baseball_linear_regression.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import libraries\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn import linear_model\n",
    "from scipy import stats\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Dataset downloaded from https://www.retrosheet.org/gamelogs/index.html. Data dictionary can be found at http://www.retrosheet.org/gamelogs/glfields.txt."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Read games logs from the 2015 season into a dataframe\n",
    "input_df = pd.read_table(\"GL2015.TXT\", sep=\",\", header=None)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Data Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>Visiting Team</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>Home Team</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>Runs Visitor</th>\n",
       "      <th>...</th>\n",
       "      <th>151</th>\n",
       "      <th>152</th>\n",
       "      <th>153</th>\n",
       "      <th>154</th>\n",
       "      <th>155</th>\n",
       "      <th>156</th>\n",
       "      <th>157</th>\n",
       "      <th>158</th>\n",
       "      <th>159</th>\n",
       "      <th>160</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>20150405</td>\n",
       "      <td>0</td>\n",
       "      <td>Sun</td>\n",
       "      <td>SLN</td>\n",
       "      <td>NL</td>\n",
       "      <td>1</td>\n",
       "      <td>CHN</td>\n",
       "      <td>NL</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>...</td>\n",
       "      <td>David Ross</td>\n",
       "      <td>2</td>\n",
       "      <td>lestj001</td>\n",
       "      <td>Jon Lester</td>\n",
       "      <td>1</td>\n",
       "      <td>lastt001</td>\n",
       "      <td>Tommy La Stella</td>\n",
       "      <td>4</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Y</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>20150406</td>\n",
       "      <td>0</td>\n",
       "      <td>Mon</td>\n",
       "      <td>MIN</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>DET</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>Nick Castellanos</td>\n",
       "      <td>5</td>\n",
       "      <td>avila001</td>\n",
       "      <td>Alex Avila</td>\n",
       "      <td>2</td>\n",
       "      <td>iglej001</td>\n",
       "      <td>Jose Iglesias</td>\n",
       "      <td>6</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Y</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>20150406</td>\n",
       "      <td>0</td>\n",
       "      <td>Mon</td>\n",
       "      <td>CLE</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>HOU</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>Jed Lowrie</td>\n",
       "      <td>6</td>\n",
       "      <td>rasmc001</td>\n",
       "      <td>Colby Rasmus</td>\n",
       "      <td>7</td>\n",
       "      <td>marij002</td>\n",
       "      <td>Jake Marisnick</td>\n",
       "      <td>8</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Y</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>20150406</td>\n",
       "      <td>0</td>\n",
       "      <td>Mon</td>\n",
       "      <td>CHA</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>KCA</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>Alex Rios</td>\n",
       "      <td>9</td>\n",
       "      <td>peres002</td>\n",
       "      <td>Salvador Perez</td>\n",
       "      <td>2</td>\n",
       "      <td>infao001</td>\n",
       "      <td>Omar Infante</td>\n",
       "      <td>4</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Y</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>20150406</td>\n",
       "      <td>0</td>\n",
       "      <td>Mon</td>\n",
       "      <td>TOR</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>NYA</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>...</td>\n",
       "      <td>Alex Rodriguez</td>\n",
       "      <td>10</td>\n",
       "      <td>drews001</td>\n",
       "      <td>Stephen Drew</td>\n",
       "      <td>4</td>\n",
       "      <td>gregd001</td>\n",
       "      <td>Didi Gregorius</td>\n",
       "      <td>6</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Y</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>20150406</td>\n",
       "      <td>0</td>\n",
       "      <td>Mon</td>\n",
       "      <td>TEX</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>OAK</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>Stephen Vogt</td>\n",
       "      <td>2</td>\n",
       "      <td>semim001</td>\n",
       "      <td>Marcus Semien</td>\n",
       "      <td>6</td>\n",
       "      <td>sogae001</td>\n",
       "      <td>Eric Sogard</td>\n",
       "      <td>4</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Y</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>20150406</td>\n",
       "      <td>0</td>\n",
       "      <td>Mon</td>\n",
       "      <td>ANA</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>SEA</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>Mike Zunino</td>\n",
