6.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>29.0</td>\n",
       "      <td>343.0</td>\n",
       "      <td>236.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>21</td>\n",
       "      <td>0.0</td>\n",
       "      <td>184.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>69.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>0.6</td>\n",
       "      <td>1.2</td>\n",
       "      <td>0.94</td>\n",
       "      <td>0.88</td>\n",
       "      <td>53.4</td>\n",
       "      <td>9.6</td>\n",
       "      <td>3.0</td>\n",
       "      <td>18.14</td>\n",
       "      <td>0.80</td>\n",
       "      <td>-0.34</td>\n",
       "      <td>3.2</td>\n",
       "      <td>2.2</td>\n",
       "      <td>64.0</td>\n",
       "      <td>35.2</td>\n",
       "      <td>2.6</td>\n",
       "      <td>0.76</td>\n",
       "      <td>-0.24</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.54</td>\n",
       "      <td>0.58</td>\n",
       "      <td>7.2</td>\n",
       "      <td>30.6</td>\n",
       "      <td>6.2</td>\n",
       "      <td>2.4</td>\n",
       "      <td>40.8</td>\n",
       "      <td>594.4</td>\n",
       "      <td>500.2</td>\n",
       "      <td>4.6</td>\n",
       "      <td>3.8</td>\n",
       "      <td>22.2</td>\n",
       "      <td>0.2</td>\n",
       "      <td>334.4</td>\n",
       "      <td>17.4</td>\n",
       "      <td>139.4</td>\n",
       "      <td>9.8</td>\n",
       "      <td>11.2</td>\n",
       "      <td>0.8</td>\n",
       "      <td>Leeds United</td>\n",
       "      <td>1.2</td>\n",
       "      <td>1.80</td>\n",
       "      <td>61.4</td>\n",
       "      <td>14.4</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1.80</td>\n",
       "      <td>4.2</td>\n",
       "      <td>1.36</td>\n",
       "      <td>1.22</td>\n",
       "      <td>11.2</td>\n",
       "      <td>36.4</td>\n",
       "      <td>11.6</td>\n",
       "      <td>55.2</td>\n",
       "      <td>587.0</td>\n",
       "      <td>475.0</td>\n",
       "      <td>30.6</td>\n",
       "      <td>318.2</td>\n",
       "      <td>28.6</td>\n",
       "      <td>180.6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2020-11-29</td>\n",
       "      <td>19:15</td>\n",
       "      <td>Matchweek 10</td>\n",
       "      <td>Sun</td>\n",
       "      <td>Home</td>\n",
       "      <td>L</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>1.5</td>\n",
       "      <td>1.8</td>\n",
       "      <td>56.0</td>\n",
       "      <td>4-2-3-1</td>\n",
       "      <td>13.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>15.6</td>\n",
       "      <td>1.5</td>\n",
       "      <td>-0.5</td>\n",
       "      <td>5.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>60.0</td>\n",
       "      <td>17.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1.8</td>\n",
       "      <td>-0.2</td>\n",
       "      <td>1</td>\n",
       "      <td>0.8</td>\n",
       "      <td>1.6</td>\n",
       "      <td>6.0</td>\n",
       "      <td>49.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>63.0</td>\n",
       "      <td>557.0</td>\n",
       "      <td>447.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>22</td>\n",
       "      <td>0.0</td>\n",
       "      <td>307.0</td>\n",
       "      <td>35.0</td>\n",
       "      <td>219.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2021</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>23</td>\n",
       "      <td>0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.00</td>\n",
       "      <td>1.20</td>\n",
       "      <td>47.4</td>\n",
       "      <td>10.2</td>\n",
       "      <td>2.4</td>\n",
       "      <td>18.80</td>\n",
       "      <td>0.86</td>\n",
       "      <td>-0.80</td>\n",
       "      <td>3.6</td>\n",
       "      <td>2.8</td>\n",
       "      <td>74.0</td>\n",
       "      <td>32.8</td>\n",
       "      <td>2.6</td>\n",
       "      <td>0.84</td>\n",
       "      <td>0.04</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.56</td>\n",
       "      <td>0.56</td>\n",
       "      <td>7.2</td>\n",
       "      <td>25.4</td>\n",
       "      <td>6.0</td>\n",
       "      <td>2.2</td>\n",
       "      <td>36.8</td>\n",
       "      <td>514.8</td>\n",
       "      <td>418.2</td>\n",
       "      <td>4.8</td>\n",
       "      <td>4.4</td>\n",
       "      <td>22.8</td>\n",
       "      <td>0.0</td>\n",
       "      <td>289.2</td>\n",
       "      <td>17.2</td>\n",
       "      <td>119.4</td>\n",
       "      <td>10.8</td>\n",
       "      <td>11.4</td>\n",
       "      <td>1.2</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.94</td>\n",
       "      <td>47.8</td>\n",
       "      <td>13.2</td>\n",
       "      <td>4.4</td>\n",
       "      <td>0.94</td>\n",
       "      <td>2.2</td>\n",
       "      <td>0.74</td>\n",
       "      <td>0.62</td>\n",
       "      <td>10.2</td>\n",
       "      <td>37.0</td>\n",
       "      <td>5.8</td>\n",
       "      <td>40.8</td>\n",
       "      <td>511.2</td>\n",
       "      <td>414.6</td>\n",
       "      <td>26.8</td>\n",
       "      <td>306.2</td>\n",
       "      <td>20.8</td>\n",
       "      <td>148.6</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        date   time         round  day venue result   gf   ga        opponent  \\\n",
       "0 2020-10-25  19:15   Matchweek 6  Sun  Home      L  0.0  1.0  Leicester City   \n",
       "1 2020-11-01  16:30   Matchweek 7  Sun  Away      W  1.0  0.0  Manchester Utd   \n",
       "2 2020-11-08  19:15   Matchweek 8  Sun  Home      L  0.0  3.0     Aston Villa   \n",
       "3 2020-11-22  16:30   Matchweek 9  Sun  Away      D  0.0  0.0    Leeds United   \n",
       "4 2020-11-29  19:15  Matchweek 10  Sun  Home      L  1.0  2.0          Wolves   \n",
       "\n",
       "   xg_x  xga  poss formation    sh  sot  dist  npxg  g-xg  sota  saves  save%  \\\n",
       "0   0.7  0.7  56.0     3-4-3  12.0  4.0  15.0   0.7  -0.7   2.0    1.0   50.0   \n",
       "1   1.0  0.4  46.0     3-4-3   6.0  1.0  24.6   0.3   0.0   2.0    2.0  100.0   \n",
       "2   1.5  1.8  59.0     3-4-3  13.0  2.0  16.4   1.5  -1.5   5.0    3.0   40.0   \n",
       "3   0.9  1.8  34.0   4-2-3-1   9.0  2.0  19.6   0.9  -0.9   4.0    4.0  100.0   \n",
       "4   1.5  1.8  56.0   4-2-3-1  13.0  2.0  15.6   1.5  -0.5   5.0    3.0   60.0   \n",
       "\n",
       "   att (gk)  thr  psxg  psxg+/-  ast  xag   xa    kp   1/3   ppa  crspa  prgp  \\\n",
       "0      24.0  6.0   0.3     -0.7    0  0.6  0.7  10.0  41.0  11.0    5.0  50.0   \n",
       "1      43.0  0.0   0.1      0.1    0  0.2  0.4   5.0  24.0   7.0    2.0  40.0   \n",
       "2      24.0  1.0   1.6     -0.4    0  0.6  0.6   8.0  30.0   4.0    1.0  43.0   \n",
       "3      25.0  2.0   0.8      0.8    0  0.7  0.5   6.0  15.0   3.0    0.0  29.0   \n",
       "4      17.0  5.0   1.8     -0.2    1  0.8  1.6   6.0  49.0  15.0    2.0  63.0   \n",
       "\n",
       "   attempted passes  completed passes  corner kicks  shots blocked  tkl+int  \\\n",
       "0             621.0             525.0           9.0            1.0       27   \n",
       "1             526.0             438.0           3.0            2.0       32   \n",
       "2             588.0             481.0           3.0           10.0       15   \n",
       "3             343.0             236.0           3.0            5.0       21   \n",
       "4             557.0             447.0           8.0            4.0       22   \n",
       "\n",
       "   err  mid 3rd  att pen  att 3rd   fls   fld  off  season     team  target  \\\n",
       "0  0.0    437.0     21.0    153.0  13.0   9.0  1.0    2021  Arsenal       0   \n",
       "1  0.0    298.0     14.0    130.0  12.0  11.0  0.0    2021  Arsenal       1   \n",
       "2  0.0    288.0     25.0    153.0  11.0  13.0  1.0    2021  Arsenal       0   \n",
       "3  0.0    184.0     13.0     69.0   8.0   9.0  2.0    2021  Arsenal       0   \n",
       "4  0.0    307.0     35.0    219.0  13.0  12.0  2.0    2021  Arsenal       0   \n",
       "\n",
       "   venue_code  opp_code  team_code  gf_rolling  ga_rolling  xg_x_rolling  \\\n",
       "0           1        11          0         1.6         1.2          1.16   \n",
       "1           0        15          0         1.0         1.4          0.92   \n",
       "2           1         1          0         0.8         1.2          0.90   \n",
       "3           0        10          0         0.6         1.2          0.94   \n",
       "4           1        23          0         0.2         1.0          1.00   \n",
       "\n",
       "   xga_rolling  poss_rolling  sh_rolling  sot_rolling  dist_rolling  \\\n",
       "0         1.26          51.2         8.2          3.8         15.72   \n",
       "1         1.38          51.6         8.0          3.6         15.90   \n",
       "2         1.06          48.4         7.8          3.2         17.86   \n",
       "3         0.88          53.4         9.6          3.0         18.14   \n",
       "4         1.20          47.4        10.2          2.4         18.80   \n",
       "\n",
       "   npxg_rolling  g-xg_rolling  sota_rolling  saves_rolling  save%_rolling  \\\n",
       "0          1.16          0.44           3.8            2.6           68.5   \n",
       "1          0.92          0.08           3.8            2.4           58.5   \n",
       "2          0.76         -0.10           3.8            2.6           68.5   \n",
       "3          0.80         -0.34           3.2            2.2           64.0   \n",
       "4          0.86         -0.80           3.6            2.8           74.0   \n",
       "\n",
       "   att (gk)_rolling  thr_rolling  psxg_rolling  psxg+/-_rolling  ast_rolling  \\\n",
       "0              36.0          4.2          1.04            -0.16          1.2   \n",
       "1              35.0          4.8          1.08            -0.32          0.8   \n",
       "2              37.2          3.6          0.96            -0.24          0.4   \n",
       "3              35.2          2.6          0.76            -0.24          0.4   \n",
       "4              32.8          2.6          0.84             0.04          0.0   \n",
       "\n",
       "   xag_rolling  xa_rolling  kp_rolling  1/3_rolling  ppa_rolling  \\\n",
       "0         0.80        0.88         6.0         29.0          7.4   \n",
       "1         0.76        0.82         6.4         29.6          7.4   \n",
       "2         0.58        0.58         6.2         26.8          5.6   \n",
       "3         0.54        0.58         7.2         30.6          6.2   \n",
       "4         0.56        0.56         7.2         25.4          6.0   \n",
       "\n",
       "   crspa_rolling  prgp_rolling  attempted passes_rolling  \\\n",
       "0            1.6          36.6                     608.0   \n",
       "1            2.4          37.2                     598.6   \n",
       "2            2.2          34.8                     561.8   \n",
       "3            2.4          40.8                     594.4   \n",
       "4            2.2          36.8                     514.8   \n",
       "\n",
       "   completed passes_rolling  corner kicks_rolling  shots blocked_rolling  \\\n",
       "0                     511.4                   4.2                    3.8   \n",
       "1                     498.6                   5.4                    3.8   \n",
       "2                     466.4                   4.6                    3.0   \n",
       "3                     500.2                   4.6                    3.8   \n",
       "4                     418.2                   4.8                    4.4   \n",
       "\n",
       "   tkl+int_rolling  err_rolling  mid 3rd_rolling  att pen_rolling  \\\n",
       "0             20.8          0.6            317.8             18.2   \n",
       "1             20.2          0.6            336.6             17.2   \n",
       "2             22.8          0.4            311.6             14.4   \n",
       "3             22.2          0.2            334.4             17.4   \n",
       "4             22.8          0.0            289.2             17.2   \n",
       "\n",
       "   att 3rd_rolling  fls_rolling  fld_rolling  off_rolling  \\\n",
       "0            123.2          8.6         11.6          2.8   \n",
       "1            124.8          8.8         11.0          2.6   \n",
       "2            117.6          9.0         10.8          1.6   \n",
       "3            139.4          9.8         11.2          0.8   \n",
       "4            119.4         10.8         11.4          1.2   \n",
       "\n",
       "       opponent_normalized  gf_rolling_opponent  xg_x_rolling_opponent  \\\n",
       "0           Leicester City                  2.4                   1.62   \n",
       "1        Manchester United                  1.8                   1.30   \n",
       "2              Aston Villa                  2.8                   1.62   \n",
       "3             Leeds United                  1.2                   1.80   \n",
       "4  Wolverhampton Wanderers                  1.0                   0.94   \n",
       "\n",
       "   poss_rolling_opponent  sh_rolling_opponent  sot_rolling_opponent  \\\n",
       "0                   55.6                  8.8                   4.0   \n",
       "1                   55.0                 13.6                   4.4   \n",
       "2                   42.8                 14.0                   6.4   \n",
       "3                   61.4                 14.4                   5.0   \n",
       "4                   47.8                 13.2                   4.4   \n",
       "\n",
       "   npxg_rolling_opponent  sota_rolling_opponent  xag_rolling_opponent  \\\n",
       "0                   0.84                    4.2                  0.54   \n",
       "1                   0.86                    4.6                  0.70   \n",
       "2                   1.48                    5.6                  1.28   \n",
       "3                   1.80                    4.2                  1.36   \n",
       "4                   0.94                    2.2                  0.74   \n",
       "\n",
       "   xa_rolling_opponent  kp_rolling_opponent  1/3_rolling_opponent  \\\n",
       "0                 0.60                  6.0                  29.6   \n",
       "1                 0.66                 11.2                  37.4   \n",
       "2                 1.04                 11.2                  21.2   \n",
       "3                 1.22                 11.2                  36.4   \n",
       "4                 0.62                 10.2                  37.0   \n",
       "\n",
       "   ppa_rolling_opponent  prgp_rolling_opponent  \\\n",