       "      <td>2</td>\n",
       "      <td>ackld001</td>\n",
       "      <td>Dustin Ackley</td>\n",
       "      <td>7</td>\n",
       "      <td>millb002</td>\n",
       "      <td>Brad Miller</td>\n",
       "      <td>6</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Y</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>20150406</td>\n",
       "      <td>0</td>\n",
       "      <td>Mon</td>\n",
       "      <td>BAL</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>TBA</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>...</td>\n",
       "      <td>Rene Rivera</td>\n",
       "      <td>2</td>\n",
       "      <td>forsl001</td>\n",
       "      <td>Logan Forsythe</td>\n",
       "      <td>4</td>\n",
       "      <td>kierk001</td>\n",
       "      <td>Kevin Kiermaier</td>\n",
       "      <td>8</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Y</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>20150406</td>\n",
       "      <td>0</td>\n",
       "      <td>Mon</td>\n",
       "      <td>SFN</td>\n",
       "      <td>NL</td>\n",
       "      <td>1</td>\n",
       "      <td>ARI</td>\n",
       "      <td>NL</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>...</td>\n",
       "      <td>Tuffy Gosewisch</td>\n",
       "      <td>2</td>\n",
       "      <td>ahmen001</td>\n",
       "      <td>Nick Ahmed</td>\n",
       "      <td>6</td>\n",
       "      <td>collj001</td>\n",
       "      <td>Josh Collmenter</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Y</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>20150406</td>\n",
       "      <td>0</td>\n",
       "      <td>Mon</td>\n",
       "      <td>PIT</td>\n",
       "      <td>NL</td>\n",
       "      <td>1</td>\n",
       "      <td>CIN</td>\n",
       "      <td>NL</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>Brandon Phillips</td>\n",
       "      <td>4</td>\n",
       "      <td>cozaz001</td>\n",
       "      <td>Zack Cozart</td>\n",
       "      <td>6</td>\n",
       "      <td>cuetj001</td>\n",
       "      <td>Johnny Cueto</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Y</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>10 rows × 161 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          0  1    2 Visiting Team   4  5 Home Team   7  8  Runs Visitor ...  \\\n",
       "0  20150405  0  Sun           SLN  NL  1       CHN  NL  1             3 ...   \n",
       "1  20150406  0  Mon           MIN  AL  1       DET  AL  1             0 ...   \n",
       "2  20150406  0  Mon           CLE  AL  1       HOU  AL  1             0 ...   \n",
       "3  20150406  0  Mon           CHA  AL  1       KCA  AL  1             1 ...   \n",
       "4  20150406  0  Mon           TOR  AL  1       NYA  AL  1             6 ...   \n",
       "5  20150406  0  Mon           TEX  AL  1       OAK  AL  1             0 ...   \n",
       "6  20150406  0  Mon           ANA  AL  1       SEA  AL  1             1 ...   \n",
       "7  20150406  0  Mon           BAL  AL  1       TBA  AL  1             6 ...   \n",
       "8  20150406  0  Mon           SFN  NL  1       ARI  NL  1             5 ...   \n",
       "9  20150406  0  Mon           PIT  NL  1       CIN  NL  1             2 ...   \n",
       "\n",
       "                151  152       153             154  155       156  \\\n",
       "0        David Ross    2  lestj001      Jon Lester    1  lastt001   \n",
       "1  Nick Castellanos    5  avila001      Alex Avila    2  iglej001   \n",
       "2        Jed Lowrie    6  rasmc001    Colby Rasmus    7  marij002   \n",
       "3         Alex Rios    9  peres002  Salvador Perez    2  infao001   \n",
       "4    Alex Rodriguez   10  drews001    Stephen Drew    4  gregd001   \n",
       "5      Stephen Vogt    2  semim001   Marcus Semien    6  sogae001   \n",
       "6       Mike Zunino    2  ackld001   Dustin Ackley    7  millb002   \n",
       "7       Rene Rivera    2  forsl001  Logan Forsythe    4  kierk001   \n",
       "8   Tuffy Gosewisch    2  ahmen001      Nick Ahmed    6  collj001   \n",
       "9  Brandon Phillips    4  cozaz001     Zack Cozart    6  cuetj001   \n",
       "\n",
       "               157  158  159 160  \n",
       "0  Tommy La Stella    4  NaN   Y  \n",
       "1    Jose Iglesias    6  NaN   Y  \n",
       "2   Jake Marisnick    8  NaN   Y  \n",
       "3     Omar Infante    4  NaN   Y  \n",
       "4   Didi Gregorius    6  NaN   Y  \n",
       "5      Eric Sogard    4  NaN   Y  \n",
       "6      Brad Miller    6  NaN   Y  \n",
       "7  Kevin Kiermaier    8  NaN   Y  \n",
       "8  Josh Collmenter    1  NaN   Y  \n",
       "9     Johnny Cueto    1  NaN   Y  \n",
       "\n",
       "[10 rows x 161 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def rename_columns(input_df):\n",
    "    \"\"\" \n",
    "    Rename columns of an input dataframe for readability.\n",
    "  \n",
    "    Parameters: \n",
    "    input_df (dataframe): Input dataframe.\n",
    "  \n",
    "    Returns: \n",