       "0                   9.2                   41.6   \n",
       "1                   8.8                   46.2   \n",
       "2                   8.2                   30.0   \n",
       "3                  11.6                   55.2   \n",
       "4                   5.8                   40.8   \n",
       "\n",
       "   attempted passes_rolling_opponent  completed passes_rolling_opponent  \\\n",
       "0                              561.6                              448.6   \n",
       "1                              597.2                              494.2   \n",
       "2                              374.6                              264.4   \n",
       "3                              587.0                              475.0   \n",
       "4                              511.2                              414.6   \n",
       "\n",
       "   tkl+int_rolling_opponent  mid 3rd_rolling_opponent  \\\n",
       "0                      30.0                     306.2   \n",
       "1                      26.0                     333.0   \n",
       "2                      29.2                     192.0   \n",
       "3                      30.6                     318.2   \n",
       "4                      26.8                     306.2   \n",
       "\n",
       "   att pen_rolling_opponent  att 3rd_rolling_opponent  \n",
       "0                      21.8                     144.4  \n",
       "1                      24.4                     199.2  \n",
       "2                      28.8                     131.6  \n",
       "3                      28.6                     180.6  \n",
       "4                      20.8                     148.6  "
      ]
     },
     "execution_count": 2070,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Mapping to match naming converntion in opponent and team columns\n",
    "opponent_to_team = {\n",
    "    \"Brighton\": \"Brighton and Hove Albion\",\n",
    "    \"Manchester Utd\": \"Manchester United\",\n",
    "    \"Newcastle Utd\": \"Newcastle United\",\n",
    "    \"Tottenham\": \"Tottenham Hotspur\",\n",
    "    \"West Ham\": \"West Ham United\",\n",
    "    \"Wolves\": \"Wolverhampton Wanderers\",\n",
    "    \"Nott'ham Forest\": 'Nottingham Forest',\n",
    "    'Sheffield Utd': 'Sheffield United',\n",
    "    'West Brom': 'West Bromwich Albion'\n",
    "}\n",
    "\n",
    "# Adding 'date' and 'team' to the list for merging purposes\n",
    "columns_for_subset = selected_features + ['date', 'team']\n",
    "\n",
    "# Creating a subset of the dataset with only the necessary columns\n",
    "rolling_data_subset = matches_rolling[columns_for_subset]\n",
    "\n",
    "# Normalizing team names in the original dataset's 'opponent' column\n",
    "matches_rolling['opponent_normalized'] = matches_rolling['opponent'].replace(opponent_to_team)\n",
    "\n",
    "# Merging the original dataset with the rolling data subset\n",
    "# Join on the date and the normalized opponent name\n",
    "merged_data = pd.merge(matches_rolling, rolling_data_subset, left_on=['date', 'opponent_normalized'], right_on=['date', 'team'], suffixes=('', '_opponent'))\n",
    "\n",
    "# Drop the extra 'team_opponent' column after merging\n",
    "merged_data.drop(columns=['team_opponent'], inplace=True)\n",
    "\n",
    "merged_data.head()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.5 Split Data into training and test\n",
    "\n",
    "Before training our predictive model, we need to divide our dataset into three distinct sets: training, validation, and test sets. This split is crucial for several reasons:\n",
    "\n",
    "**Training Set**: This is the largest portion of the dataset and is used to train the model. The model learns to recognise patterns and make predictions based on this data.\n",
    "\n",
    "**Validation Set**: After the model is trained, the validation set is used to evaluate its performance. This helps in fine-tuning the model's hyperparameters without testing on the final test set. It acts as a checkpoint to prevent overfitting.\n",
    "\n",
    "**Test Set**: This is the final set used to evaluate the model. It should only be used once to assess the performance of the fully trained model. The test set provides an unbiased evaluation of the final model fit on the training dataset.\n",
    "\n",
    "We will split our dataset based on dates, ensuring that our model is tested on data it hasn't seen during training or validation, simulating real-world predictions. Here's how we split the dataset:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2071,
   "metadata": {},
   "outputs": [],
   "source": [
    "train = merged_data[merged_data[\"date\"] < '2023-03-01']\n",
    "validation = merged_data[(merged_data[\"date\"] >= '2023-03-01') & (merged_data[\"date\"] < '2023-06-01')] \n",
    "test = merged_data[merged_data[\"date\"] >= '2023-06-01']\n",
    "\n",
    "# Define the features and target variable\n",
    "features = rolling_columns + ['venue_code', 'opp_code', 'team_code']\n",
    "\n",
    "train_features = train[features]\n",
    "train_target = train[\"target\"]\n",
    "\n",
    "validation_features = validation[features]\n",
    "validation_target = validation[\"target\"]\n",
    "\n",
    "test_features = test[features]\n",
    "test_target = test[\"target\"] "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now that we have used the power of pandas' extensive libraries to handle a wide range of datatypes, easier methods of preprocessing, we will convert our data into a numpy array. This will allow for faster mathematical computations when we start using our neural network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2072,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Converting to NumPy arrays\n",
    "train_features_np = train_features.values.astype(np.float32)\n",
    "train_target_np = train_target.values.astype(np.float32)\n",
    "validation_features_np = validation_features.values.astype(np.float32)\n",
    "validation_target_np = validation_target.values.astype(np.float32)\n",
    "test_features_np = test_features.values.astype(np.float32)\n",
    "test_target_np = test_target.values.astype(np.float32)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.6 Normalise Data\n",
    "\n",
    "Before feeding our data into a machine learning model, it's important to normalise it. Normalisation is a preprocessing step used to scale numeric features in the dataset into a common range, typically between 0 and 1. It ensures that each feature contributes proportionately to the final prediction. This process helps in speeding up training and can lead to better model performance since many machine learning algorithms are sensitive to the scale of input data.\n",
    "\n",
    "Additionally, we will fill any NaN values with zero. This is a simple imputation strategy to handle missing values in our data. With knowledge of our dataset from the source tutorial [1], we know that nan values are as a result of the event not occuring hence 0 is appropriate substittution. However, keep in mind that different strategies might be more suitable depending on the nature of your data and the expected impact of these missing values.\n",
    "\n",
    "After normalization, it's important to convert the data into tensors, which are the primary data structures used by neural networks in PyTorch. Here's how we can normalize our data and then convert it into tensors:\n",
    "\n",
    "Here's the code implementation for normalising our data:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2073,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import MinMaxScaler\n",
    "\n",
    "# Creating the scaler instance\n",
    "min_max_scaler = MinMaxScaler()\n",
    "\n",
    "# fill nan values with 0\n",
    "train_features_np = np.nan_to_num(train_features_np)\n",
    "test_features_np = np.nan_to_num(test_features_np)\n",
    "validation_features_np = np.nan_to_num(validation_features_np)\n",
    "\n",
    "\n",
    "# Fitting the scaler to the training data and transforming both training and test data\n",
    "train_features_min_max = min_max_scaler.fit_transform(train_features_np)\n",
    "validation_features_min_max = min_max_scaler.transform(validation_features_np)\n",
    "test_features_min_max = min_max_scaler.transform(test_features_np)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2074,
   "metadata": {},
   "outputs": [],
   "source": [
    "# convert the data to tensors to feed into the model\n",
    "training_input = torch.FloatTensor(train_features_min_max)\n",
    "training_output = torch.FloatTensor(train_target_np)\n",
    "validation_input = torch.FloatTensor(validation_features_min_max)\n",
    "validation_output = torch.FloatTensor(validation_target_np)\n",
    "test_input = torch.FloatTensor(test_features_min_max)\n",
    "test_output = torch.FloatTensor(test_target_np)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='6-building-neural-network-with-pytorch'></a>\n",
    "### 6. Building Neural Network with Pytorch"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we shall define our neural network for our predictive model. Before we start building our model, we set a seed value for reproducibility. This helps to ensure that our results are consistent each time we run our code. It's crucial for the experimental process, allowing us to compare results under the same initial conditions.\n",
    "\n",
    "Here's the code for setting the seed and defining our neural network architecture:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2075,
   "metadata": {},
   "outputs": [],
   "source": [
    "# for reproducibility\n",
    "seed_value = 42\n",
    "torch.manual_seed(seed_value)\n",
    "np.random.seed(seed_value)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2076,
   "metadata": {},
   "outputs": [],
   "source": [
    "# source [2] \n",
    "# https://medium.com/@andreluiz_4916/pytorch-neural-networks-to-predict-matches-results-in-soccer-championships-part-ii-3d02b2ddd538\n",
    "\n",
    "import torch\n",
    "class Net(torch.nn.Module):\n",
    "    def __init__(self, input_size, hidden_size):\n",
    "        super(Net, self).__init__()\n",
    "        self.input_size = input_size\n",
    "        self.hidden_size = hidden_size\n",
    "        self.fc1 = torch.nn.Linear(self.input_size, self.hidden_size)\n",
    "        self.relu = torch.nn.ReLU()\n",
    "        self.fc2 = torch.nn.Linear(self.hidden_size, 1)\n",
    "        self.sigmoid = torch.nn.Sigmoid() \n",
    "    def forward(self, x):\n",
    "        hidden = self.fc1(x)\n",
    "        relu = self.relu(hidden)\n",
    "        output = self.fc2(relu)\n",
    "        output = self.sigmoid(output)\n",
    "        return output\n",
    "    \n",
    "#  end of source [2]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='7-train-model'></a>\n",
    "### 7. Train Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Training a neural network is a critical step where the model learns to make predictions by adjusting its weights based on the input data. In this section, we'll walk through the process of training our neural network, which includes setting the parameters, defining the loss function, choosing an optimizer, and iterating through the epochs.\n",
    "\n",
    "Here are the key steps in our training process:\n",
    "\n",
    "Defining the Model Architecture: We instantiate our neural network model by specifying the input size (number of features) and the hidden layer size. The hidden layer size is a hyperparameter that we can tune.\n",
    "\n",
    "Loss Function and Optimizer: For our binary classification task, we use the Binary Cross-Entropy Loss (BCELoss), which is well-suited for binary outcomes. The optimizer updates the model's weights; here, we use Stochastic Gradient Descent (SGD) with a learning rate and momentum parameter.\n",
    "\n",
    "Evaluating the Model Before Training: It's useful to assess the model's performance before training to establish a baseline. We use precision as one of our evaluation metrics, which measures the accuracy of the positive predictions,  alongside the standard accuracy metric and a confusion matrix.\n",
    "\n",
    "Training Loop: The training loop involves repeatedly passing the training data through the model (forward pass), calculating the loss, and updating the model's weights (backward pass). We'll track the loss at each epoch to monitor the training process.\n",
    "\n",
    "Evaluating the Model After Training: After training, we evaluate the model again to see the improvements made through the training process.\n",
    "\n",
    "Here's the implementation of the training process:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2077,
   "metadata": {},
   "outputs": [],
   "source": [
    "# source [2] \n",
    "# https://medium.com/@andreluiz_4916/pytorch-neural-networks-to-predict-matches-results-in-soccer-championships-part-ii-3d02b2ddd538\n",
    "\n",
    "input_size = training_input.size()[1] # number of features selected\n",
    "hidden_size = 30 # number of nodes/neurons in the hidden layer\n",
    "model = Net(input_size, hidden_size) # create the model\n",
    "criterion = torch.nn.BCELoss() # works for binary classification\n",
    "optimizer = torch.optim.SGD(model.parameters(), lr = 0.9, momentum=0.2)\n",
    "\n",
    "# end of source [2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2078,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test loss before training 0.6917647123336792\n"
     ]
    }
   ],
   "source": [
    "# source [2] \n",
    "# https://medium.com/@andreluiz_4916/pytorch-neural-networks-to-predict-matches-results-in-soccer-championships-part-ii-3d02b2ddd538\n",
    "\n",
    "# evaluate the model before training\n",
    "model.eval()\n",
    "y_pred = model(test_input)\n",
    "before_train = criterion(y_pred.squeeze(), test_output)\n",
    "print('Test loss before training' , before_train.item())\n",
    "\n",
    "# end of source [2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2079,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 500x500 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "import seaborn as sns\n",
    "\n",
    "# Convert predictions to binary format\n",
    "y_pred_binary = (y_pred.squeeze() > 0.5).float()\n",
    "\n",
    "# Create the confusion matrix\n",
    "cm = confusion_matrix(test_output.numpy(), y_pred_binary.numpy())\n",
    "\n",
    "# Plot the confusion matrix\n",
    "plt.figure(figsize=(5, 5))\n",
    "sns.heatmap(cm, annot=True, fmt='g')\n",
    "plt.title('Confusion Matrix')\n",
    "plt.xlabel('Predicted Label')\n",
    "plt.ylabel('True Label')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2080,
   "metadata": {},
   "outputs": [],
   "source": [
    "def evaluate_model(model, val_inputs, val_outputs):\n",
    "    model.eval()\n",
    "    with torch.no_grad():\n",
    "        predictions = model(val_inputs)\n",
    "        accuracy = ((predictions.squeeze() > 0.5) == val_outputs).float().mean().item()\n",
    "    return accuracy\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2081,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 62.63%\n"
     ]
    }
   ],
   "source": [
    "accuracy = evaluate_model(model, validation_input, validation_output)\n",
    "print(f\"Accuracy: {accuracy * 100:.2f}%\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2082,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision before training: 18.92%\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import precision_score\n",
    "\n",
    "# Make predictions\n",
    "model.eval()\n",
    "y_pred = model(test_input)\n",
    "y_pred = (y_pred.squeeze() > 0.5).float()  # Convert probabilities to binary predictions\n",
    "\n",
    "# Calculate precision\n",
    "precision = precision_score(test_output, y_pred)\n",
    "print(f\"Precision before training: {precision * 100:.2f}%\" )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2083,
   "metadata": {},
   "outputs": [],
   "source": [
    "def trainFunction(model, optimizer, criterion, training_input, training_output, epochs):\n",
    "    model.train()\n",
    "    errors = []\n",
    "    for epoch in range(epochs):\n",
    "        optimizer.zero_grad()\n",
    "        # Forward pass\n",
    "        y_pred = model(training_input)\n",
    "        # Compute Loss\n",
    "        loss = criterion(y_pred.squeeze(), training_output)\n",
    "        errors.append(loss.item())\n",
    "        # print('Epoch {}: train loss: {}'.format(epoch, loss.item()))\n",
    "        # Backward pass\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "    return errors\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2084,
   "metadata": {},
   "outputs": [],
   "source": [
    "# source [2] \n",
    "# https://medium.com/@andreluiz_4916/pytorch-neural-networks-to-predict-matches-results-in-soccer-championships-part-ii-3d02b2ddd538\n",
    "\n",
    "# torch.set_grad_enabled(True)\n",
    "input_size = training_input.size()[1] # number of features selected\n",
    "hidden_size = 30 # number of nodes/neurons in the hidden layer\n",
    "model = Net(input_size, hidden_size) # create the model\n",
    "criterion = torch.nn.BCELoss() # works for binary classification\n",
    "optimizer = torch.optim.SGD(model.parameters(), lr = 0.9, momentum=0.2)\n",
    "\n",
    "#  end of source [2]\n",
    "\n",
    "model.train()\n",
    "# Call the train function\n",
    "epochs = 400\n",
    "errors = trainFunction(model, optimizer, criterion, training_input, training_output, epochs)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2085,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test loss after Training 0.645779550075531\n"
     ]
    }
   ],
   "source": [
    "# evaluate the model after training\n",
    "model.eval()\n",
    "y_pred = model(test_input)\n",
    "after_train = criterion(y_pred.squeeze(), test_output)\n",
    "print('Test loss after Training' , after_train.item())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='8-plot-the-loss-of-your-model-over-training'></a> \n",
    "### 8. Plot the loss of your model over training"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Visualizing the training progress of a neural network is important for understanding how well the model is learning. One of the most informative visualizations in this context is the plot of the model's loss over each training epoch. A loss plot can provide insights into how quickly or effectively the model is converging towards minimizing the error.\n",
    "\n",
    "The second chart represents the real values (Win at x = 1, Draw-Defeat at x = 0) with yellow circles; and the predicted values with blue crosses. The closer the blue cross is to the yellow circle, the more this prediction is accurated"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2086,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x500 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# source [2] \n",
    "# https://medium.com/@andreluiz_4916/pytorch-neural-networks-to-predict-matches-results-in-soccer-championships-part-ii-3d02b2ddd538\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "def plotcharts(errors):\n",
    "    errors = np.array(errors)\n",
    "    plt.figure(figsize=(12, 5))\n",
    "    graf02 = plt.subplot(1, 2, 1) # nrows, ncols, index\n",
    "    graf02.set_title('Errors')\n",
    "    plt.plot(errors, '-')\n",
    "    plt.xlabel('Epochs')\n",
    "    graf03 = plt.subplot(1, 2, 2)\n",
    "    graf03.set_title('Tests')\n",
    "    a = plt.plot(test_output.numpy(), 'yo', label='Real')\n",
    "    plt.setp(a, markersize=10)\n",
    "    a = plt.plot(y_pred.detach().numpy(), 'b+', label='Predicted')\n",
    "    plt.setp(a, markersize=10)\n",
    "    plt.legend(loc=7)\n",
    "    plt.show()\n",
    "plotcharts(errors)\n",
    "# end of source [2]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Through these plots, we get a visual feedback loop that aids in tuning our model for better performance. The error plot helps us in understanding the learning curve, and the test plot visually demonstrates how close the model's predictions are to the actual outcomes."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='9-evaluate-performance-of-the-model'></a> \n",
    "### 9. Evaluate Performance of the Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "After training our neural network, we evaluate the models performance to understand how well it has learned to predict match outcomes. In this section, we will use different metrics to gauge the model's effectiveness on our test dataset. These metrics provide us with insights into various aspects of the model's predictive capabilities.\n",
    "\n",
    "Accuracy: This is the most intuitive performance measure and it is simply a ratio of correctly predicted observation to the total observations. Accuracy gives us a baseline for the effectiveness of our model.\n",
    "\n",
    "Precision: Precision is the ratio of correctly predicted positive observations to the total predicted positive observations. Good precision indicates our model has an understanding of what factors lead to a team winning a match.\n",
    "\n",
    "Confusion Matrix: Additionally, we visualize the model's performance using a confusion matrix, which not only highlights the successes (true positives and true negatives) but also sheds light on the errors made (false positives and false negatives).\n",
    "\n",
    "Here's how we can compute these metrics using PyTorch and sklearn:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2087,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 64.45%\n"
     ]
    }
   ],
   "source": [
    "# source [2] \n",
    "# https://medium.com/@andreluiz_4916/pytorch-neural-networks-to-predict-matches-results-in-soccer-championships-part-ii-3d02b2ddd538\n",
    "model.eval()  # Set the model to evaluation mode\n",
    "with torch.no_grad():  # Disable gradient computation\n",
    "    # Get predictions\n",
    "    outputs = model(test_input)\n",
    "    predicted = (outputs.squeeze() > 0.5).long()  # Convert probabilities to class labels\n",
    "\n",
    "# Calculate accuracy\n",
    "accuracy = (predicted == test_output).float().mean()  # Mean of correct predictions\n",
    "print(f\"Accuracy: {accuracy * 100:.2f}%\")\n",
    "# end of source [2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2088,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision after training: 54.88%\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import precision_score\n",
    "\n",
    "# Make predictions\n",
    "model.eval()\n",
    "y_pred = model(test_input)\n",
    "y_pred = (y_pred.squeeze() > 0.5).float()  # Convert probabilities to binary predictions\n",
    "\n",
    "# Calculate precision\n",
    "precision = precision_score(test_output, y_pred)\n",
    "print(f\"Precision after training: {precision * 100:.2f}%\" )\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2089,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 500x500 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "import seaborn as sns\n",
    "\n",
    "# Convert predictions to binary format\n",
    "y_pred_binary = (y_pred.squeeze() > 0.5).float()\n",
    "\n",
    "# Create the confusion matrix\n",
    "cm = confusion_matrix(test_output.numpy(), y_pred_binary.numpy())\n",
    "\n",
    "# Plot the confusion matrix\n",
    "plt.figure(figsize=(5, 5))\n",
    "sns.heatmap(cm, annot=True, fmt='g')\n",
    "plt.title('Confusion Matrix')\n",
    "plt.xlabel('Predicted Label')\n",
    "plt.ylabel('True Label')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Our evaluation of the model's performance post-training reveals a notable improvement over its initial, untrained state. Significantly, the precision has surged from 18% to 54.8%, underscoring the model's enhanced proficiency in correctly predicting wins. While the accuracy has seen a moderate ascent from 62.63% to 64.45%, the improvement is best understood in the context of the model's initial tendency to predict non-winning outcomes, a likely consequence of the dataset's imbalance where approximately 60% of the instances belong to the draw-lose class. Prior to training, this bias towards the majority class resulted in a deceptively high accuracy, masking the model's lack of discriminative power. Prior it predicted 91% of matches to belong to class 0.\n",
    "\n",
    "Post-training, the model exhibits a more varied prediction pattern, indicating a shift away from a simplistic strategy of favoring the more common class. This change reflects a deeper comprehension of the factors influencing match victories and a move towards more nuanced predictions, as displayed by the model only predicting class 0 62% of the time now. Crucially, the post-training confusion matrix illustrates the model's newfound ability to discern true positives and negatives more effectively, affirming a balanced approach to prediction. The model now not only achieves higher accuracy but does so with a richer understanding of match dynamics, representing a significant stride towards capturing the complexity of real-world match outcomes."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='10-hyperparameter-tuning'></a> \n",
    "### 10. Hyperparameter Tuning"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Hyperparameter tuning is a vital process in machine learning, where we systematically vary the parameters of our model to find the optimal configuration for the best performance. Unlike model parameters, which are learned during training, hyperparameters are set prior to the training process and can have a significant impact on the learning and prediction ability of the model.\n",
    "\n",
    "In this section, we will conduct hyperparameter tuning using a grid search approach. Grid search tests every combination of a predefined list of values of hyperparameters. It's a comprehensive, though computationally expensive, method to find the best combination for our model.\n",
    "\n",
    "We will record the accuracy of each combination then plot a graph at the end to see how changing values of each hyperparameter while keeping the others controlled affected performance. This approach provides a deeper insight into the sensitivity and influence of each hyperparameter on our model, allowing us to fine-tune our network more effectively.\n",
    "\n",
    "We'll focus on tuning several hyperparameters:\n",
    "\n",
    "- **Learning Rate (lr)**: Determines how much to adjust the model in response to the estimated error each time the model weights are updated.\n",
    "- **Hidden Layer Size (hidden_size)**: The number of nodes in the hidden layer.\n",
    "- **Number of Epochs (epochs)**: The number of complete passes through the training dataset.\n",
    "- **Optimizer Type (optimizer)**: Algorithm to update the weights of the network.\n",
    "- **Momentum (momentum)**: Helps to accelerate gradients vectors in the right direction, leading to faster converging.\n",
    "Here's how we implement the grid search:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2090,
   "metadata": {},
   "outputs": [],
   "source": [
    "# initialise search space\n",
    "\n",
    "hyperparameters = {\n",
    "    'lr': [0.001, 0.01, 0.1],  # Learning rate\n",
    "    'hidden_size': [10, 20, 30], # Number of nodes in the hidden layer\n",
    "    'epochs': [100, 200 ,400, 600, 1000 ,2000], # Number of epochs\n",
    "    'optimizer' : ['Adam', 'SGD'], # Optimizer\n",
    "    'momentum': [0.0, 0.5, 0.9] # Momentum\n",
    "    # Add other hyperparameters here\n",
    "}\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2091,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import itertools\n",
    "# from collections import defaultdict\n",
    "\n",
    "# def grid_search(hyperparameters, training_data, training_output, input_size, validation_input, validation_output):\n",
    "#     results = []\n",
    "#     best_performance = float('-inf')\n",
    "#     best_params = None\n",
    "#     performance_dict = defaultdict(list)\n",
    "\n",
    "#     for lr, hidden_size, epochs, optimizer_type, momentum in itertools.product(hyperparameters['lr'], hyperparameters['hidden_size'], hyperparameters['epochs'], hyperparameters['optimizer'], hyperparameters['momentum']):\n",
    "#         # Initialize model\n",
    "#         model = Net(input_size, hidden_size)\n",
    "\n",
    "#         # Initialize optimizer\n",
    "#         if optimizer_type == 'Adam':\n",
    "#             optimizer = torch.optim.Adam(model.parameters(), lr=lr)\n",
    "#         elif optimizer_type == 'SGD':\n",
    "#             optimizer = torch.optim.SGD(model.parameters(), lr=lr, momentum=momentum)\n",
    "\n",
    "#         # Train the model\n",
    "#         trainFunction(model, optimizer, criterion, training_data, training_output, epochs)\n",
    "\n",
    "#         # Evaluate the model\n",
    "#         performance = evaluate_model(model, validation_input, validation_output)\n",
    "\n",
    "#         # Store performance in the dictionary\n",
    "#         performance_dict['lr'].append((lr, performance))\n",
    "#         performance_dict['hidden_size'].append((hidden_size, performance))\n",
    "#         performance_dict['epochs'].append((epochs, performance))\n",
    "#         performance_dict['optimizer'].append((optimizer_type, performance))\n",
    "#         performance_dict['momentum'].append((momentum, performance))\n",