    "    dataframe: Dataframe with renamed column names.\n",
    "  \n",
    "    \"\"\"\n",
    "    \n",
    "    input_df.rename(columns = {3: 'Visiting Team', 6: 'Home Team', 9: 'Runs Visitor', 10: 'Runs Home'}, inplace=True)\n",
    "    return input_df\n",
    "\n",
    "# Invoke the function to rename columns\n",
    "input_df = rename_columns(input_df)\n",
    "\n",
    "# Display first ten rows\n",
    "input_df.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>Visiting Team</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>Home Team</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>Runs Visitor</th>\n",
       "      <th>...</th>\n",
       "      <th>153</th>\n",
       "      <th>154</th>\n",
       "      <th>155</th>\n",
       "      <th>156</th>\n",
       "      <th>157</th>\n",
       "      <th>158</th>\n",
       "      <th>159</th>\n",
       "      <th>160</th>\n",
       "      <th>Home Win</th>\n",
       "      <th>Visitor Win</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>20150405</td>\n",
       "      <td>0</td>\n",
       "      <td>Sun</td>\n",
       "      <td>SLN</td>\n",
       "      <td>NL</td>\n",
       "      <td>1</td>\n",
       "      <td>CHN</td>\n",
       "      <td>NL</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>...</td>\n",
       "      <td>lestj001</td>\n",
       "      <td>Jon Lester</td>\n",
       "      <td>1</td>\n",
       "      <td>lastt001</td>\n",
       "      <td>Tommy La Stella</td>\n",
       "      <td>4</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Y</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>20150406</td>\n",
       "      <td>0</td>\n",
       "      <td>Mon</td>\n",
       "      <td>MIN</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>DET</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>avila001</td>\n",
       "      <td>Alex Avila</td>\n",
       "      <td>2</td>\n",
       "      <td>iglej001</td>\n",
       "      <td>Jose Iglesias</td>\n",
       "      <td>6</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Y</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>20150406</td>\n",
       "      <td>0</td>\n",
       "      <td>Mon</td>\n",
       "      <td>CLE</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>HOU</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>rasmc001</td>\n",
       "      <td>Colby Rasmus</td>\n",
       "      <td>7</td>\n",
       "      <td>marij002</td>\n",
       "      <td>Jake Marisnick</td>\n",
       "      <td>8</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Y</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>20150406</td>\n",
       "      <td>0</td>\n",
       "      <td>Mon</td>\n",
       "      <td>CHA</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>KCA</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>peres002</td>\n",
       "      <td>Salvador Perez</td>\n",
       "      <td>2</td>\n",
       "      <td>infao001</td>\n",
       "      <td>Omar Infante</td>\n",
       "      <td>4</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Y</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>20150406</td>\n",
       "      <td>0</td>\n",
       "      <td>Mon</td>\n",
       "      <td>TOR</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>NYA</td>\n",
       "      <td>AL</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>...</td>\n",
       "      <td>drews001</td>\n",
       "      <td>Stephen Drew</td>\n",
       "      <td>4</td>\n",
       "      <td>gregd001</td>\n",
       "      <td>Didi Gregorius</td>\n",
       "      <td>6</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Y</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 163 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          0  1    2 Visiting Team   4  5 Home Team   7  8  Runs Visitor  \\\n",
       "0  20150405  0  Sun           SLN  NL  1       CHN  NL  1             3   \n",
       "1  20150406  0  Mon           MIN  AL  1       DET  AL  1             0   \n",
       "2  20150406  0  Mon           CLE  AL  1       HOU  AL  1             0   \n",
       "3  20150406  0  Mon           CHA  AL  1       KCA  AL  1             1   \n",
       "4  20150406  0  Mon           TOR  AL  1       NYA  AL  1             6   \n",
       "\n",
       "      ...           153             154 155       156              157 158  \\\n",
       "0     ...      lestj001      Jon Lester   1  lastt001  Tommy La Stella   4   \n",
       "1     ...      avila001      Alex Avila   2  iglej001    Jose Iglesias   6   \n",
       "2     ...      rasmc001    Colby Rasmus   7  marij002   Jake Marisnick   8   \n",
       "3     ...      peres002  Salvador Perez   2  infao001     Omar Infante   4   \n",
       "4     ...      drews001    Stephen Drew   4  gregd001   Didi Gregorius   6   \n",
       "\n",
       "   159  160  Home Win Visitor Win  \n",
       "0  NaN    Y     False        True  \n",
       "1  NaN    Y      True       False  \n",
       "2  NaN    Y      True       False  \n",
       "3  NaN    Y      True       False  \n",
       "4  NaN    Y     False        True  \n",
       "\n",
       "[5 rows x 163 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def add_new_columns(input_df):\n",
    "    \"\"\" \n",
    "    Add new columns to indicate which team won the game. \n",
    "  \n",
    "    Parameters: \n",
    "    input_df (dataframe): Input dataframe.\n",
    "  \n",
    "    Returns: \n",
    "    dataframe: Dataframe with newly-added column.\n",
    "  \n",
    "    \"\"\"\n",
    "    \n",
    "    input_df['Home Win'] = (input_df['Runs Home'] > input_df['Runs Visitor'])\n",
    "    input_df['Visitor Win'] = (input_df['Runs Visitor'] > input_df['Runs Home'])\n",
    "    return input_df\n",
    "\n",
    "# Invoke method to add new columns\n",
    "input_df = add_new_columns(input_df)\n",
    "\n",
    "# Display\n",
    "input_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Team</th>\n",
       "      <th>Runs by Visitor</th>\n",
       "      <th>Runs at Home</th>\n",
       "      <th>Wins at Home</th>\n",
       "      <th>Home Games</th>\n",
       "      <th>RD at Home</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ANA</td>\n",
       "      <td>298.0</td>\n",
       "      <td>320.0</td>\n",
       "      <td>49.0</td>\n",
       "      <td>81</td>\n",
       "      <td>22.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ARI</td>\n",
       "      <td>372.0</td>\n",
       "      <td>366.0</td>\n",
       "      <td>39.0</td>\n",
       "      <td>81</td>\n",
       "      <td>-6.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ATL</td>\n",
       "      <td>359.0</td>\n",
       "      <td>286.0</td>\n",
       "      <td>42.0</td>\n",
       "      <td>81</td>\n",
       "      <td>-73.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>BAL</td>\n",
       "      <td>353.0</td>\n",
       "      <td>408.0</td>\n",
       "      <td>49.0</td>\n",
       "      <td>81</td>\n",
       "      <td>55.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>BOS</td>\n",
       "      <td>383.0</td>\n",
       "      <td>433.0</td>\n",
       "      <td>43.0</td>\n",
       "      <td>81</td>\n",
       "      <td>50.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>CHA</td>\n",
       "      <td>338.0</td>\n",
       "      <td>290.0</td>\n",
       "      <td>40.0</td>\n",
       "      <td>81</td>\n",
       "      <td>-48.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>CHN</td>\n",
       "      <td>306.0</td>\n",
       "      <td>326.0</td>\n",
       "      <td>49.0</td>\n",
       "      <td>81</td>\n",
       "      <td>20.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>CIN</td>\n",
       "      <td>391.0</td>\n",
       "      <td>344.0</td>\n",
       "      <td>34.0</td>\n",
       "      <td>81</td>\n",
       "      <td>-47.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>CLE</td>\n",
       "      <td>359.0</td>\n",
       "      <td>367.0</td>\n",
       "      <td>39.0</td>\n",
       "      <td>80</td>\n",
       "      <td>8.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>COL</td>\n",
       "      <td>483.0</td>\n",
       "      <td>449.0</td>\n",
       "      <td>36.0</td>\n",
       "      <td>81</td>\n",
       "      <td>-34.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Team  Runs by Visitor  Runs at Home  Wins at Home  Home Games  RD at Home\n",
       "0  ANA            298.0         320.0          49.0          81        22.0\n",
       "1  ARI            372.0         366.0          39.0          81        -6.0\n",
       "2  ATL            359.0         286.0          42.0          81       -73.0\n",
       "3  BAL            353.0         408.0          49.0          81        55.0\n",
       "4  BOS            383.0         433.0          43.0          81        50.0\n",
       "5  CHA            338.0         290.0          40.0          81       -48.0\n",
       "6  CHN            306.0         326.0          49.0          81        20.0\n",
       "7  CIN            391.0         344.0          34.0          81       -47.0\n",
       "8  CLE            359.0         367.0          39.0          80         8.0\n",
       "9  COL            483.0         449.0          36.0          81       -34.0"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def proc_home_team_data(input_df):\n",
    "    \"\"\" \n",
    "    Group the data by home team and compute relevant statistics.\n",
    "  \n",
    "    Parameters: \n",
    "    input_df (dataframe): Input dataframe.\n",
    "  \n",
    "    Returns: \n",
    "    dataframe: Dataframe with stats grouped by home team. \n",
    "  \n",
    "    \"\"\"\n",
    "    \n",
    "    # Group by home team\n",
    "    home_group = input_df.groupby(input_df['Home Team'])\n",
    "\n",
    "    # Compute stats: Number of games, runs scored, runs conceded, wins, run differential\n",
    "    home_df = home_group[['Runs Visitor', 'Runs Home', 'Home Win']].apply(sum)\n",
    "    home_df['Home Games'] = home_group['Home Win'].count()\n",