    "\n",
    "#         # Update best performance\n",
    "#         if performance > best_performance:\n",
    "#             best_performance = performance\n",
    "#             best_params = {\n",
    "#                 'lr': lr, 'hidden_size': hidden_size, 'epochs': epochs,\n",
    "#                 'optimizer': optimizer_type, 'momentum': momentum\n",
    "#             }\n",
    "\n",
    "#     return best_params, best_performance, performance_dict\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2092,
   "metadata": {},
   "outputs": [],
   "source": [
    "import itertools\n",
    "from collections import defaultdict\n",
    "\n",
    "def grid_search(hyperparameters, training_data, training_output, input_size, validation_input, validation_output):\n",
    "    # results = []  # This is not used in the original code, so we can remove it\n",
    "    best_performance = float('-inf')\n",
    "    best_params = None\n",
    "    performance_dict = defaultdict(list)\n",
    "    performance_list = []  # To capture the performance for each combination\n",
    "\n",
    "    for lr, hidden_size, epochs, optimizer_type, momentum in itertools.product(\n",
    "        hyperparameters['lr'],\n",
    "        hyperparameters['hidden_size'],\n",
    "        hyperparameters['epochs'],\n",
    "        hyperparameters['optimizer'],\n",
    "        hyperparameters['momentum']\n",
    "    ):\n",
    "        # Initialize model\n",
    "        model = Net(input_size, hidden_size)\n",
    "\n",
    "        # Initialize optimizer\n",
    "        if optimizer_type == 'Adam':\n",
    "            optimizer = torch.optim.Adam(model.parameters(), lr=lr)\n",
    "            current_momentum = None  # Momentum is not used with Adam\n",
    "        elif optimizer_type == 'SGD':\n",
    "            optimizer = torch.optim.SGD(model.parameters(), lr=lr, momentum=momentum)\n",
    "            current_momentum = momentum  # Use the provided momentum value\n",
    "\n",
    "        # Train the model\n",
    "        trainFunction(model, optimizer, criterion, training_data, training_output, epochs)\n",
    "\n",
    "        # Evaluate the model\n",
    "        performance = evaluate_model(model, validation_input, validation_output)\n",
    "\n",
    "        # Store performance in the list with corresponding hyperparameters\n",
    "        performance_list.append({\n",
    "            'lr': lr,\n",
    "            'hidden_size': hidden_size,\n",
    "            'epochs': epochs,\n",
    "            'optimizer': optimizer_type,\n",
    "            'momentum': current_momentum,\n",
    "            'performance': performance\n",
    "        })\n",
    "\n",
    "        # Update best performance\n",
    "        if performance > best_performance:\n",
    "            best_performance = performance\n",
    "            best_params = {\n",
    "                'lr': lr,\n",
    "                'hidden_size': hidden_size,\n",
    "                'epochs': epochs,\n",
    "                'optimizer': optimizer_type,\n",
    "                'momentum': current_momentum\n",
    "            }\n",
    "\n",
    "    return best_params, best_performance, performance_list\n",
    "\n",
    "# Now, let's define the hyperparameters and run the grid search\n",
    "# Assuming you have a dictionary called `hyperparameters` with your ranges defined\n",
    "# Also assuming the functions `Net`, `trainFunction`, and `evaluate_model` are defined, as well as `training_data`, `training_output`, etc.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2093,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best hyperparameters: {'lr': 0.001, 'hidden_size': 10, 'epochs': 600, 'optimizer': 'Adam', 'momentum': None}, Best performance: 0.6605504751205444\n"
     ]
    }
   ],
   "source": [
    "best_params, best_performance, results = grid_search(hyperparameters, training_input, training_output, input_size, test_input, test_output)\n",
    "print(f\"Best hyperparameters: {best_params}, Best performance: {best_performance}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2094,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Print out the performance of each combination\n",
    "# for combo in results:\n",
    "#     print(combo)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2095,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 611.111x900 with 5 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Convert the performance list to a DataFrame for easy plotting\n",
    "performance_df = pd.DataFrame(results)\n",
    "\n",
    "# Now, create separate columns for each hyperparameter in the DataFrame\n",
    "performance_df['lr'] = [d['lr'] for d in results]\n",
    "performance_df['hidden_size'] = [d['hidden_size'] for d in results]\n",
    "performance_df['epochs'] = [d['epochs'] for d in results]\n",
    "performance_df['optimizer'] = [d['optimizer'] for d in results]\n",
    "performance_df['momentum'] = [d['momentum'] for d in results]\n",
    "performance_df['Performance'] = [d['performance'] for d in results]\n",
    "\n",
    "# Plot the performance for different hyperparameters using seaborn\n",
    "# Here we need to melt the DataFrame to have 'Hyperparameter' and 'Value' columns\n",
    "performance_melted_df = performance_df.melt(id_vars=['Performance'], \n",
    "                                            value_vars=['lr', 'hidden_size', 'epochs', 'optimizer', 'momentum'],\n",
    "                                            var_name='Hyperparameter', value_name='Value')\n",
    "\n",
    "g = sns.FacetGrid(performance_melted_df, col=\"Hyperparameter\", col_wrap=2, sharex=False, sharey=False)\n",
    "g.map_dataframe(sns.lineplot, x='Value', y='Performance')\n",
    "g.add_legend()\n",
    "g.set_titles(col_template=\"{col_name}\")\n",
    "\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2096,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Convert the performance dictionary to a DataFrame for easy plotting\n",
    "# performance_df = pd.DataFrame({\n",
    "#     'Hyperparameter': np.repeat(list(results.keys()), [len(v) for v in results.values()]),\n",
    "#     'Value': [item for sublist in results.values() for item, _ in sublist],\n",
    "#     'Performance': [perf for sublist in results.values() for _, perf in sublist]\n",
    "# })\n",
    "\n",
    "# g = sns.FacetGrid(performance_df, col=\"Hyperparameter\", col_wrap=2, sharex=False, sharey=False)\n",
    "# g.map_dataframe(sns.lineplot, x='Value', y='Performance')\n",
    "# g.add_legend()\n",
    "# g.set_titles(col_template=\"{col_name}\")\n",
    "\n",
    "# plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='11-final-model'></a> \n",
    "### 11. Final Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this section, we'll use the selected hyperparameters to configure our neural network and train it with our dataset. This final iteration of the model should ideally represent the most effective version, considering all the adjustments and improvements we've made."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "training set Best hyperparameters: {'lr': 0.1, 'hidden_size': 20, 'epochs': 50, 'optimizer': 'Adam', 'momentum': None}, Best performance: 0.6674311757087708\n",
    "\n",
    "training Best hyperparameters: {'lr': 0.001, 'hidden_size': 10, 'epochs': 600, 'optimizer': 'Adam', 'momentum': None}, Best performance: 0.6605504751205444\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2097,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # train with best hyperparameters\n",
    "\n",
    "# selected_features\n",
    "\n",
    "# input_size = training_input.size()[1] # number of features selected\n",
    "# hidden_size = 30 # number of nodes/neurons in the hidden layer\n",
    "# model = Net(input_size, hidden_size) # create the model\n",
    "# criterion = torch.nn.BCELoss() # works for binary classification\n",
    "# optimizer = torch.optim.SGD(model.parameters(), lr=0.1, momentum=0.2)\n",
    "\n",
    "# # Call the train function\n",
    "# epochs = 400\n",
    "# errors = trainFunction(model, optimizer, criterion, training_input, training_output, epochs)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2111,
   "metadata": {},
   "outputs": [],
   "source": [
    "# for reproducibility\n",
    "seed_value = 42\n",
    "torch.manual_seed(seed_value)\n",
    "np.random.seed(seed_value)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2132,
   "metadata": {},
   "outputs": [],
   "source": [
    "# train with best hyperparameters\n",
    "\n",
    "selected_features\n",
    "\n",
    "input_size = training_input.size()[1] # number of features selected\n",
    "hidden_size = 10 # number of nodes/neurons in the hidden layer\n",
    "model = Net(input_size, hidden_size) # create the model\n",
    "criterion = torch.nn.BCELoss() # works for binary classification\n",
    "optimizer = torch.optim.Adam(model.parameters(), lr=0.001)\n",
    "\n",
    "# Call the train function\n",
    "epochs = 600\n",
    "errors = trainFunction(model, optimizer, criterion, training_input, training_output, epochs)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2133,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 65.60%\n"
     ]
    }
   ],
   "source": [
    "accuracy = evaluate_model(model, test_input, test_output)\n",
    "print(f\"Accuracy: {accuracy * 100:.2f}%\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2134,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision after training: 59.81%\n"
     ]
    }
   ],
   "source": [
    "# Make predictions\n",
    "model.eval()\n",
    "y_pred = model(test_input)\n",
    "y_pred = (y_pred.squeeze() > 0.5).float()  # Convert probabilities to binary predictions\n",
    "\n",
    "# Calculate precision\n",
    "precision = precision_score(test_output, y_pred)\n",
    "print(f\"Precision after training: {precision * 100:.2f}%\" )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2135,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAcAAAAHWCAYAAADgqln1AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/SrBM8AAAACXBIWXMAAA9hAAAPYQGoP6dpAABE3ElEQVR4nO3deVhUZf8/8PcgMCDIIAoMk4K4ZKAmpj6KG5Ik7pL2GOaCK2mgKWpGuadNmuWWSlmpmfaUlXu5hAuZuIe7uKGUOKASICjDMuf3hz/m2wjqjA4zwv1+Pde5ruY+97nPZ+by8eP9Ofc5RyZJkgQiIiLB2Fg7ACIiImtgAiQiIiExARIRkZCYAImISEhMgEREJCQmQCIiEhITIBERCYkJkIiIhMQESEREQmICpArl4sWL6Ny5MxQKBWQyGTZu3GjW8a9evQqZTIZVq1aZddyKrGPHjujYsaO1wyAyOyZAMtnly5fx5ptvom7dunBwcICLiwvatm2LRYsW4d69e+V67oiICJw6dQpz5szBmjVr0KJFi3I9nyUNGTIEMpkMLi4uZf6OFy9ehEwmg0wmw/z5800ePy0tDTNmzEBSUpIZoiWq+GytHQBVLNu2bcN///tfyOVyDB48GI0bN0ZBQQH279+PSZMm4cyZM/jiiy/K5dz37t1DYmIi3n//fURHR5fLOXx8fHDv3j3Y2dmVy/iPY2tri7t372LLli3o16+fwb61a9fCwcEB+fn5TzR2WloaZs6ciTp16iAgIMDo43bu3PlE5yN61jEBktFSUlIQHh4OHx8f7N69G15eXvp9UVFRuHTpErZt21Zu57958yYAwNXVtdzOIZPJ4ODgUG7jP45cLkfbtm3x3XfflUqA69atQ/fu3fHTTz9ZJJa7d++iatWqsLe3t8j5iCyNJVAy2rx585Cbm4uvvvrKIPmVqF+/Pt5++23956KiInzwwQeoV68e5HI56tSpg/feew9ardbguDp16qBHjx7Yv38//vOf/8DBwQF169bFN998o+8zY8YM+Pj4AAAmTZoEmUyGOnXqALhfOiz573+bMWMGZDKZQduuXbvQrl07uLq6wtnZGQ0bNsR7772n3/+wa4C7d+9G+/bt4eTkBFdXV/Tu3Rvnzp0r83yXLl3CkCFD4OrqCoVCgaFDh+Lu3bsP/2Ef8MYbb+DXX39FVlaWvu3IkSO4ePEi3njjjVL9MzMzMXHiRDRp0gTOzs5wcXFB165dceLECX2fvXv3omXLlgCAoUOH6kupJd+zY8eOaNy4MY4dO4YOHTqgatWq+t/lwWuAERERcHBwKPX9Q0NDUb16daSlpRn9XYmsiQmQjLZlyxbUrVsXbdq0Mar/iBEjMG3aNLz00ktYsGABgoKCoFarER4eXqrvpUuX8Nprr+GVV17BJ598gurVq2PIkCE4c+YMAKBPnz5YsGABAKB///5Ys2YNFi5caFL8Z86cQY8ePaDVajFr1ix88skn6NWrF/74449HHvfbb78hNDQUGRkZmDFjBmJiYnDgwAG0bdsWV69eLdW/X79+uHPnDtRqNfr164dVq1Zh5syZRsfZp08fyGQy/Pzzz/q2devW4YUXXsBLL71Uqv+VK1ewceNG9OjRA59++ikmTZqEU6dOISgoSJ+M/Pz8MGvWLABAZGQk1qxZgzVr1qBDhw76cW7fvo2uXbsiICAACxcuRHBwcJnxLVq0CO7u7oiIiEBxcTEA4PPPP8fOnTuxZMkSqFQqo78rkVVJREbIzs6WAEi9e/c2qn9SUpIEQBoxYoRB+8SJEyUA0u7du/VtPj4+EgApISFB35aRkSHJ5XJpwoQJ+raUlBQJgPTxxx8bjBkRESH5+PiUimH69OnSv/+IL1iwQAIg3bx586Fxl5xj5cqV+raAgADJw8NDun37tr7txIkTko2NjTR48OBS5xs2bJjBmK+++qpUo0aNh57z39/DyclJkiRJeu2116ROnTpJkiRJxcXFklKplGbOnFnmb5Cfny8VFxeX+h5yuVyaNWuWvu3IkSOlvluJoKAgCYAUFxdX5r6goCCDth07dkgApNmzZ0tXrlyRnJ2dpbCwsMd+R6JnCWeAZJScnBwAQLVq1Yzq/8svvwAAYmJiDNonTJgAAKWuFfr7+6N9+/b6z+7u7mjYsCGuXLnyxDE/qOTa4aZNm6DT6Yw65saNG0hKSsKQIUPg5uamb3/xxRfxyiuv6L/nv40aNcrgc/v27XH79m39b2iMN954A3v37oVGo8Hu3buh0WjKLH8C968b2tjc/79ycXExbt++rS/vHj9+3OhzyuVyDB061Ki+nTt3xptvvolZs2ahT58+cHBwwOeff270uYieBUyAZBQXFxcAwJ07d4zqf+3aNdjY2KB+/foG7UqlEq6urrh27ZpBu7e3d6kxqlevjn/++ecJIy7t9ddfR9u2bTFixAh4enoiPDwcP/zwwyOTYUmcDRs2LLXPz88Pt27dQl5enkH7g9+levXqAGDSd+nWrRuqVauG77//HmvXrkXLli1L/ZYldDodFixYgAYNGkAul6NmzZpwd3fHyZMnkZ2dbfQ5n3vuOZMWvMyfPx9ubm5ISkrC4sWL4eHhYfSxRM8CJkAyiouLC1QqFU6fPm3ScQ8uQnmYKlWqlNkuSdITn6Pk+lQJR0dHJCQk4LfffsOgQYNw8uRJvP7663jllVdK9X0aT/NdSsjlcvTp0werV6/Ghg0bHjr7A4APP/wQMTEx6NChA7799lvs2LEDu3btQqNGjYye6QL3fx9T/Pnnn8jIyAAAnDp1yqRjiZ4FTIBktB49euDy5ctITEx8bF8fHx/odDpcvHjRoD09PR1ZWVn6FZ3mUL16dYMVkyUenGUCgI2NDTp16oRPP/0UZ8+exZw5c7B7927s2bOnzLFL4kxOTi617/z586hZsyacnJye7gs8xBtvvIE///wTd+7cKXPhUIkff/wRwcHB+OqrrxAeHo7OnTsjJCSk1G9i7D9GjJGXl4ehQ4fC398fkZGRmDdvHo4cOWK28YksgQmQjPbOO+/AyckJI0aMQHp6eqn9ly9fxqJFiwDcL+EBKLVS89NPPwUAdO/e3Wxx1atXD9nZ2Th58qS+7caNG9iwYYNBv8zMzFLHltwQ/uCtGSW8vLwQEBCA1atXGySU06dPY+fOnfrvWR6Cg4PxwQcf4LPPPoNSqXxovypVqpSaXa5fvx7Xr183aCtJ1GX9Y8FUkydPRmpqKlavXo1PP/0UderUQURExEN/R6JnEW+EJ6PVq1cP69atw+uvvw4/Pz+DJ8EcOHAA69evx5AhQwAATZs2RUREBL744gtkZWUhKCgIhw8fxurVqxEWFvbQJfZPIjw8HJMnT8arr76KsWPH4u7du1i+fDmef/55g0Ugs2bNQkJCArp37w4fHx9kZGRg2bJlqFWrFtq1a/fQ8T/++GN07doVgYGBGD58OO7du4clS5ZAoVBgxowZZvseD7KxscGUKVMe269Hjx6YNWsWhg4dijZt2uDUqVNYu3Yt6tata9CvXr16cHV1RVxcHKpVqwYnJye0atUKvr6+JsW1e/duLFu2DNOnT9fflrFy5Up07NgRU6dOxbx580waj8hqrLwKlSqgCxcuSCNHjpTq1Kkj2dvbS9WqVZPatm0rLVmyRMrPz9f3KywslGbOnCn5+vpKdnZ2Uu3ataXY2FiDPpJ0/zaI7t27lzrPg8vvH3YbhCRJ0s6dO6XGjRtL9vb2UsOGDaVvv/221G0Q8fHxUu/evSWVSiXZ29tLKpVK6t+/v3ThwoVS53jwVoHffvtNatu2reTo6Ci5uLhIPXv2lM6ePWvQp+R8D95msXLlSgmAlJKS8tDfVJIMb4N4mIfdBjFhwgTJy8tLcnR0lNq2bSslJiaWefvCpk2bJH9/f8nW1tbgewYFBUmNGjUq85z/HicnJ0fy8fGRXnrpJamwsNCg3/jx4yUbGxspMTHxkd+B6FkhkyQTrswTERFVErwGSEREQmICJCIiITEBEhGRkJgAiYhISEyAREQkJCZAIiISEhMgEREJqVI+CabwlvleoUP0KI6q9o/vRGQGRQXXH9/JBOb8e9KuZt3Hd/r/1Go1fv75Z5w/fx6Ojo5o06YN5s6dq3/jSmZmJqZPn46dO3ciNTUV7u7uCAsLwwcffACFQqEfJzU1FaNHj8aePXvg7OyMiIgIqNVq2Noan9YqZQIkIqLH0JnvDSim2LdvH6KiotCyZUsUFRXhvffeQ+fOnXH27Fk4OTkhLS0NaWlpmD9/Pvz9/XHt2jWMGjUKaWlp+PHHHwHcf9NL9+7doVQqceDAAdy4cQODBw+GnZ0dPvzwQ6NjqZRPguEMkCyFM0CyFLPPADMuPr6Tkew8GjzxsTdv3oSHhwf27duHDh06lNln/fr1GDhwIPLy8mBra4tff/0VPXr0QFpaGjw9PQEAcXFxmDx5Mm7evGn0ey15DZCISESSzmybVqtFTk6OwWbsm0FKXtrs5ub2yD4uLi768mZiYiKaNGmiT34AEBoaipycHJw5c8bon4AJkIhIRDqd2Ta1Wg2FQmGwqdVqI0LQYdy4cWjbti0aN25cZp9bt27hgw8+QGRkpL5No9EYJD8A+s8ajcbon4DXAImI6KnExsYiJibGoE0ulz/2uKioKJw+fRr79+8vc39OTg66d+8Of3//cnn1GBMgEZGAJElntrHkcrlRCe/foqOjsXXrViQkJKBWrVql9t+5cwddunRBtWrVsGHDBtjZ2en3KZVKHD582KB/yUu6H/Xy6AexBEpEJCIzlkBNIUkSoqOjsWHDBuzevbvMFzLn5OSgc+fOsLe3x+bNm+Hg4GCwPzAwEKdOnUJGRoa+bdeuXXBxcYG/v7/RsXAGSEREFhMVFYV169Zh06ZNqFatmv6anUKhgKOjoz753b17F99++61+UQ0AuLu7o0qVKujcuTP8/f0xaNAgzJs3DxqNBlOmTEFUVJRJM1HeBkH0FHgbBFmKuW+DKPjrhNnGsq/d1Oi+MpmszPaVK1diyJAh2Lt3L4KDg8vsk5KSgjp16gAArl27htGjR2Pv3r1wcnJCREQEPvroI5NuhGcCJHoKTIBkKWZPgNeOm20se5+XzDaWJfEaIBERCYnXAImIRGTGVaAVFRMgEZGITFy9WRmxBEpERELiDJCISEDmvBG+omICJCISEUugLIESEZGYOAMkIhIRS6BMgEREQrLSG+GfJSyBEhGRkDgDJCISEUugTIBERELiKlCWQImISEycARIRiYglUCZAIiIhsQTKEigREYmJM0AiIgFJEu8DZAIkIhIRrwGyBEpERGLiDJCISERcBMMESEQkJJZAWQIlIiIxcQZIRCQivg2CCZCISEgsgbIESkREYuIMkIhIRFwFygRIRCQklkBZAiUiIjFxBkhEJCKWQJkAiYiExATIEigREYmJM0AiIgHxdUicARIRiUmnM99mArVajZYtW6JatWrw8PBAWFgYkpOTDfrk5+cjKioKNWrUgLOzM/r27Yv09HSDPqmpqejevTuqVq0KDw8PTJo0CUVFRSbFwgRIREQWs2/fPkRFReHgwYPYtWsXCgsL0blzZ+Tl5en7jB8/Hlu2bMH69euxb98+pKWloU+fPvr9xcXF6N69OwoKCnDgwAGsXr0aq1atwrRp00yKRSZJkmS2b/aMKLx1xdohkCAcVe2tHQIJoqjgulnHu7fnS7ON5Rg84omPvXnzJjw8PLBv3z506NAB2dnZcHd3x7p16/Daa68BAM6fPw8/Pz8kJiaidevW+PXXX9GjRw+kpaXB09MTABAXF4fJkyfj5s2bsLe3N+rcnAESEYnIjCVQrVaLnJwcg02r1RoVRnZ2NgDAzc0NAHDs2DEUFhYiJCRE3+eFF16At7c3EhMTAQCJiYlo0qSJPvkBQGhoKHJycnDmzBmjfwImQCIieipqtRoKhcJgU6vVjz1Op9Nh3LhxaNu2LRo3bgwA0Gg0sLe3h6urq0FfT09PaDQafZ9/J7+S/SX7jMVVoEREIjLjo9BiY2MRExNj0CaXyx97XFRUFE6fPo39+/ebLRZTMAESEYnIjDfCy+VyoxLev0VHR2Pr1q1ISEhArVq19O1KpRIFBQXIysoymAWmp6dDqVTq+xw+fNhgvJJVoiV9jMESKBERWYwkSYiOjsaGDRuwe/du+Pr6Guxv3rw57OzsEB8fr29LTk5GamoqAgMDAQCBgYE4deoUMjIy9H127doFFxcX+Pv7Gx0LZ4BERCKy0tsgoqKisG7dOmzatAnVqlXTX7NTKBRwdHSEQqHA8OHDERMTAzc3N7i4uGDMmDEIDAxE69atAQCdO3eGv78/Bg0ahHnz5kGj0WDKlCmIiooyaSbKBEhEJCIrPQt0+fLlAICOHTsatK9cuRJDhgwBACxYsAA2Njbo27cvtFotQkNDsWzZMn3fKlWqYOvWrRg9ejQCAwPh5OSEiIgIzJo1y6RYeB8g0VPgfYBkKWa/D/DXxWYby7HrWLONZUmcARIRiYhvg2ACJCISEt8Iz1WgREQkJs4AiYhExBIoEyARkZBYAmUJlIiIxMQZIBGRiFgCZQIkIhISS6AsgRIRkZg4AyQiEhFLoEyARERCYgJkCZSIiMTEGSARkYgq33sQTMYESEQkIpZAWQIlIiIxcQZIRCQizgCZAImIhMQb4VkCJSIiMXEGSEQkIpZAmQCJiITE2yBYAiUiIjFxBkhEJCKWQJkAiYiExATIEigREYmJM0AiIhHxPkAmQCIiEUk6rgJlCZSIiITEGSARkYi4CIYJkIhISLwGyBIoERGJiTNAIiIRcREMEyARkZB4DZAlUCIispyEhAT07NkTKpUKMpkMGzduNNifm5uL6Oho1KpVC46OjvD390dcXJxBn/z8fERFRaFGjRpwdnZG3759kZ6ebnIsTIBERCLS6cy3mSAvLw9NmzbF0qVLy9wfExOD7du349tvv8W5c+cwbtw4REdHY/Pmzfo+48ePx5YtW7B+/Xrs27cPaWlp6NOnj8k/AUugREQistLrkLp27YquXbs+dP+BAwcQERGBjh07AgAiIyPx+eef4/Dhw+jVqxeys7Px1VdfYd26dXj55ZcBACtXroSfnx8OHjyI1q1bGx0LZ4BERPRUtFotcnJyDDatVvtEY7Vp0wabN2/G9evXIUkS9uzZgwsXLqBz584AgGPHjqGwsBAhISH6Y1544QV4e3sjMTHRpHMxARIRiciMJVC1Wg2FQmGwqdXqJwpryZIl8Pf3R61atWBvb48uXbpg6dKl6NChAwBAo9HA3t4erq6uBsd5enpCo9GYdC6WQCuxFd98j9/2/YGUa3/DQW6PgCb+GD96GHx9agEAsnPuYOmXa3Dg8HHcSL+J6tUVeLl9IMaMHIxqzk4AgPMXr+Crb3/A8ZNnkJWVA5WXJ/qFdcOgfmFW/GZU0bwzKQofznkPixZ/iQkTpwMAli2di04vt4NK5Ync3LtIPHgUse/NQXLyZStHKwgz3gYRGxuLmJgYgza5XP5EYy1ZsgQHDx7E5s2b4ePjg4SEBERFRUGlUhnM+syBCbASO5p0Cv379ERjv+dRVFyMRZ+vQuT497Fp7eeo6uiAjFu3kXErExOjR6BuHW/cSM/ArI8/w81bt7FgzhQAwNnki3Cr7oqPpk2C0sMdSafPYebcxahiY4M3Xutl5W9IFUGL5k0xcsRAnDh51qD9+PGT+O67n5H613W4VXfFtGkT8Ou271D/+dbQcYl+hSKXy5844f3bvXv38N5772HDhg3o3r07AODFF19EUlIS5s+fj5CQECiVShQUFCArK8tgFpieng6lUmnS+ZgAK7HPP51t8HnO+zHo0KM/ziZfRIuAJmhQtw4WfjhFv9+7lgpjIyPw7qx5KCoqhq1tFfTpEWowRu3nvHDi9Dn8tu8AEyA9lpNTVXzzzWcYNfodvBc71mDfl1+t1f/3tWt/Y9r0efjz2G+oU6c2rly5ZulQxfMMPgqtsLAQhYWFsLExvDpXpUoV/T+KmjdvDjs7O8THx6Nv374AgOTkZKSmpiIwMNCk81k1Ad66dQtff/01EhMT9bVbpVKJNm3aYMiQIXB3d7dmeJVObt5dAIDCpdpD+9zJzYOzU1XY2lZ5ZB+Fi7PZ46PKZ8niD/HrL/GI3/17qQT4b1WrOmLI4Ndx5co1/PVXmgUjFJiVngSTm5uLS5cu6T+npKQgKSkJbm5u8Pb2RlBQECZNmgRHR0f4+Phg3759+Oabb/Dpp58CABQKBYYPH46YmBi4ubnBxcUFY8aMQWBgoEkrQAErJsAjR44gNDQUVatWRUhICJ5//nkA96exixcvxkcffYQdO3agRYsWjxxHq9WWWm1ko9WaZTpemeh0Ony06HM0e9EfDerWKbPPP1nZ+HzVd3it18OXKP956ix2xCdg6cczyylSqiz69euFZs0ao3Vg94f2GfVmBD5Svw9nZyecT76ELt36o7Cw0IJRkqUdPXoUwcHB+s8l1w4jIiKwatUq/O9//0NsbCwGDBiAzMxM+Pj4YM6cORg1apT+mAULFsDGxgZ9+/aFVqtFaGgoli1bZnIsMkmyzs0grVu3RtOmTREXFweZTGawT5IkjBo1CidPnnzsstYZM2Zg5kzDv4ynTBqLae+8bfaYK7JZHy/B/oNH8c3y+VB6lJ5Z5+blYeS496FwqYYlc6fDzrb0v40uXrmKYWMmY+B/w/DmkP6WCPuZ56hqb+0Qnkm1aqlwKPEXdOnWH6dOnQMAxO9aj6QTZ/WLYADAxaUaPDxqwkvpgZiYUVCplOgQFPbES+grs6KC62YdL08dYbaxnGJXm20sS7JaAnR0dMSff/6JF154ocz958+fR7NmzXDv3r1HjlPmDPDOdc4A/2XOJ8uwe38iVi/9GLVUpS8S5+XdRWTMFDg6yLF03kzI5fal+lxOuYahY95F356hePvNIRaIumJgAixbr16h+PnHr1FUVKRvs7W1hU6ng06nQ1Vn31ILXezs7HAr4ywiR03E999vsnTIzzyzJ8A5g802ltP735htLEuyWglUqVTi8OHDD02Ahw8fhqen52PHKWv1UWHBLbPEWNFJkoQPP12O+IQDWPnZ3DKTX25eHt4cPwV29nZYMnd6mcnv0pVrGDb2XfTuGsLkR0bZvXs/mjZ72aDtyxWfIjn5Mj6ev7TMVZ4ymQwymQxye/7jlSzDaglw4sSJiIyMxLFjx9CpUyd9sktPT0d8fDxWrFiB+fPnWyu8SmH2J0vxy669WPzRNDhVdcSt25kAAGdnJzjI5cjNy0PkuPdxT6vFommTkJd3F3n/f6FMdVcFqlSpgotXrmL4mHfRplVzRIS/qh/DxsYGbtVdrfXV6BmXm5uHM2eSDdru5t3F7dv/4MyZZPj6eqPff3th1659uHnrNmo9p8I770Th3r18/Lo93kpRC+YZXAVqaVZLgFFRUahZsyYWLFiAZcuWobi4GMD95a7NmzfHqlWr0K9fP2uFVyl8v2EbAGBo9GSD9tnvxSCs+ys4m3wZJ8/e/0uq2+vDDfrs+HEVnvPyxM49+5GZlY2tO3Zj647d+v0qpQd2/lQx6/5kffn5WrRr+x+MHTMC1asrkJ5+C7/vP4j2Qb1x8+Zta4cnBr4P0HrXAP+tsLAQt27dL1vWrFkTdnZ2TzferSvmCIvosXgNkCzF7NcAZw0w21hO09Y+vtMz6Jm4Ed7Ozg5eXl7WDoOISBx82s6zkQCJiMjCWALl2yCIiEhMnAESEYmIq0CZAImIhMQSKEugREQkJs4AiYgEJHEVKGeAREQkJs4AiYhExGuATIBEREJiAmQJlIiIxMQZIBGRiHgfIBMgEZGQWAJlCZSIiMTEGSARkYAkzgCZAImIhMQEyBIoERGJiTNAIiIR8VFoTIBEREJiCZQlUCIiEhNngEREIuIMkAmQiEhEksQEyBIoEREJiTNAIiIRsQTKBEhEJCQmQJZAiYhITJwBEhEJiM8C5QyQiEhMOsl8mwkSEhLQs2dPqFQqyGQybNy4sVSfc+fOoVevXlAoFHByckLLli2Rmpqq35+fn4+oqCjUqFEDzs7O6Nu3L9LT003+CZgAiYjIYvLy8tC0aVMsXbq0zP2XL19Gu3bt8MILL2Dv3r04efIkpk6dCgcHB32f8ePHY8uWLVi/fj327duHtLQ09OnTx+RYZFIlvBmk8NYVa4dAgnBUtbd2CCSIooLrZh0ve1Ans42lWBP/RMfJZDJs2LABYWFh+rbw8HDY2dlhzZo1ZR6TnZ0Nd3d3rFu3Dq+99hoA4Pz58/Dz80NiYiJat25t9Pk5AyQiEpCkk8y2abVa5OTkGGxardbkmHQ6HbZt24bnn38eoaGh8PDwQKtWrQzKpMeOHUNhYSFCQkL0bS+88AK8vb2RmJho0vmYAImI6Kmo1WooFAqDTa1WmzxORkYGcnNz8dFHH6FLly7YuXMnXn31VfTp0wf79u0DAGg0Gtjb28PV1dXgWE9PT2g0GpPOx1WgREQiMuMq0NjYWMTExBi0yeVyk8fR/f9XNPXu3Rvjx48HAAQEBODAgQOIi4tDUFDQ0wf7L0yAREQiMuPrAOVy+RMlvAfVrFkTtra28Pf3N2j38/PD/v37AQBKpRIFBQXIysoymAWmp6dDqVSadD6WQImI6Jlgb2+Pli1bIjk52aD9woUL8PHxAQA0b94cdnZ2iI//v4U3ycnJSE1NRWBgoEnn4wyQiEhA1roRPjc3F5cuXdJ/TklJQVJSEtzc3ODt7Y1Jkybh9ddfR4cOHRAcHIzt27djy5Yt2Lt3LwBAoVBg+PDhiImJgZubG1xcXDBmzBgEBgaatAIUYAIkIhKTGUugpjh69CiCg4P1n0uuHUZERGDVqlV49dVXERcXB7VajbFjx6Jhw4b46aef0K5dO/0xCxYsgI2NDfr27QutVovQ0FAsW7bM5Fh4HyDRU+B9gGQp5r4P8J++Hc02VvWf9pptLEviDJCISEB8FigTIBGRmKxUAn2WcBUoEREJiTNAIiIBSZwBMgESEQmJCZAlUCIiEhNngEREAmIJlAmQiEhMTIAsgRIRkZg4AyQiEhBLoEyARERCYgJkCZSIiATFGSARkYA4A2QCJCISkySzdgRWxxIoEREJiTNAIiIBsQTKBEhEJCRJxxIoS6BERCQkzgCJiATEEqiRCfDkyZNGD/jiiy8+cTBERGQZEleBGpcAAwICIJPJIElSmftL9slkMhQXF5s1QCIiovJgVAJMSUkp7ziIiMiCWAI1MgH6+PiUdxxERGRBXAX6hKtA16xZg7Zt20KlUuHatWsAgIULF2LTpk1mDY6IiKi8mJwAly9fjpiYGHTr1g1ZWVn6a36urq5YuHChueMjIqJyIEnm2yoqkxPgkiVLsGLFCrz//vuoUqWKvr1FixY4deqUWYMjIqLyIelkZtsqKpMTYEpKCpo1a1aqXS6XIy8vzyxBERERlTeTE6Cvry+SkpJKtW/fvh1+fn7miImIiMoZZ4BP8CSYmJgYREVFIT8/H5Ik4fDhw/juu++gVqvx5ZdflkeMRERkZhX52p25mJwAR4wYAUdHR0yZMgV3797FG2+8AZVKhUWLFiE8PLw8YiQiIjI7mfSwx7sY4e7du8jNzYWHh4c5Y3pqhbeuWDsEEoSjqr21QyBBFBVcN+t4V5p0NttYdU/tNNtYlvTED8POyMhAcnIygPuPQnN3dzdbUEREVL74LNAnWARz584dDBo0CCqVCkFBQQgKCoJKpcLAgQORnZ1dHjESEVElkZCQgJ49e0KlUkEmk2Hjxo0P7Ttq1CjIZLJS95hnZmZiwIABcHFxgaurK4YPH47c3FyTYzE5AY4YMQKHDh3Ctm3bkJWVhaysLGzduhVHjx7Fm2++aXIARERkeZLOfJsp8vLy0LRpUyxduvSR/TZs2ICDBw9CpVKV2jdgwACcOXMGu3btwtatW5GQkIDIyEjTAsETlEC3bt2KHTt2oF27dvq20NBQrFixAl26dDE5ACIisjydlUqgXbt2RdeuXR/Z5/r16xgzZgx27NiB7t27G+w7d+4ctm/fjiNHjqBFixYA7j+gpVu3bpg/f36ZCfNhTJ4B1qhRAwqFolS7QqFA9erVTR2OiIgqOK1Wi5ycHINNq9U+0Vg6nQ6DBg3CpEmT0KhRo1L7ExMT4erqqk9+ABASEgIbGxscOnTIpHOZnACnTJmCmJgYaDQafZtGo8GkSZMwdepUU4cjIiIrkCSZ2Ta1Wg2FQmGwqdXqJ4pr7ty5sLW1xdixY8vcr9FoSt15YGtrCzc3N4O8ZAyjSqDNmjWDTPZ/0+WLFy/C29sb3t7eAIDU1FTI5XLcvHmT1wGJiCoAcz7BJTY2FjExMQZtcrnc5HGOHTuGRYsW4fjx4wY5p7wYlQDDwsLKOQwiIqqo5HL5EyW8B/3+++/IyMjQT64AoLi4GBMmTMDChQtx9epVKJVKZGRkGBxXVFSEzMxMKJVKk85nVAKcPn26SYMSEdGz7Vl8FNqgQYMQEhJi0BYaGopBgwZh6NChAIDAwEBkZWXh2LFjaN68OQBg9+7d0Ol0aNWqlUnne+Ib4YmIqOKy1kOsc3NzcenSJf3nlJQUJCUlwc3NDd7e3qhRo4ZBfzs7OyiVSjRs2BAA4Ofnhy5dumDkyJGIi4tDYWEhoqOjER4ebtIKUOAJEmBxcTEWLFiAH374AampqSgoKDDYn5mZaeqQREQkiKNHjyI4OFj/ueTaYUREBFatWmXUGGvXrkV0dDQ6deoEGxsb9O3bF4sXLzY5FpMT4MyZM/Hll19iwoQJmDJlCt5//31cvXoVGzduxLRp00wOgIiILM9a9wF27NgRpjyC+urVq6Xa3NzcsG7duqeOxeTbINauXYsVK1ZgwoQJsLW1Rf/+/fHll19i2rRpOHjw4FMHRERE5c+ct0FUVCYnQI1GgyZNmgAAnJ2d9c//7NGjB7Zt22be6IiIiMqJyQmwVq1auHHjBgCgXr162Lnz/mswjhw5YpZlsEREVP4kyXxbRWVyAnz11VcRHx8PABgzZgymTp2KBg0aYPDgwRg2bJjZAyQiIvPTSTKzbRXVU70QFwAOHjyIAwcOoEGDBujZs6e54noqfCEuWQpfiEuWYu4X4ib59DLbWAHXNpttLEsyeQb4oNatWyMmJgatWrXChx9+aI6YiIionHERjBkSYIkbN27wYdhERBUErwGaMQESERFVJHwUGhGRgCry4hVzqZQJ8LOX+EQasgx/N+/HdyJ6BlXka3fmYnQCfPBdTw+6efPmUwdDRERkKUYnwD///POxfTp06PBUwRARkWWwBGpCAtyzZ095xkFERBZUgRdvmg1XgRIRkZAq5SIYIiJ6NJZAmQCJiITEVaAsgRIRkaA4AyQiEpDO2gE8A55oBvj7779j4MCBCAwMxPXr959QvmbNGuzfv9+swRERUfmQIDPbVlGZnAB/+uknhIaGwtHREX/++Se0Wi0AIDs7m2+DICKiCsPkBDh79mzExcVhxYoVsLOz07e3bdsWx48fN2twRERUPnSS+baKyuRrgMnJyWU+8UWhUCArK8scMRERUTnTVeDSpbmYPANUKpW4dOlSqfb9+/ejbt26ZgmKiIiovJmcAEeOHIm3334bhw4dgkwmQ1paGtauXYuJEydi9OjR5REjERGZGRfBPEEJ9N1334VOp0OnTp1w9+5ddOjQAXK5HBMnTsSYMWPKI0YiIjIz3gbxBAlQJpPh/fffx6RJk3Dp0iXk5ubC398fzs7O5REfERFRuXjiG+Ht7e3h7+9vzliIiMhCKnLp0lxMToDBwcGQyR7+w+3evfupAiIiovLHEugTJMCAgACDz4WFhUhKSsLp06cRERFhrriIiIjKlckJcMGCBWW2z5gxA7m5uU8dEBERlT/OAM34NoiBAwfi66+/NtdwRERUjngbhBkTYGJiIhwcHMw1HBERUbkyOQH26dPHYHv11VfRunVrDB06FG+++WZ5xEhERGamk5lvM0VCQgJ69uwJlUoFmUyGjRs36vcVFhZi8uTJaNKkCZycnKBSqTB48GCkpaUZjJGZmYkBAwbAxcUFrq6uGD58+BNdgjM5ASoUCoPNzc0NHTt2xC+//ILp06ebHAAREVmeDjKzbabIy8tD06ZNsXTp0lL77t69i+PHj2Pq1Kk4fvw4fv75ZyQnJ6NXr14G/QYMGIAzZ85g165d2Lp1KxISEhAZGWnybyCTJMnoZ3kXFxfjjz/+QJMmTVC9enWTT2YpC7wHWjsEEsTKgsvWDoEEcVKTaNbxNinfMNtYvTXrnug4mUyGDRs2ICws7KF9jhw5gv/85z+4du0avL29ce7cOfj7++PIkSNo0aIFAGD79u3o1q0b/v77b6hUKqPPb9IMsEqVKujcuTPf+kBEVMFJZty0Wi1ycnIMtpJ3xT6t7OxsyGQyuLq6Ari/3sTV1VWf/AAgJCQENjY2OHTokEljm1wCbdy4Ma5cuWLqYURE9AzRmXFTq9WlLo+p1eqnjjE/Px+TJ09G//794eLiAgDQaDTw8PAw6Gdraws3NzdoNBqTxjf5PsDZs2dj4sSJ+OCDD9C8eXM4OTkZ7C8JkoiIxBAbG4uYmBiDNrlc/lRjFhYWol+/fpAkCcuXL3+qsR7G6AQ4a9YsTJgwAd26dQMA9OrVy+CRaJIkQSaTobi42PxREhGRWeke8UhLU8nl8qdOeP9WkvyuXbuG3bt3G0yslEolMjIyDPoXFRUhMzMTSqXSpPMYnQBnzpyJUaNGYc+ePSadgIiInj1Gr360sJLkd/HiRezZswc1atQw2B8YGIisrCwcO3YMzZs3B3D/GdQ6nQ6tWrUy6VxGJ8CSxaJBQUEmnYCIiKhEbm4uLl26pP+ckpKCpKQkuLm5wcvLC6+99hqOHz+OrVu3ori4WH9dz83NDfb29vDz80OXLl0wcuRIxMXFobCwENHR0QgPDzdpBShg4jXAR70FgoiIKg5rPQv06NGjCA4O1n8uuXYYERGBGTNmYPPmzQBKv3hhz5496NixIwBg7dq1iI6ORqdOnWBjY4O+ffti8eLFJsdiUgJ8/vnnH5sEMzMzTQ6CiIgsy9QnuJhLx44d8ajbz425Nd3NzQ3r1j3ZvYf/ZlICnDlzJhQKxVOflIiIyNpMSoDh4eGl7r8gIqKKx9RHmFVGRidAXv8jIqo8ntVVoJZk9JNgTHhkKBER0TPP6BmgTsf3BxMRVRbWWgTzLDH5UWhERFTxcUpjxjfCExERVSScARIRCYirOpgAiYiExGuALIESEZGgOAMkIhIQF8EwARIRCYkJkCVQIiISFGeAREQCkrgIhgmQiEhELIGyBEpERILiDJCISECcATIBEhEJiU+CYQmUiIgExRkgEZGA+Cg0JkAiIiHxGiBLoEREJCjOAImIBMQZIBMgEZGQuAqUJVAiIhIUZ4BERALiKlAmQCIiIfEaIEugREQkKM4AiYgExEUwTIBERELSMQWyBEpERGLiDJCISEBcBMMZIBGRkCQzbqZISEhAz549oVKpIJPJsHHjRsO4JAnTpk2Dl5cXHB0dERISgosXLxr0yczMxIABA+Di4gJXV1cMHz4cubm5JkbCBEhERBaUl5eHpk2bYunSpWXunzdvHhYvXoy4uDgcOnQITk5OCA0NRX5+vr7PgAEDcObMGezatQtbt25FQkICIiMjTY6FJVAiIgFZqwTatWtXdO3atcx9kiRh4cKFmDJlCnr37g0A+Oabb+Dp6YmNGzciPDwc586dw/bt23HkyBG0aNECALBkyRJ069YN8+fPh0qlMjoWzgCJiASkk5lv02q1yMnJMdi0Wq3JMaWkpECj0SAkJETfplAo0KpVKyQmJgIAEhMT4erqqk9+ABASEgIbGxscOnTIpPMxARIR0VNRq9VQKBQGm1qtNnkcjUYDAPD09DRo9/T01O/TaDTw8PAw2G9raws3Nzd9H2OxBEpEJCBz3gf4fmwsYmJiDNrkcrnZxi8vTIBERAIy523wcrncLAlPqVQCANLT0+Hl5aVvT09PR0BAgL5PRkaGwXFFRUXIzMzUH28slkCJiOiZ4OvrC6VSifj4eH1bTk4ODh06hMDAQABAYGAgsrKycOzYMX2f3bt3Q6fToVWrViadjzNAIiIBWWsVaG5uLi5duqT/nJKSgqSkJLi5ucHb2xvjxo3D7Nmz0aBBA/j6+mLq1KlQqVQICwsDAPj5+aFLly4YOXIk4uLiUFhYiOjoaISHh5u0AhRgAiQiEpK1ngV69OhRBAcH6z+XXDuMiIjAqlWr8M477yAvLw+RkZHIyspCu3btsH37djg4OOiPWbt2LaKjo9GpUyfY2Nigb9++WLx4scmxyCRJqnRPRF3gPdDaIZAgVhZctnYIJIiTmkSzjje5Tn+zjTX36ndmG8uSOAMkIhJQpZv5PAEmQCIiAfFh2FwFSkREguIMkIhIQHwhLhMgEZGQmP5YAiUiIkFxBkhEJCAugmECJCISksQiKEugREQkJs4AiYgExBIoEyARkZB4GwRLoEREJCjOAImIBMT5HxMgEZGQWAJlAqzUnvtPQ7QY1R0eTXzh7Fkdm0cswOWdxwz6BMb0RZM3giF3qYq0oxcQ/95KZF1NBwDUau2H//7wfpljr+sxDeknr5T7d6CKy0PpjnFT3kK7lwPh4OiAv67+janjZuPsifOl+k6Z+w76RbyKeVMX4tsV31shWhIRE2AlZldVjptnU3H6+wT0WjGu1P4Wo3sgYGhn7Ij5HDl/3USbia+hz7eTsbrTZBRrC5F27AI+bx5lcEybia/Bu20jJj96pGqKali95XMc+eMY3hoQg39u/wNv39rIybpTqu/LXYPwYvNGSL9x0wqRiourQJkAK7Wre0/i6t6TD93/0vAuOLxkE67sOg4A2D4+Dm8eW4p6nZvjwpaD0BUW4+7NbH1/G9sqqPfKS0hatavcY6eKbVj0QKRfT8e0cXP0bddTb5Tq56F0R+ycGIzqPw6fffuJJUMUHm+E5ypQYSm83eHk4YrU/af1bQV37kGTdBmq5g3KPKbuKy/BoXo1nPkhwVJhUgXVMbQ9zpw4j/kr5mDv6W34ftdq9B3Qy6CPTCbDh59Nw6pla3E5OcVKkZLIKvwMUKvVQqvVGrQVScWwlVWxUkQVQ1V3VwDA3Vs5Bu13b+WgqruizGMavx6Ea/tOIleTWd7hUQVXy1uFfhGvYs3n/8OXi1ajUYAfJs+OQWFhETb/8AsAYFj0IBQVFWPtlz9YOVoxsQT6jM8A//rrLwwbNuyRfdRqNRQKhcH2W84ZC0UoDmelG3yCXsTp7/dZOxSqAGxsbHDu1AUsVsfh/OkL+OnbTfhp7Sb8d3AYAMDvxYYYMLIfpr4927qBCkwy4/8qqmc6AWZmZmL16tWP7BMbG4vs7GyDLcSlkYUirLju3swCAFSt6WLQXrWmi8F1vxKN+nVA/j939NcLiR7lZsYtXLlgWNZMuXgVyueUAIDmrQLgVrM6dhzbgON//47jf/+O52p7YcKMMfj1yM/WCJkEZNUS6ObNmx+5/8qVx680lMvlkMvlBm0sfz5edupN5GVkoXbbRrh5NhUAYO/sCGVAPZxYE1+qf6N+HXD2p/3QFRVbOlSqgJIOn0Kdet4GbT51vXHjbw0AYMuPv+Lg70cM9i//biG2/vgrNv1vm8XiFBlLoFZOgGFhYZDJZJCkh0+hZTKZBSOqXOyqyuFax1P/2aW2O9z9vZGflYc7abdx/KvtaDU2DFlX05GdmoE2E19DXkZWqXsFa7dtBIW3B07/b6+FvwFVVGu++B++2fIFRoyNwI7N8WjSzB+vDeqNmRM/AgBk/5OD7H8Mrz8XFRXhdkYmrl5OtUbIwtE94u9dUVg1AXp5eWHZsmXo3bt3mfuTkpLQvHlzC0dVeXi+WNfgRvaO0wcCAM6sT8DOCV/g6PKtsHOUI0Q9TH8j/M+D5qFYW2gwTuPXg5B29AL+uVx6GTtRWc4kncP4Ye/i7fdG482YobieegPzpi7ELz/vtHZoRHoy6VHTr3LWq1cvBAQEYNasWWXuP3HiBJo1awadzrTJ+gLvgeYIj+ixVhZctnYIJIiTmkSzjjfQp4/Zxvr2WsW8bmvVGeCkSZOQl5f30P3169fHnj17LBgREZEY+CxQKyfA9u3bP3K/k5MTgoKCLBQNERGJpMLfCE9ERKaryPfvmQsTIBGRgHgbxDN+IzwREVF54QyQiEhAXATDGSAREQmKCZCISEDWehh2cXExpk6dCl9fXzg6OqJevXr44IMPDJ4IJkkSpk2bBi8vLzg6OiIkJAQXL14090/ABEhEJCKdGTdTzJ07F8uXL8dnn32Gc+fOYe7cuZg3bx6WLFmi7zNv3jwsXrwYcXFxOHToEJycnBAaGor8/Pyn+cql8BogERFZzIEDB9C7d290794dAFCnTh189913OHz4MID7s7+FCxdiypQp+sdkfvPNN/D09MTGjRsRHh5utlg4AyQiEpAkSWbbtFotcnJyDLYHX1Reok2bNoiPj8eFCxcA3H/k5f79+9G1a1cAQEpKCjQaDUJCQvTHKBQKtGrVComJ5n0cHBMgEZGAdJDMtpX1YnK1Wl3med99912Eh4fjhRdegJ2dHZo1a4Zx48ZhwIABAACN5v4rszw9PQ2O8/T01O8zF5ZAiYjoqcTGxiImJsag7cH3tJb44YcfsHbtWqxbtw6NGjVCUlISxo0bB5VKhYiICEuEq8cESEQkIHM+CaasF5M/zKRJk/SzQABo0qQJrl27BrVajYiICCiVSgBAeno6vLy89Melp6cjICDAjFGzBEpEJCRr3QZx9+5d2NgYpp4qVaroX3vn6+sLpVKJ+Ph4/f6cnBwcOnQIgYGBT//F/4UzQCIispiePXtizpw58Pb2RqNGjfDnn3/i008/xbBhwwAAMpkM48aNw+zZs9GgQQP4+vpi6tSpUKlUCAsLM2ssTIBERAKy1qPQlixZgqlTp+Ktt95CRkYGVCoV3nzzTUybNk3f55133kFeXh4iIyORlZWFdu3aYfv27XBwcDBrLFZ9I3x54RvhyVL4RniyFHO/Eb5r7a5mG+vXv34121iWxGuAREQkJJZAiYgExPcBMgESEQmJb4RnCZSIiATFGSARkYD4QlwmQCIiIVXCGwBMxhIoEREJiTNAIiIBsQTKBEhEJCSuAmUJlIiIBMUZIBGRgHRcBMMESEQkIqY/lkCJiEhQnAESEQmIq0CZAImIhMQEyBIoEREJijNAIiIB8VFoTIBEREJiCZQlUCIiEhRngEREAuKj0JgAiYiExGuALIESEZGgOAMkIhIQF8EwARIRCYklUJZAiYhIUJwBEhEJiCVQJkAiIiHxNgiWQImISFCcARIRCYhvhGcCJCISEkugLIESEZGgOAMkIhIQS6CcARIRCUky4/9Mdf36dQwcOBA1atSAo6MjmjRpgqNHj/5fbJKEadOmwcvLC46OjggJCcHFixfN+fUBMAESEZEF/fPPP2jbti3s7Ozw66+/4uzZs/jkk09QvXp1fZ958+Zh8eLFiIuLw6FDh+Dk5ITQ0FDk5+ebNRaWQImIBGStEujcuXNRu3ZtrFy5Ut/m6+ur/29JkrBw4UJMmTIFvXv3BgB888038PT0xMaNGxEeHm62WDgDJCISkDlLoFqtFjk5OQabVqst87ybN29GixYt8N///hceHh5o1qwZVqxYod+fkpICjUaDkJAQfZtCoUCrVq2QmJho1t+ACZCIiJ6KWq2GQqEw2NRqdZl9r1y5guXLl6NBgwbYsWMHRo8ejbFjx2L16tUAAI1GAwDw9PQ0OM7T01O/z1xYAiUiEpA5S6CxsbGIiYkxaJPL5WWfV6dDixYt8OGHHwIAmjVrhtOnTyMuLg4RERFmi8kYnAESEQnInCVQuVwOFxcXg+1hCdDLywv+/v4GbX5+fkhNTQUAKJVKAEB6erpBn/T0dP0+c2ECJCIii2nbti2Sk5MN2i5cuAAfHx8A9xfEKJVKxMfH6/fn5OTg0KFDCAwMNGssLIESEQlIknRWOe/48ePRpk0bfPjhh+jXrx8OHz6ML774Al988QUAQCaTYdy4cZg9ezYaNGgAX19fTJ06FSqVCmFhYWaNhQmQiEhA1nofYMuWLbFhwwbExsZi1qxZ8PX1xcKFCzFgwAB9n3feeQd5eXmIjIxEVlYW2rVrh+3bt8PBwcGsscgkqfI9D2eB90Brh0CCWFlw2dohkCBOasx7C4BPjRfNNta12yfNNpYlcQZIRCSgSjj3MRkTIBGRgKxVAn2WcBUoEREJiTNAIiIBsQTKBEhEJCS+D5AlUCIiEhRngEREAnqSF9lWNkyAREQC4jVAlkCJiEhQnAESEQmI9wEyARIRCYklUJZAiYhIUJwBEhEJiPcBMgESEQmJJVCWQImISFCcARIRCYirQJkAiYiExBIoS6BERCQozgCJiATEVaBMgEREQuLDsFkCJSIiQXEGSEQkIJZAmQCJiITEVaAsgRIRkaA4AyQiEhAXwTABEhEJiSVQlkCJiEhQnAESEQmIM0AmQCIiITH9sQRKRESCkkmcBxMArVYLtVqN2NhYyOVya4dDlRj/rNGzggmQAAA5OTlQKBTIzs6Gi4uLtcOhSox/1uhZwRIoEREJiQmQiIiExARIRERCYgIkAIBcLsf06dO5KIHKHf+s0bOCi2CIiEhInAESEZGQmACJiEhITIBERCQkJkAiIhISEyBh6dKlqFOnDhwcHNCqVSscPnzY2iFRJZSQkICePXtCpVJBJpNh48aN1g6JBMcEKLjvv/8eMTExmD59Oo4fP46mTZsiNDQUGRkZ1g6NKpm8vDw0bdoUS5cutXYoRAB4G4TwWrVqhZYtW+Kzzz4DAOh0OtSuXRtjxozBu+++a+XoqLKSyWTYsGEDwsLCrB0KCYwzQIEVFBTg2LFjCAkJ0bfZ2NggJCQEiYmJVoyMiKj8MQEK7NatWyguLoanp6dBu6enJzQajZWiIiKyDCZAIiISEhOgwGrWrIkqVaogPT3doD09PR1KpdJKURERWQYToMDs7e3RvHlzxMfH69t0Oh3i4+MRGBhoxciIiMqfrbUDIOuKiYlBREQEWrRogf/85z9YuHAh8vLyMHToUGuHRpVMbm4uLl26pP+ckpKCpKQkuLm5wdvb24qRkah4GwThs88+w8cffwyNRoOAgAAsXrwYrVq1snZYVMns3bsXwcHBpdojIiKwatUqywdEwmMCJCIiIfEaIBERCYkJkIiIhMQESEREQmICJCIiITEBEhGRkJgAiYhISEyAREQkJCZAIiISEhMgVVpDhgwxeOFqx44dMW7cOIvHsXfvXshkMmRlZZXbOR78rk/CEnESPUuYAMmihgwZAplMBplMBnt7e9SvXx+zZs1CUVFRuZ/7559/xgcffGBUX0sngzp16mDhwoUWORcR3ceHYZPFdenSBStXroRWq8Uvv/yCqKgo2NnZITY2tlTfgoIC2Nvbm+W8bm5uZhmHiCoHzgDJ4uRyOZRKJXx8fDB69GiEhIRg8+bNAP6vlDdnzhyoVCo0bNgQAPDXX3+hX79+cHV1hZubG3r37o2rV6/qxywuLkZMTAxcXV1Ro0YNvPPOO3jwMbcPlkC1Wi0mT56M2rVrQy6Xo379+vjqq69w9epV/UObq1evDplMhiFDhgC4/7ootVoNX19fODo6omnTpvjxxx8NzvPLL7/g+eefh6OjI4KDgw3ifBLFxcUYPny4/pwNGzbEokWLyuw7c+ZMuLu7w8XFBaNGjUJBQYF+nzGxE4mEM0CyOkdHR9y+fVv/OT4+Hi4uLti1axcAoLCwEKGhoQgMDMTvv/8OW1tbzJ49G126dMHJkydhb2+PTz75BKtWrcLXX38NPz8/fPLJJ9iwYQNefvnlh5538ODBSExMxOLFi9G0aVOkpKTg1q1bqF27Nn766Sf07dsXycnJcHFxgaOjIwBArVbj22+/RVxcHBo0aICEhAQMHDgQ7u7uCAoKwl9//YU+ffogKioKkZGROHr0KCZMmPBUv49Op0OtWrWwfv161KhRAwcOHEBkZCS8vLzQr18/g9/NwcEBe/fuxdWrVzF06FDUqFEDc+bMMSp2IuFIRBYUEREh9e7dW5IkSdLpdNKuXbskuVwuTZw4Ub/f09NT0mq1+mPWrFkjNWzYUNLpdPo2rVYrOTo6Sjt27JAkSZK8vLykefPm6fcXFhZKtWrV0p9LkiQpKChIevvttyVJkqTk5GQJgLRr164y49yzZ48EQPrnn3/0bfn5+VLVqlWlAwcOGPQdPny41L9/f0mSJCk2Nlby9/c32D958uRSYz3Ix8dHWrBgwUP3PygqKkrq27ev/nNERITk5uYm5eXl6duWL18uOTs7S8XFxUbFXtZ3JqrMOAMki9u6dSucnZ1RWFgInU6HN954AzNmzNDvb9KkicF1vxMnTuDSpUuoVq2awTj5+fm4fPkysrOzcePGDYN3GNra2qJFixalyqAlkpKSUKVKFZNmPpcuXcLdu3fxyiuvGLQXFBSgWbNmAIBz586VepdiYGCg0ed4mKVLl+Lrr79Gamoq7t27h4KCAgQEBBj0adq0KapWrWpw3tzcXPz111/Izc19bOxEomECJIsLDg7G8uXLYW9vD5VKBVtbwz+GTk5OBp9zc3PRvHlzrF27ttRY7u7uTxRDSUnTFLm5uQCAbdu24bnnnjPYJ5fLnygOY/zvf//DxIkT8cknnyAwMBDVqlXDxx9/jEOHDhk9hrViJ3qWMQGSxTk5OaF+/fpG93/ppZfw/fffw8PDAy4uLmX28fLywqFDh9ChQwcAQFFREY4dO4aXXnqpzP5NmjSBTqfDvn37EBISUmp/yQy0uLhY3+bv7w+5XI7U1NSHzhz9/Pz0C3pKHDx48PFf8hH++OMPtGnTBm+99Za+7fLly6X6nThxAvfu3dMn94MHD8LZ2Rm1a9eGm5vbY2MnEg1XgdIzb8CAAahZsyZ69+6N33//HSkpKdi7dy/Gjh2Lv//+GwDw9ttv46OPPsLGjRtx/vx5vPXWW4+8h69OnTqIiIjAsGHDsHHjRv2YP/zwAwDAx8cHMpkMW7duxc2bN5Gbm4tq1aph4sSJGD9+PFavXo3Lly/j+PHjWLJkCVavXg0AGDVqFC5evIhJkyYhOTkZ69atw6pVq4z6ntevX0dSUpLB9s8//6BBgwY4evQoduzYgQsXLmDq1Kk4cuRIqeMLCgowfPhwnD17Fr/88gumT5+O6Oho2NjYGBU7kXCsfRGSxPLvRTCm7L9x44Y0ePBgqWbNmpJcLpfq1q0rjRw5UsrOzpYk6f6il7fffltycXGRXF1dpZiYGGnw4MEPXQQjSZJ07949afz48ZKXl5dkb28v1a9fX/r666/1+2fNmiUplUpJJpNJERERkiTdX7izcOFCqWHDhpKdnZ3k7u4uhYaGSvv27dMft2XLFql+/fqSXC6X2rdvL3399ddGLYIBUGpbs2aNlJ+fLw0ZMkRSKBSSq6urNHr0aOndd9+VmjZtWup3mzZtmlSjRg3J2dlZGjlypJSfn6/v87jYuQiGRCOTpIesEiAiIqrEWAIlIiIhMQESEZGQmACJiEhITIBERCQkJkAiIhISEyAREQmJCZCIiITEBEhEREJiAiQiIiExARIRkZCYAImISEj/DwRRz/kKI7SOAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 500x500 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "import seaborn as sns\n",
    "\n",
    "# Convert predictions to binary format\n",
    "y_pred_binary = (y_pred.squeeze() > 0.5).float()\n",
    "\n",
    "# Create the confusion matrix\n",
    "cm = confusion_matrix(test_output.numpy(), y_pred_binary.numpy())\n",
    "\n",
    "# Plot the confusion matrix\n",
    "plt.figure(figsize=(5, 5))\n",
    "sns.heatmap(cm, annot=True, fmt='g')\n",
    "plt.title('Confusion Matrix')\n",
    "plt.xlabel('Predicted Label')\n",
    "plt.ylabel('True Label')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='12-display-Predictions'></a> \n",
    "### 12. Display Predictions"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The final step in our predictive analysis is to present our neural network's forecasts in a user-friendly manner. This entails translating the network's output into a comprehensible format, which involves mapping the predicted and actual match results to the corresponding match data.\n",
    "\n",
    "We will utilise our model's output to craft a DataFrame combining match details such as date, team, opponent, and round, with both predicted and actual outcomes. To reflect the complete picture of the match, we will merge the home and away predictions to deduce a unified outcome. The final match result, from the perspective of the home team, will be distilled into a win, draw, or loss based on logical decision rules.\n",
    "\n",
    "The following code segment walks through these steps, including DataFrame creation, data merging, and the application of a function to interpret the final predictions."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2103,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predictions vs Actual:\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>Date</th>\n",
       "      <th>Team</th>\n",
       "      <th>Venue</th>\n",
       "      <th>Opponent</th>\n",
       "      <th>Round</th>\n",
       "      <th>Predicted</th>\n",
       "      <th>Actual</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>57</th>\n",
       "      <td>2023-09-24</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Home</td>\n",
       "      <td>Tottenham</td>\n",
       "      <td>Matchweek 6</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>58</th>\n",
       "      <td>2023-09-30</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Away</td>\n",
       "      <td>Bournemouth</td>\n",
       "      <td>Matchweek 7</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>59</th>\n",
       "      <td>2023-10-08</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Home</td>\n",
       "      <td>Manchester City</td>\n",
       "      <td>Matchweek 8</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>60</th>\n",
       "      <td>2023-12-09</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Away</td>\n",
       "      <td>Aston Villa</td>\n",
       "      <td>Matchweek 16</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>61</th>\n",
       "      <td>2023-12-17</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Home</td>\n",
       "      <td>Brighton</td>\n",
       "      <td>Matchweek 17</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1471</th>\n",
       "      <td>2023-12-17</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Away</td>\n",
       "      <td>West Ham</td>\n",
       "      <td>Matchweek 17</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1472</th>\n",
       "      <td>2023-12-24</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Home</td>\n",
       "      <td>Chelsea</td>\n",
       "      <td>Matchweek 18</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1473</th>\n",
       "      <td>2023-12-27</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Away</td>\n",
       "      <td>Brentford</td>\n",
       "      <td>Matchweek 19</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1474</th>\n",
       "      <td>2023-12-30</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Home</td>\n",
       "      <td>Everton</td>\n",
       "      <td>Matchweek 20</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1475</th>\n",
       "      <td>2024-02-25</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Home</td>\n",
       "      <td>Sheffield Utd</td>\n",
       "      <td>Matchweek 26</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>436 rows × 7 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           Date                     Team Venue         Opponent         Round  \\\n",
       "57   2023-09-24                  Arsenal  Home        Tottenham   Matchweek 6   \n",
       "58   2023-09-30                  Arsenal  Away      Bournemouth   Matchweek 7   \n",
       "59   2023-10-08                  Arsenal  Home  Manchester City   Matchweek 8   \n",
       "60   2023-12-09                  Arsenal  Away      Aston Villa  Matchweek 16   \n",
       "61   2023-12-17                  Arsenal  Home         Brighton  Matchweek 17   \n",
       "...         ...                      ...   ...              ...           ...   \n",
       "1471 2023-12-17  Wolverhampton Wanderers  Away         West Ham  Matchweek 17   \n",
       "1472 2023-12-24  Wolverhampton Wanderers  Home          Chelsea  Matchweek 18   \n",
       "1473 2023-12-27  Wolverhampton Wanderers  Away        Brentford  Matchweek 19   \n",
       "1474 2023-12-30  Wolverhampton Wanderers  Home          Everton  Matchweek 20   \n",
       "1475 2024-02-25  Wolverhampton Wanderers  Home    Sheffield Utd  Matchweek 26   \n",
       "\n",
       "      Predicted  Actual  \n",
       "57          1.0     0.0  \n",
       "58          1.0     1.0  \n",
       "59          1.0     1.0  \n",
       "60          0.0     0.0  \n",
       "61          1.0     1.0  \n",
       "...         ...     ...  \n",
       "1471        0.0     0.0  \n",
       "1472        0.0     1.0  \n",
       "1473        0.0     1.0  \n",
       "1474        0.0     1.0  \n",
       "1475        0.0     1.0  \n",
       "\n",
       "[436 rows x 7 columns]"
      ]
     },
     "execution_count": 2103,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "# Set the model to evaluation mode\n",
    "model.eval()\n",
    "\n",
    "# Generate predictions\n",
    "with torch.no_grad():  # Disable gradient computation\n",
    "    predictions = model(test_input)\n",
    "\n",
    "# Convert predictions to a binary format if using sigmoid in the output layer\n",
    "predicted_labels = torch.round(predictions).squeeze()\n",
    "\n",
    "# Convert tensors to numpy arrays for easier handling\n",
    "predicted_labels = predicted_labels.numpy()\n",
    "actual_labels = test_output.numpy()  # Assuming test_output is your actual labels tensor\n",
    "\n",
    "# Combine team names and dates from the test set with predictions and actual labels\n",
    "results_df = pd.DataFrame({\n",
    "    'Date': test['date'],      \n",
    "    'Team': test['team'], \n",
    "    'Venue': test['venue'], \n",
    "    'Opponent': test['opponent'],\n",
    "    'Round': test['round'],     \n",
    "    'Predicted': predicted_labels,\n",
    "    'Actual': actual_labels\n",
    "})\n",
    "\n",
    "# Print or process the results\n",
    "print(\"Predictions vs Actual:\")\n",
    "results_df\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we have predicitons for all the matches, but there are still duplicates for each match as there is a prediciton from the home teams perspective and the away team's perpective. Now we will go onto merge the two predictions and deduce a final prediction from there. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2104,
   "metadata": {},
   "outputs": [],
   "source": [
    "class MissingDict(dict):\n",
    "    __missing__ = lambda self, key: key\n",
    "\n",
    "# Naming convention for teams on opponent column is not consistent with the team column.\n",
    "# We will use a dictionary to map the names to the same convention to allow us to merge the dataframes\n",
    "\n",
    "map_values = {\"Brighton and Hove Albion\": \"Brighton\", \"Manchester United\": \"Manchester Utd\", \n",
    "              \"Newcastle United\": \"Newcastle Utd\", \"Tottenham Hotspur\": \"Tottenham\", \n",
    "              \"West Ham United\": \"West Ham\", \"Wolverhampton Wanderers\": \"Wolves\", \n",
    "              'Nottingham Forest': 'Nott\\'ham Forest', 'Sheffield United': 'Sheffield Utd', \n",
    "              'West Bromwich Albion': 'West Brom'} \n",
    "mapping = MissingDict(**map_values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2105,
   "metadata": {},
   "outputs": [],
   "source": [
    "# merge the dataframes\n",
    "results_df[\"New_team\"] = results_df[\"Team\"].map(mapping)\n",
    "results_df[\"Opponent\"] = results_df[\"Opponent\"].map(mapping)\n",
    "merged = results_df.merge(results_df, left_on=[\"Date\", \"New_team\"], right_on=[\"Date\", \"Opponent\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2106,
   "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>Date</th>\n",
       "      <th>Team_x</th>\n",
       "      <th>Venue_x</th>\n",
       "      <th>Opponent_x</th>\n",
       "      <th>Round_x</th>\n",
       "      <th>Predicted_x</th>\n",
       "      <th>Actual_x</th>\n",
       "      <th>New_team_x</th>\n",
       "      <th>Team_y</th>\n",
       "      <th>Venue_y</th>\n",
       "      <th>Opponent_y</th>\n",
       "      <th>Round_y</th>\n",
       "      <th>Predicted_y</th>\n",
       "      <th>Actual_y</th>\n",
       "      <th>New_team_y</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2023-09-24</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Home</td>\n",
       "      <td>Tottenham</td>\n",
       "      <td>Matchweek 6</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Tottenham Hotspur</td>\n",
       "      <td>Away</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Matchweek 6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Tottenham</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2023-09-30</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Away</td>\n",
       "      <td>Bournemouth</td>\n",
       "      <td>Matchweek 7</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Bournemouth</td>\n",
       "      <td>Home</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Matchweek 7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Bournemouth</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2023-10-08</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Home</td>\n",
       "      <td>Manchester City</td>\n",
       "      <td>Matchweek 8</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Manchester City</td>\n",