    "    home_df.rename(columns = {'Runs Visitor': 'Runs by Visitor', 'Runs Home': 'Runs at Home', 'Home Win': 'Wins at Home'}, inplace=True)\n",
    "    home_df['RD at Home'] = home_df['Runs at Home'] - home_df['Runs by Visitor']\n",
    "    home_df.index.rename('Team', inplace=True)\n",
    "    home_df.reset_index(inplace=True)\n",
    "\n",
    "    return home_df\n",
    "\n",
    "# Invoke method to group data by home team and compute statistics\n",
    "home_df = proc_home_team_data(input_df)\n",
    "\n",
    "# Display first ten rows\n",
    "home_df.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Team</th>\n",
       "      <th>Runs as Visitor</th>\n",
       "      <th>Runs by Home</th>\n",
       "      <th>Wins as Visitor</th>\n",
       "      <th>Road Games</th>\n",
       "      <th>RD as Visitor</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ANA</td>\n",
       "      <td>341.0</td>\n",
       "      <td>377.0</td>\n",
       "      <td>36.0</td>\n",
       "      <td>81</td>\n",
       "      <td>-36.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ARI</td>\n",
       "      <td>354.0</td>\n",
       "      <td>341.0</td>\n",
       "      <td>40.0</td>\n",
       "      <td>81</td>\n",
       "      <td>13.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ATL</td>\n",
       "      <td>287.0</td>\n",
       "      <td>401.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>81</td>\n",
       "      <td>-114.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>BAL</td>\n",
       "      <td>305.0</td>\n",
       "      <td>340.0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>81</td>\n",
       "      <td>-35.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>BOS</td>\n",
       "      <td>315.0</td>\n",
       "      <td>370.0</td>\n",
       "      <td>35.0</td>\n",
       "      <td>81</td>\n",
       "      <td>-55.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>CHA</td>\n",
       "      <td>332.0</td>\n",
       "      <td>363.0</td>\n",
       "      <td>36.0</td>\n",
       "      <td>81</td>\n",
       "      <td>-31.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>CHN</td>\n",
       "      <td>363.0</td>\n",
       "      <td>302.0</td>\n",
       "      <td>48.0</td>\n",
       "      <td>81</td>\n",
       "      <td>61.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>CIN</td>\n",
       "      <td>296.0</td>\n",
       "      <td>363.0</td>\n",
       "      <td>30.0</td>\n",
       "      <td>81</td>\n",
       "      <td>-67.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>CLE</td>\n",
       "      <td>302.0</td>\n",
       "      <td>281.0</td>\n",
       "      <td>42.0</td>\n",
       "      <td>81</td>\n",
       "      <td>21.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>COL</td>\n",
       "      <td>288.0</td>\n",
       "      <td>361.0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>81</td>\n",
       "      <td>-73.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Team  Runs as Visitor  Runs by Home  Wins as Visitor  Road Games  \\\n",
       "0  ANA            341.0         377.0             36.0          81   \n",
       "1  ARI            354.0         341.0             40.0          81   \n",
       "2  ATL            287.0         401.0             25.0          81   \n",
       "3  BAL            305.0         340.0             32.0          81   \n",
       "4  BOS            315.0         370.0             35.0          81   \n",
       "5  CHA            332.0         363.0             36.0          81   \n",
       "6  CHN            363.0         302.0             48.0          81   \n",
       "7  CIN            296.0         363.0             30.0          81   \n",
       "8  CLE            302.0         281.0             42.0          81   \n",
       "9  COL            288.0         361.0             32.0          81   \n",
       "\n",
       "   RD as Visitor  \n",
       "0          -36.0  \n",
       "1           13.0  \n",
       "2         -114.0  \n",
       "3          -35.0  \n",
       "4          -55.0  \n",
       "5          -31.0  \n",
       "6           61.0  \n",
       "7          -67.0  \n",
       "8           21.0  \n",
       "9          -73.0  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def proc_visiting_team_data(input_df):\n",
    "    \"\"\" \n",
    "    Method to group data by visiting team and compute relevant statistics.\n",
    "  \n",
    "    Parameters: \n",
    "    input_df (dataframe): Input dataframe.\n",
    "  \n",
    "    Returns: \n",
    "    dataframe: Dataframe with stats grouped by visiting team. \n",
    "  \n",
    "    \"\"\"\n",
    "    \n",
    "    # Group by visiting team\n",
    "    visit_group = input_df.groupby(input_df['Visiting Team'])\n",
    "\n",
    "    # Compute stats: Number of games, runs scored, runs conceded, wins, run differential\n",
    "    visit_df = visit_group[['Runs Visitor', 'Runs Home', 'Visitor Win']].apply(sum)\n",