       "      <td>Away</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Matchweek 8</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Manchester City</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2023-12-09</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Away</td>\n",
       "      <td>Aston Villa</td>\n",
       "      <td>Matchweek 16</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Aston Villa</td>\n",
       "      <td>Home</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Matchweek 16</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Aston Villa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2023-12-17</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Home</td>\n",
       "      <td>Brighton</td>\n",
       "      <td>Matchweek 17</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Brighton and Hove Albion</td>\n",
       "      <td>Away</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Matchweek 17</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Brighton</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>431</th>\n",
       "      <td>2023-12-17</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Away</td>\n",
       "      <td>West Ham</td>\n",
       "      <td>Matchweek 17</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>West Ham United</td>\n",
       "      <td>Home</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>Matchweek 17</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>West Ham</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>432</th>\n",
       "      <td>2023-12-24</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Home</td>\n",
       "      <td>Chelsea</td>\n",
       "      <td>Matchweek 18</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>Chelsea</td>\n",
       "      <td>Away</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>Matchweek 18</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Chelsea</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>433</th>\n",
       "      <td>2023-12-27</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Away</td>\n",
       "      <td>Brentford</td>\n",
       "      <td>Matchweek 19</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>Brentford</td>\n",
       "      <td>Home</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>Matchweek 19</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Brentford</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>434</th>\n",
       "      <td>2023-12-30</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Home</td>\n",
       "      <td>Everton</td>\n",
       "      <td>Matchweek 20</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>Everton</td>\n",
       "      <td>Away</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>Matchweek 20</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Everton</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>435</th>\n",
       "      <td>2024-02-25</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Home</td>\n",
       "      <td>Sheffield Utd</td>\n",
       "      <td>Matchweek 26</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>Sheffield United</td>\n",
       "      <td>Away</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>Matchweek 26</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Sheffield Utd</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>436 rows × 15 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          Date                   Team_x Venue_x       Opponent_x  \\\n",
       "0   2023-09-24                  Arsenal    Home        Tottenham   \n",
       "1   2023-09-30                  Arsenal    Away      Bournemouth   \n",
       "2   2023-10-08                  Arsenal    Home  Manchester City   \n",
       "3   2023-12-09                  Arsenal    Away      Aston Villa   \n",
       "4   2023-12-17                  Arsenal    Home         Brighton   \n",
       "..         ...                      ...     ...              ...   \n",
       "431 2023-12-17  Wolverhampton Wanderers    Away         West Ham   \n",
       "432 2023-12-24  Wolverhampton Wanderers    Home          Chelsea   \n",
       "433 2023-12-27  Wolverhampton Wanderers    Away        Brentford   \n",
       "434 2023-12-30  Wolverhampton Wanderers    Home          Everton   \n",
       "435 2024-02-25  Wolverhampton Wanderers    Home    Sheffield Utd   \n",
       "\n",
       "          Round_x  Predicted_x  Actual_x New_team_x                    Team_y  \\\n",
       "0     Matchweek 6          1.0       0.0    Arsenal         Tottenham Hotspur   \n",
       "1     Matchweek 7          1.0       1.0    Arsenal               Bournemouth   \n",
       "2     Matchweek 8          1.0       1.0    Arsenal           Manchester City   \n",
       "3    Matchweek 16          0.0       0.0    Arsenal               Aston Villa   \n",
       "4    Matchweek 17          1.0       1.0    Arsenal  Brighton and Hove Albion   \n",
       "..            ...          ...       ...        ...                       ...   \n",
       "431  Matchweek 17          0.0       0.0     Wolves           West Ham United   \n",
       "432  Matchweek 18          0.0       1.0     Wolves                   Chelsea   \n",
       "433  Matchweek 19          0.0       1.0     Wolves                 Brentford   \n",
       "434  Matchweek 20          0.0       1.0     Wolves                   Everton   \n",
       "435  Matchweek 26          0.0       1.0     Wolves          Sheffield United   \n",
       "\n",
       "    Venue_y Opponent_y       Round_y  Predicted_y  Actual_y       New_team_y  \n",
       "0      Away    Arsenal   Matchweek 6          0.0       0.0        Tottenham  \n",
       "1      Home    Arsenal   Matchweek 7          0.0       0.0      Bournemouth  \n",
       "2      Away    Arsenal   Matchweek 8          1.0       0.0  Manchester City  \n",
       "3      Home    Arsenal  Matchweek 16          0.0       1.0      Aston Villa  \n",
       "4      Away    Arsenal  Matchweek 17          1.0       0.0         Brighton  \n",
       "..      ...        ...           ...          ...       ...              ...  \n",
       "431    Home     Wolves  Matchweek 17          1.0       1.0         West Ham  \n",
       "432    Away     Wolves  Matchweek 18          0.0       0.0          Chelsea  \n",
       "433    Home     Wolves  Matchweek 19          0.0       0.0        Brentford  \n",
       "434    Away     Wolves  Matchweek 20          0.0       0.0          Everton  \n",
       "435    Away     Wolves  Matchweek 26          0.0       0.0    Sheffield Utd  \n",
       "\n",
       "[436 rows x 15 columns]"
      ]
     },
     "execution_count": 2106,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "merged"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2107,
   "metadata": {},
   "outputs": [],
   "source": [
    "#  final prediction logic getting net result from the two points of view\n",
    "def final_prediction(row):\n",
    "    if row['Predicted_x'] == 1 and row['Predicted_y'] == 0:\n",
    "        # Team wins, Opponent loses\n",
    "        final_pred = \"W\"\n",
    "    elif row['Predicted_x'] == 0 and row['Predicted_y'] == 1:\n",
    "        # Team loses, Opponent wins\n",
    "        final_pred = \"L\"\n",
    "    else:\n",
    "        # Both predicted to have same result, contradictory\n",
    "        final_pred = \"D\"\n",
    "    return final_pred\n",
    "  \n",
    "merged['Final_prediction'] = merged.apply(final_prediction, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2108,
   "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>Date</th>\n",
       "      <th>Team_x</th>\n",
       "      <th>Venue_x</th>\n",
       "      <th>Opponent_x</th>\n",
       "      <th>Round_x</th>\n",
       "      <th>Predicted_x</th>\n",
       "      <th>Actual_x</th>\n",
       "      <th>New_team_x</th>\n",
       "      <th>Team_y</th>\n",
       "      <th>Venue_y</th>\n",
       "      <th>Opponent_y</th>\n",
       "      <th>Round_y</th>\n",
       "      <th>Predicted_y</th>\n",
       "      <th>Actual_y</th>\n",
       "      <th>New_team_y</th>\n",
       "      <th>Final_prediction</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2023-09-24</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Home</td>\n",
       "      <td>Tottenham</td>\n",
       "      <td>Matchweek 6</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Tottenham Hotspur</td>\n",
       "      <td>Away</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Matchweek 6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Tottenham</td>\n",
       "      <td>W</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2023-09-30</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Away</td>\n",
       "      <td>Bournemouth</td>\n",
       "      <td>Matchweek 7</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Bournemouth</td>\n",
       "      <td>Home</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Matchweek 7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Bournemouth</td>\n",
       "      <td>W</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2023-10-08</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Home</td>\n",
       "      <td>Manchester City</td>\n",
       "      <td>Matchweek 8</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Manchester City</td>\n",
       "      <td>Away</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Matchweek 8</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Manchester City</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2023-12-09</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Away</td>\n",
       "      <td>Aston Villa</td>\n",
       "      <td>Matchweek 16</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Aston Villa</td>\n",
       "      <td>Home</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Matchweek 16</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Aston Villa</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2023-12-17</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Home</td>\n",
       "      <td>Brighton</td>\n",
       "      <td>Matchweek 17</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Brighton and Hove Albion</td>\n",
       "      <td>Away</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Matchweek 17</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Brighton</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>431</th>\n",
       "      <td>2023-12-17</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Away</td>\n",
       "      <td>West Ham</td>\n",
       "      <td>Matchweek 17</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>West Ham United</td>\n",
       "      <td>Home</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>Matchweek 17</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>West Ham</td>\n",
       "      <td>L</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>432</th>\n",
       "      <td>2023-12-24</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Home</td>\n",
       "      <td>Chelsea</td>\n",
       "      <td>Matchweek 18</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>Chelsea</td>\n",
       "      <td>Away</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>Matchweek 18</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Chelsea</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>433</th>\n",
       "      <td>2023-12-27</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Away</td>\n",
       "      <td>Brentford</td>\n",
       "      <td>Matchweek 19</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>Brentford</td>\n",
       "      <td>Home</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>Matchweek 19</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Brentford</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>434</th>\n",
       "      <td>2023-12-30</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Home</td>\n",
       "      <td>Everton</td>\n",
       "      <td>Matchweek 20</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>Everton</td>\n",
       "      <td>Away</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>Matchweek 20</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Everton</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>435</th>\n",
       "      <td>2024-02-25</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Home</td>\n",
       "      <td>Sheffield Utd</td>\n",
       "      <td>Matchweek 26</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>Sheffield United</td>\n",
       "      <td>Away</td>\n",
       "      <td>Wolves</td>\n",
       "      <td>Matchweek 26</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Sheffield Utd</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>436 rows × 16 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          Date                   Team_x Venue_x       Opponent_x  \\\n",
       "0   2023-09-24                  Arsenal    Home        Tottenham   \n",
       "1   2023-09-30                  Arsenal    Away      Bournemouth   \n",
       "2   2023-10-08                  Arsenal    Home  Manchester City   \n",
       "3   2023-12-09                  Arsenal    Away      Aston Villa   \n",
       "4   2023-12-17                  Arsenal    Home         Brighton   \n",
       "..         ...                      ...     ...              ...   \n",
       "431 2023-12-17  Wolverhampton Wanderers    Away         West Ham   \n",
       "432 2023-12-24  Wolverhampton Wanderers    Home          Chelsea   \n",
       "433 2023-12-27  Wolverhampton Wanderers    Away        Brentford   \n",
       "434 2023-12-30  Wolverhampton Wanderers    Home          Everton   \n",
       "435 2024-02-25  Wolverhampton Wanderers    Home    Sheffield Utd   \n",
       "\n",
       "          Round_x  Predicted_x  Actual_x New_team_x                    Team_y  \\\n",
       "0     Matchweek 6          1.0       0.0    Arsenal         Tottenham Hotspur   \n",
       "1     Matchweek 7          1.0       1.0    Arsenal               Bournemouth   \n",
       "2     Matchweek 8          1.0       1.0    Arsenal           Manchester City   \n",
       "3    Matchweek 16          0.0       0.0    Arsenal               Aston Villa   \n",
       "4    Matchweek 17          1.0       1.0    Arsenal  Brighton and Hove Albion   \n",
       "..            ...          ...       ...        ...                       ...   \n",
       "431  Matchweek 17          0.0       0.0     Wolves           West Ham United   \n",
       "432  Matchweek 18          0.0       1.0     Wolves                   Chelsea   \n",
       "433  Matchweek 19          0.0       1.0     Wolves                 Brentford   \n",
       "434  Matchweek 20          0.0       1.0     Wolves                   Everton   \n",
       "435  Matchweek 26          0.0       1.0     Wolves          Sheffield United   \n",
       "\n",
       "    Venue_y Opponent_y       Round_y  Predicted_y  Actual_y       New_team_y  \\\n",
       "0      Away    Arsenal   Matchweek 6          0.0       0.0        Tottenham   \n",
       "1      Home    Arsenal   Matchweek 7          0.0       0.0      Bournemouth   \n",
       "2      Away    Arsenal   Matchweek 8          1.0       0.0  Manchester City   \n",
       "3      Home    Arsenal  Matchweek 16          0.0       1.0      Aston Villa   \n",
       "4      Away    Arsenal  Matchweek 17          1.0       0.0         Brighton   \n",
       "..      ...        ...           ...          ...       ...              ...   \n",
       "431    Home     Wolves  Matchweek 17          1.0       1.0         West Ham   \n",
       "432    Away     Wolves  Matchweek 18          0.0       0.0          Chelsea   \n",
       "433    Home     Wolves  Matchweek 19          0.0       0.0        Brentford   \n",
       "434    Away     Wolves  Matchweek 20          0.0       0.0          Everton   \n",
       "435    Away     Wolves  Matchweek 26          0.0       0.0    Sheffield Utd   \n",
       "\n",
       "    Final_prediction  \n",
       "0                  W  \n",
       "1                  W  \n",
       "2                  D  \n",
       "3                  D  \n",
       "4                  D  \n",
       "..               ...  \n",
       "431                L  \n",
       "432                D  \n",
       "433                D  \n",
       "434                D  \n",
       "435                D  \n",
       "\n",
       "[436 rows x 16 columns]"
      ]
     },
     "execution_count": 2108,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "merged"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2109,
   "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>Date</th>\n",
       "      <th>Round</th>\n",
       "      <th>Venue</th>\n",
       "      <th>Team</th>\n",
       "      <th>Opponent</th>\n",
       "      <th>Prediction</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2023-09-24</td>\n",
       "      <td>Matchweek 6</td>\n",
       "      <td>Home</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Tottenham</td>\n",
       "      <td>W</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2023-09-30</td>\n",
       "      <td>Matchweek 7</td>\n",
       "      <td>Away</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Bournemouth</td>\n",
       "      <td>W</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2023-10-08</td>\n",
       "      <td>Matchweek 8</td>\n",
       "      <td>Home</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Manchester City</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2023-12-09</td>\n",
       "      <td>Matchweek 16</td>\n",
       "      <td>Away</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Aston Villa</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2023-12-17</td>\n",
       "      <td>Matchweek 17</td>\n",
       "      <td>Home</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Brighton</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>431</th>\n",
       "      <td>2023-12-17</td>\n",
       "      <td>Matchweek 17</td>\n",
       "      <td>Away</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>West Ham</td>\n",
       "      <td>L</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>432</th>\n",
       "      <td>2023-12-24</td>\n",
       "      <td>Matchweek 18</td>\n",
       "      <td>Home</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Chelsea</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>433</th>\n",
       "      <td>2023-12-27</td>\n",
       "      <td>Matchweek 19</td>\n",
       "      <td>Away</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Brentford</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>434</th>\n",
       "      <td>2023-12-30</td>\n",
       "      <td>Matchweek 20</td>\n",
       "      <td>Home</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Everton</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>435</th>\n",
       "      <td>2024-02-25</td>\n",
       "      <td>Matchweek 26</td>\n",
       "      <td>Home</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Sheffield Utd</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>436 rows × 6 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          Date         Round Venue                     Team         Opponent  \\\n",
       "0   2023-09-24   Matchweek 6  Home                  Arsenal        Tottenham   \n",
       "1   2023-09-30   Matchweek 7  Away                  Arsenal      Bournemouth   \n",
       "2   2023-10-08   Matchweek 8  Home                  Arsenal  Manchester City   \n",
       "3   2023-12-09  Matchweek 16  Away                  Arsenal      Aston Villa   \n",
       "4   2023-12-17  Matchweek 17  Home                  Arsenal         Brighton   \n",
       "..         ...           ...   ...                      ...              ...   \n",
       "431 2023-12-17  Matchweek 17  Away  Wolverhampton Wanderers         West Ham   \n",
       "432 2023-12-24  Matchweek 18  Home  Wolverhampton Wanderers          Chelsea   \n",
       "433 2023-12-27  Matchweek 19  Away  Wolverhampton Wanderers        Brentford   \n",
       "434 2023-12-30  Matchweek 20  Home  Wolverhampton Wanderers          Everton   \n",
       "435 2024-02-25  Matchweek 26  Home  Wolverhampton Wanderers    Sheffield Utd   \n",
       "\n",
       "    Prediction  \n",
       "0            W  \n",
       "1            W  \n",
       "2            D  \n",
       "3            D  \n",
       "4            D  \n",
       "..         ...  \n",
       "431          L  \n",
       "432          D  \n",
       "433          D  \n",
       "434          D  \n",
       "435          D  \n",
       "\n",
       "[436 rows x 6 columns]"
      ]
     },
     "execution_count": 2109,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# now simplify the dataframe back to single team perspective and maintain relevant columns\n",
    "from datetime import datetime\n",
    "\n",
    "prediction = merged[[\"Date\", \"Round_x\" , \"Venue_x\" ,\"Team_x\", \"Opponent_x\", \"Final_prediction\"]]\n",
    "prediction = prediction.rename(columns={\"Round_x\": \"Round\", \"Venue_x\": \"Venue\", \"Team_x\": \"Team\", \"Opponent_x\": \"Opponent\", \"Final_prediction\": \"Prediction\"})\n",
    "prediction.sort_values(by='Date', ascending=False)\n",
    "prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2110,
   "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>Date</th>\n",
       "      <th>Round</th>\n",
       "      <th>Team</th>\n",
       "      <th>Opponent</th>\n",
       "      <th>Prediction</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2023-09-24</td>\n",
       "      <td>Matchweek 6</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Tottenham</td>\n",
       "      <td>W</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2023-10-08</td>\n",
       "      <td>Matchweek 8</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Manchester City</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2023-12-17</td>\n",
       "      <td>Matchweek 17</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Brighton</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>2023-12-28</td>\n",
       "      <td>Matchweek 19</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>West Ham</td>\n",
       "      <td>W</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>2024-01-20</td>\n",
       "      <td>Matchweek 21</td>\n",
       "      <td>Arsenal</td>\n",
       "      <td>Crystal Palace</td>\n",
       "      <td>W</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>429</th>\n",
       "      <td>2023-12-05</td>\n",
       "      <td>Matchweek 15</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Burnley</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>430</th>\n",
       "      <td>2023-12-09</td>\n",
       "      <td>Matchweek 16</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Nott'ham Forest</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>432</th>\n",
       "      <td>2023-12-24</td>\n",
       "      <td>Matchweek 18</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Chelsea</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>434</th>\n",
       "      <td>2023-12-30</td>\n",
       "      <td>Matchweek 20</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Everton</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>435</th>\n",
       "      <td>2024-02-25</td>\n",
       "      <td>Matchweek 26</td>\n",
       "      <td>Wolverhampton Wanderers</td>\n",
       "      <td>Sheffield Utd</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>218 rows × 5 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          Date         Round                     Team         Opponent  \\\n",
       "0   2023-09-24   Matchweek 6                  Arsenal        Tottenham   \n",
       "2   2023-10-08   Matchweek 8                  Arsenal  Manchester City   \n",
       "4   2023-12-17  Matchweek 17                  Arsenal         Brighton   \n",
       "6   2023-12-28  Matchweek 19                  Arsenal         West Ham   \n",
       "8   2024-01-20  Matchweek 21                  Arsenal   Crystal Palace   \n",
       "..         ...           ...                      ...              ...   \n",
       "429 2023-12-05  Matchweek 15  Wolverhampton Wanderers          Burnley   \n",
       "430 2023-12-09  Matchweek 16  Wolverhampton Wanderers  Nott'ham Forest   \n",
       "432 2023-12-24  Matchweek 18  Wolverhampton Wanderers          Chelsea   \n",
       "434 2023-12-30  Matchweek 20  Wolverhampton Wanderers          Everton   \n",
       "435 2024-02-25  Matchweek 26  Wolverhampton Wanderers    Sheffield Utd   \n",
       "\n",
       "    Prediction  \n",
       "0            W  \n",
       "2            D  \n",
       "4            D  \n",
       "6            W  \n",
       "8            W  \n",
       "..         ...  \n",
       "429          D  \n",
       "430          D  \n",
       "432          D  \n",
       "434          D  \n",
       "435          D  \n",
       "\n",
       "[218 rows x 5 columns]"
      ]
     },
     "execution_count": 2110,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# remove redundancy by only keeping home point of view\n",
    "prediction = prediction[prediction['Venue'] != 'Away']\n",
    "prediction = prediction.drop(columns=['Venue'])\n",
    "prediction"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='13-conclusiion'></a> \n",
    "### 13. Conclusion"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This tutorial has successfully walked through the process of building a machine learning model to predict Premier League match outcomes. Starting with data collection and preprocessing, we highlighted the importance of contextual data by integrating both team and opponent statistics. Feature selection and normalisation streamlined our model, making it efficient and effective. Our exploration into neural network architecture and the subsequent hyperparameter tuning showcased the intricacies of model optimisation. We crafted visualisations to inform us on model performance and aid in further fine tuning. The final model's evaluation illustrated the potency of machine learning in sports analytics, providing valuable insights into the practical aspects of predictive modeling in football."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='14-differences-and-similarities'></a> \n",
    "### 14. Enhancements From Source Tutorials\n",
    "\n",
    "**Similarities:**\n",
    "\n",
    "- The Neural Network class structure was adapted from Andre Luiz's implementation[2].\n",
    "\n",
    "- The approach to presenting data in a reader-friendly format in Section 12 was inspired by Dataquest's implementation[3].\n",
    "\n",
    "- Transformation of string fields like venue, team names, and results into categorical values was guided by Dataquest's practices[3].\n",
    "\n",
    "- The initial calculations of rolling averages for team statistics were based on techniques presented by Dataquest[3].\n",
    "\n",
    "**Differences:**\n",
    "\n",
    "- Enhancing the rolling stats method from Dataquest[3], I integrated the opposing team's match data into the same dataset to provide a comprehensive view of both teams for more informed predictions.\n",
    "\n",
    "- Unlike the referenced sources[2][3], feature selection was implemented using a filter method to concentrate our model's learning on the most impactful features.\n",
    "\n",
    "- Data normalization was methodically applied to standardize the range of our feature sets.\n",
    "\n",
    "- Hyperparameter tuning was carried out, an optimization step not addressed by sources[2][3], to fine-tune our model's performance.\n",
    "\n",
    "- Rigorous performance evaluations, including accuracy and precision metrics, were conducted before and after training, offering valuable insights into the benefits of our modeling approach.\n",
    "\n",
    "- We integrated confusion matrix visualizations, enhancing the interpretability of our model's predictions, a step beyond the scope of sources[2][3].\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='15-references'></a> \n",
    "### 15. References"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "\n",
    "1. Andre Luiz, (Date of Publication). 'PyTorch Neural Networks to Predict Matches Results in Soccer Championships - Part I', Medium. Available at: https://medium.com/@andreluiz_4916/pytorch-neural-networks-to-predict-matches-results-in-soccer-championships-part-i-a6d0eefeca51 (Accessed: 2 March 2024).\n",
    "\n",
    "2. Andre Luiz, (Date of Publication). 'PyTorch Neural Networks to Predict Matches Results in Soccer Championships - Part II', Medium. Available at: https://medium.com/@andreluiz_4916/pytorch-neural-networks-to-predict-matches-results-in-soccer-championships-part-ii-3d02b2ddd538 (Accessed: 8 March 2024).\n",
    "\n",
    "3. Dataquest, (Date of Publication). 'Portfolio Project: Predicting EPL Football Match Winners Using Machine Learning', Dataquest. Available at: https://app.dataquest.io/m/99992/portfolio-project%3A-predicting-epl-football-match-winners-using-machine-learning/1/project-overview (Accessed: 28 February 2024).\n",
    "\n",
    "\n"
   ]
  }
 ],
 "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": "3.11.6"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}