    "    visit_df['Road Games'] = visit_group['Visitor Win'].count()\n",
    "    visit_df.rename(columns = {'Runs Visitor': 'Runs as Visitor', 'Runs Home': 'Runs by Home', \n",
    "                                 'Visitor Win': 'Wins as Visitor'}, inplace=True)\n",
    "    visit_df['RD as Visitor'] = visit_df['Runs as Visitor'] - visit_df['Runs by Home']\n",
    "    visit_df.index.rename('Team', inplace=True)\n",
    "    visit_df.reset_index(inplace=True)\n",
    "\n",
    "    return visit_df\n",
    "\n",
    "# Invoke method to group data by visiting team and compute statistics\n",
    "visit_df = proc_visiting_team_data(input_df)\n",
    "\n",
    "# Display the first ten rows\n",
    "visit_df.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Team</th>\n",
       "      <th>Runs by Visitor</th>\n",
       "      <th>Runs at Home</th>\n",
       "      <th>Wins at Home</th>\n",
       "      <th>Home Games</th>\n",
       "      <th>RD at Home</th>\n",
       "      <th>Runs as Visitor</th>\n",
       "      <th>Runs by Home</th>\n",
       "      <th>Wins as Visitor</th>\n",
       "      <th>Road Games</th>\n",
       "      <th>RD as Visitor</th>\n",
       "      <th>RD</th>\n",
       "      <th>Win Pct</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ANA</td>\n",
       "      <td>298.0</td>\n",
       "      <td>320.0</td>\n",
       "      <td>49.0</td>\n",
       "      <td>81</td>\n",
       "      <td>22.0</td>\n",
       "      <td>341.0</td>\n",
       "      <td>377.0</td>\n",
       "      <td>36.0</td>\n",
       "      <td>81</td>\n",
       "      <td>-36.0</td>\n",
       "      <td>-14.0</td>\n",
       "      <td>0.524691</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ARI</td>\n",
       "      <td>372.0</td>\n",
       "      <td>366.0</td>\n",
       "      <td>39.0</td>\n",
       "      <td>81</td>\n",
       "      <td>-6.0</td>\n",
       "      <td>354.0</td>\n",
       "      <td>341.0</td>\n",
       "      <td>40.0</td>\n",
       "      <td>81</td>\n",
       "      <td>13.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>0.487654</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ATL</td>\n",
       "      <td>359.0</td>\n",
       "      <td>286.0</td>\n",
       "      <td>42.0</td>\n",
       "      <td>81</td>\n",
       "      <td>-73.0</td>\n",
       "      <td>287.0</td>\n",
       "      <td>401.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>81</td>\n",
       "      <td>-114.0</td>\n",
       "      <td>-187.0</td>\n",
       "      <td>0.413580</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>BAL</td>\n",
       "      <td>353.0</td>\n",
       "      <td>408.0</td>\n",
       "      <td>49.0</td>\n",
       "      <td>81</td>\n",
       "      <td>55.0</td>\n",
       "      <td>305.0</td>\n",
       "      <td>340.0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>81</td>\n",
       "      <td>-35.0</td>\n",
       "      <td>20.0</td>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>BOS</td>\n",
       "      <td>383.0</td>\n",
       "      <td>433.0</td>\n",
       "      <td>43.0</td>\n",
       "      <td>81</td>\n",
       "      <td>50.0</td>\n",
       "      <td>315.0</td>\n",
       "      <td>370.0</td>\n",
       "      <td>35.0</td>\n",
       "      <td>81</td>\n",
       "      <td>-55.0</td>\n",
       "      <td>-5.0</td>\n",
       "      <td>0.481481</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Team  Runs by Visitor  Runs at Home  Wins at Home  Home Games  RD at Home  \\\n",
       "0  ANA            298.0         320.0          49.0          81        22.0   \n",
       "1  ARI            372.0         366.0          39.0          81        -6.0   \n",
       "2  ATL            359.0         286.0          42.0          81       -73.0   \n",
       "3  BAL            353.0         408.0          49.0          81        55.0   \n",
       "4  BOS            383.0         433.0          43.0          81        50.0   \n",
       "\n",
       "   Runs as Visitor  Runs by Home  Wins as Visitor  Road Games  RD as Visitor  \\\n",
       "0            341.0         377.0             36.0          81          -36.0   \n",
       "1            354.0         341.0             40.0          81           13.0   \n",
       "2            287.0         401.0             25.0          81         -114.0   \n",
       "3            305.0         340.0             32.0          81          -35.0   \n",
       "4            315.0         370.0             35.0          81          -55.0   \n",
       "\n",
       "      RD   Win Pct  \n",
       "0  -14.0  0.524691  \n",
       "1    7.0  0.487654  \n",
       "2 -187.0  0.413580  \n",
       "3   20.0  0.500000  \n",
       "4   -5.0  0.481481  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def merge_data_frames(home_df, visit_df):\n",
    "    \"\"\" \n",
    "    Merge dataframes with statistics grouped by home and visiting teams\n",
    "    and to explicitly compute explanatory and response variables.\n",
    "  \n",
    "    Parameters: \n",
    "    home_df (dataframe): Input dataframe.\n",
    "    visit_df (dataframe): Input dataframe.\n",
    "  \n",
    "    Returns: \n",
    "    dataframe: Dataframe with stats grouped by home and visiting teams. \n",
    "  \n",
    "    \"\"\"\n",
    "    \n",
    "    # Compute explanatory and response variables\n",
    "    overall_df = home_df.merge(visit_df, how='outer', left_on='Team', right_on='Team')\n",
    "    overall_df['RD'] = overall_df['RD at Home'] + overall_df['RD as Visitor']\n",
    "    overall_df['Win Pct'] = (overall_df['Wins at Home'] + overall_df['Wins as Visitor']) / (overall_df['Home Games'] + overall_df['Road Games'])\n",
    "\n",
    "    # Return dataframe with explanatory and response variables\n",
    "    return overall_df\n",
    "\n",
    "# Invoke method to merge home and visitor dataframes\n",
    "overall_df = merge_data_frames(home_df, visit_df)\n",
    "\n",
    "# Display\n",
    "overall_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def extract_linear_reg_inputs(input_df):\n",
    "    \"\"\" \n",
    "    Put together all data preprocessing steps.\n",
    "  \n",
    "    Parameters: \n",
    "    input_df (dataframe): Input dataframe.\n",
    "  \n",
    "    Returns: \n",
    "    dataframe: Dataframe with preprocessing done. \n",
    "  \n",
    "    \"\"\"\n",
    "    \n",
    "    # Rename columns\n",
    "    input_df = rename_columns(input_df)\n",
    "\n",
    "    # Add new columns\n",
    "    input_df = add_new_columns(input_df)\n",
    "\n",
    "    # Group and process data by home team\n",
    "    home_df = proc_home_team_data(input_df)\n",
    "\n",
    "    # Group and process data by visiting team\n",
    "    visit_df = proc_visiting_team_data(input_df)\n",
    "\n",
    "    # Merge home and visitor dataframes\n",
    "    overall_df = merge_data_frames(home_df, visit_df)\n",
    "\n",
    "    return overall_df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model Construction & Evaluation\n",
    "\n",
    "Get training data from 2011-2015 to train the linear regression model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Initialize arrays to hold training data\n",
    "train_run_diff = np.empty([0, 1])\n",
    "train_win_pct = np.empty([0, 1])\n",
    "\n",
    "# Loop through the years from 2011 to 2015\n",
    "for year in range(2011, 2016):\n",
    "    # Construct log file name\n",
    "    log_file = \"GL\" + str(year) + \".TXT\"\n",
    "\n",
    "    # Read log into a dataframe\n",
    "    df = pd.read_table(log_file, sep=\",\", header=None)\n",
    "\n",
    "    # Extract relevant stats into another dataframe\n",
    "    df_proc = extract_linear_reg_inputs(df)\n",
    "\n",
    "    # Add to training set\n",
    "    train_run_diff = np.vstack([train_run_diff, df_proc['RD'].values.reshape([-1, 1])])\n",
    "    train_win_pct = np.vstack([train_win_pct, df_proc['Win Pct'].values.reshape([-1, 1])])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Instantiate an object\n",
    "linear_reg = linear_model.LinearRegression(fit_intercept=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LinearRegression(copy_X=True, fit_intercept=True, n_jobs=None,\n",
       "         normalize=False)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Compute model parameters with training data\n",
    "linear_reg.fit(train_run_diff, train_win_pct)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Slope (a) =  0.0006467879497387064  Intercept (b) =  0.4999936726219291\n"
     ]
    }
   ],
   "source": [
    "# Access and display model parameters\n",
    "print(\"Slope (a) = \", float(linear_reg.coef_), \" Intercept (b) = \", float(linear_reg.intercept_))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "R-squared for linear fit =  0.8625081265362202\n"
     ]
    }
   ],
   "source": [
    "# Get regression score (R-squared)\n",
    "r_squared = linear_reg.score(train_run_diff, train_win_pct)\n",
    "print(\"R-squared for linear fit = \", r_squared)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEKCAYAAADjDHn2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJztnXl4VEX2sN+TQIQgKAR3IKCDKGuEiKCjQUFFVNxQUXTADQgwIzMug4Lihqi48HNGNp0RR+IA4gYKIjDiBqhhxAUUZYfRTyEqomEN5/vj3g6dTi+3k+50d3Le56kn99atW3UKkj5ddeqcI6qKYRiGYYQjLdECGIZhGMmPKQvDMAwjIqYsDMMwjIiYsjAMwzAiYsrCMAzDiIgpC8MwDCMicVUWItJTRFaLyBoRGRGizRUiskpEVorIC371JSKywi2z4ymnYRiGER6Jl5+FiKQDXwNnA1uAj4GrVHWVX5uWwEzgLFX9SUQOV9Uf3Ge/qurBcRHOMAzDiIp4riw6A2tUdZ2q7gGmAxcFtLkJeEpVfwLwKQrDMAwjuagVx76PATb73W8BTgloczyAiHwApAP3qOqb7rM6IlII7AMeUtVXAwcQkYHAQIB69ep1OuGEE2I7A8OIAatXrwagVatWCZbEMMqzfPnybap6WKR28VQWEqQucM+rFtAS6AY0Ad4Tkbaq+jPQTFW/FZFjgf+IyOequrZMZ6pTgCkAubm5WlhYGOs5GEal6datGwCLFy9OqByGEQwR2eilXTy3obYATf3umwDfBmnzmqruVdX1wGoc5YGqfuv+XAcsBk6Ko6yGYRhGGOK5svgYaCkiLYD/AX2BqwPavApcBUwVkcY421LrRKQhUKyqu93604BH4iirYcSNUaNGJVoEw6g0cVtZqOo+YBgwH/gSmKmqK0XkPhHp7TabDxSJyCrgbeA2VS0CTgQKReRTt/4h/1NUhpFK9OjRgx49eiRaDKMirFsHV14Jv/ySaEkSTtyOzlY1ZrMwkpUVK1YAkJOTk2BJjKiYPx+uugpUnevOnRMtUVwQkeWqmhupnXlwG0acGT58OMOHD0+0GIZXVGHsWDjvPGjSBAoLq62iiIZ42iwMwzBSi19+gQED4JVXnFXF009DvXqJliopMGVhGIYB8NVXcMkl8M038PjjMHw4SDAPgJqJKQvDMIxXX4U//AHq1IEFC+DMMxMtUdJhNgvDMKoVBQXQvDmkpTk/CwrCNC4pgVGjnBVFq1awfLkpihDYysIw4syDDz6YaBFqDAUFMHAgFBc79xs3OvcA/foFNP7pJ7j6anjzTbj+enjqKWdlYQTFjs4ahlFtaN7cURCBZGfDhg1+FZ995qwmNm+Gv/3N0Sg11D5hR2cNI0lYsmQJS5YsSbQYNYJNmzzU//vf0KUL7NoF77wDgwbVWEURDbYNZRhx5s477wQskGBV0KxZ8JVFs2bAvn1w++3wxBPw+9/Diy/CkUdWuYypiq0sDMOoNowZA5mZZesyM+HR23+As892FMWwYbBokSmKKLGVhWEY1QafEXvkSGfrqVkzmHj9x5w39lLYtg2ee845ImtEjSkLwzCqFf36+Z18+uc/YcgQZxXxwQfQsWNCZUtlbBvKMIyYE5WvQzzYvRsGD4YbboDTT3f8J0xRVApbWRhGnBk/fnyiRahSovJ1iAf/+x/06QPLlsFf/+oYMtLTq2Dg6o35WRiGEVM8+zrEg/feg8svh19/halTHaVhhMX8LAwjSVi4cCELFy5MtBhVhidfh1ij6jjXnXUWNGgAH35oiiLG2DaUYcSZBx54AKDGZMsL6+sQD3budBzrnn8eLrjA+XnooXEarOZiKwvDMGJKKF+HMWPiMNiGDXDaaTBtGtx7L7z2mimKOGErC8MwYkowX4cxY+Jg3F6wAPr2dSLHzpkD558f4wEMf2xlYRiGJ6I5Dtuvn/Olf/9+56e/ovDvp3Fjp0R1xFYVHn4YevaEo46Cjz82RVEF2MrCMIyIxOo4bGA/RUUHnnnqc8cOJ5z4rFlwxRXwj3/AwQdHNRejYtjRWcOIM6tXrwagVatWCZak4sTqOGyofjz1+fXXTljxr75yVha33GLRYmNAUhydFZGeIrJaRNaIyIgQba4QkVUislJEXvCr7y8i37ilfzzlNIx40qpVq5RWFBC747Be2gdtM2cOnHwyfP89vPUW3HqrKYoqJm7KQkTSgaeA84DWwFUi0jqgTUvgDuA0VW0DDHfrGwGjgVOAzsBoEWkYL1kNI57MmTOHOXPmJFqMkAwZArVqOZ+9tWo594GEOvYa7XFYL+3LtNm/H0aPht69oWVLJ2xH9+6exgpmYwlld0l4eJJUQFXjUoCuwHy/+zuAOwLaPALcGOTdq4DJfveTgavCjdepUyc1jGQkLy9P8/LyEi1GUPLzVR2LcdmSn1+23bRpqpmZZdtkZjr10RCsn5B9/vSTaq9ezoMBA1SLiys1TkaGau3a5cfLz4/N3FIVoFA9fKbHcxvqGGCz3/0Wt86f44HjReQDEVkmIj2jeBcRGSgihSJSuHXr1hiKbhg1gylTvNX36+fUZWc7K5DsbOc+2uOwgf1kZTmlXJ9ffAG5uc6W01NPOdFj69b1PM7IkQeM6D727IG9e8vWFRc7Ywa2LS52+jAOEM/TUME2FAOt6bWAlkA3oAnwnoi09fguqjoFmAKOgbsywhpGTaSkxHt9mdDflSBiPzNnwnXXOWE7Fi92nO6iJBpbSqh/g7iGJ0lB4rmy2AI09btvAnwbpM1rqrpXVdcDq3GUh5d3DcOoJKGCsaZF8cngZb8/lP2gcWNnVSECR2TtY9X5t8GVV0JOjmOf8KgoAvtv1Mi7/KH+DeIWniRV8bJXVZGCs2pYB7QAMoBPgTYBbXoCz7nXjXG2nrKARsB6oKFb1gONwo1nNgsjWUlFm0Vamrc9ey+2jFD2g7S0A/eN+UEXcpYq6OoeQ1R37/Y8h2D9167tjGE2i8jg0WYRN2XhyEAv4GtgLTDSrbsP6O1eC/A4sAr4HOjr9+71wBq3XBdpLFMWRrKyadMm3bRpU6LFCEm9esEVRnZ25HezsyO/G6qNr3SkUDfQTHdykPbnWU/jepEhK8t5JuL8nDbNKYF1qqHrawJelYU55RlGDSctzfl4DUTEObla2XdDtQHoz1QmMZgfOJxLeZnl5HoaN1byG0nilGcYBsyYMYMZM2YkWowy+O/xh7JPRO0TEaI+WJva7OHvDGUq1/EBp9GJ5Swn1/O40cpgVB5TFoYRZyZOnMjEiRMTLUYpvvhMGzc638hDnQbq1StyX17CkQe2OZLv+A9nMZQJjONWzmU+2zgMgIyM6EOZV2lI9BqMKQvDqGEE80EIxty5kdt48b/wb9OVJSynEyfxCVcyndsZR4l7gj8ry3GnqKzvRkV9QIzwmM3CMOJMt27dAFi8eHFC5fARzobgT0z3/FVh4kT2DB3OJppxMa+ykrbxGcuICrNZGEYNJZTfgy8GlNfvh/57/gUFTiRwn09EenrwGFJB2bnTCSs+dCjv1z2bXArLKIrAsYzkxPJZGEY1IlTeiWefhUWLvPfjv+dfUAD9+5e1bezfDz4zzIQJYTrauBEuu8xxsLv7br773Wj2Dk4Dv20wsy+kCF7O16ZCMT8LI1nZunWrbt26tUrGiuTTEMonIj8/tJ9BuD7T08MIs2iRauPGqg0aqL72Wml1TfZpSEYwPwvDqHl4tUf4E6l9pD7LPVOFxx6Dv/4VTjgBXn4ZUjyfR3XGbBaGkSRMnTqVqVOnVknOhGj3/kPFRfLaZ7n3f/0V+vaF226DSy+FZcvipiiq4t/T8lz44WX5kQrFtqGMZCUvL09POCGvSuIPhYrV1L178G2kwLwVofoUCf5+9+5+Db/5RrVtWyfo00MPqe7fH9vJeZhnLP89q2KMZIBkiA1VlcWUhZGs5OXl6UEH5VU4/lK0hLIJ5Oc7NgafrcGLovCRlRVcWZTK//rrqoccotqokepbb8V2QkHwEpMqFcZIBrwqC9uGMowqYPfu4PXxyJnQrx9s2OCcWNqwwbkvKHCc7EpKnK2jkhLn3mta0R9/DD7W5o374b774MILoUULKCyEs8+O/aQCiFVO8ESPkUqYsjCMKuCgg4LXV4V/gX94DzhwBNZ3rHbIkLLhP3z1/gojmJwN2M6bdS9xcmT36wcffOAojCqgKuJBWcypspiyMIwqoEWLxMUvChfew2ta0cD4Syeyio+lM933zIUnn4R//av8BONIVcSDsphTAXjZq0qFYjYLI1n57bff9Lfffou7f0Go/kMZpyMVkeD99+FF/VXqafEhR6i++25sJxEFVeGvURN8QjA/C8OoOQR6boPzLXjKFGeF4NuCCobPhhFIVpYT4mPTJmfr5cH7S7h65Uh4+GHo0gVmzYJjjon9ZIwqxfwsDCNJmDBhAhPCxsSoPMG2mnxbScG2U3xkZjpKJvB5Rgb88ssBO8aOjUUcOaCnoygGDYLFi01R1DBMWRhGnJk5cyYzZ86M6xjhTu74h/CGA450vlDeEyaUD/Fdvz7s3eu0y+ETltOJ0/a/y831noFJk0Jb7I1qiykLw6gGRDq54ztOqwr79jk/fcdq/Z/7jtv6jspew/Ms4VTSKeF03uPJ326o2V7MNRhTFoaRQGIVTqJXL2dV4E/gyR1fiHIR52e4EOPHNt3L//EnnucPLKMLnVjOx3QGDpySCiW7lzlZGI0UxIsVPBWKnYYykpW8vDzNy8srVx+rcBLB+hEp66Gdnx/8xFNQL+7vvtPvW52uCvoYf9Z09pbrO5Ts+fmR51RTwmikCthpKMNIDkJlymvePPgppexsZyvIK176qVUr+Imn9HRnW6qUpUuhTx/46ScGpT/DlF+vDtovBB8z1Mkqf1liNW8jNiTFaSgR6Skiq0VkjYiMCPJ8gIhsFZEVbrnR71mJX/3seMppGPFk8eLFQVOqxiqchJd+gn2Al6lXhcmTIS/PMV4vXcoZk64O6ZQWasxQ4/i3tzAaqUnclIWIpANPAecBrYGrRKR1kKYzVDXHLc/41e/0q+8dLzkNI1GEM0oH7ukPGRJ6j99LWIq0EH/p6enArl1w000weDB07+7Ed+rQocwpKt8pqSlTHGN4qDFDhTz3b29hNFIUL3tVFSlAV2C+3/0dwB0BbQYAfw/x/q/RjGc2CyNZGTdunI4bN65cfTT7/oHFf48/kg1g2rQD0WYDy53XblI9+WTnZuRI1X37PM3JbBbVBxIdohzoAzzjd39toGJwlcV3wGfALKCp37N9QCGwDLg4xBgD3TaFzZo1i9M/pWFUjlAGbtXg4SS8pkb1D5UdLixFqP561nlb9bDDVOvXV33llajnFWpMLyEyakIYjVTBq7KIm4FbRC4HzlXVG937a4HOqvpHvzZZ7gpit4gMBq5Q1bPcZ0er6rcicizwH6C7qq4NNZ4ZuI1kJZSBOxReU6OKOH4R0fenDGc847iNWie0hFdecdKfGjWSZDBwbwGa+t03Ab71b6CqRarqi/T/NNDJ79m37s91wGLgpDjKahgJI9A+0aiRt/f89/j9fSh8xWfb8G+XyW8U0I8n+AsL6/aGDz+MmaKIte+E+WIkGV6WHxUpQC1gHdACyAA+BdoEtDnK7/oSYJl73RA4yL1uDHwDtA43ntksjGQl0jZU4P59RoZq7drht6D89/hD+VAE2hGOZY2uoL2WIHp37TE67V8lMZtjrO0QZteoOki0zcKRgV7A18BaYKRbdx/Q270eC6x0FcnbwAlu/anA527958ANkcYyZWEkKz179tSePXsGfRbKnpCVVXZPPz8/9B5/KOO1v21j0W3z9Oe0Q7WIhnrt4W/G/EM31ilIa0pK02TAq7IwpzzDSCCh7BNe7RG+tiGfsZ87GcsDche0bw8vvwzHHlsxYcMQi3nEsz8jNMlgszAMIwKx8DkI5dtQn194ict4gFG8Wvcq6n26hOZnHVtm7z9WdoFY+06YL0byYcrCMOLM/fffz/333x/0WSxSdw4cWL7uBL7kIzpzIXO4Nf0JLimeRjGZZfJr++fmVg2ee9srsU5BailNkxAve1WpUMxmYSQr4QzcqrHxOcjPP2C7uJiX9RcO1q1ph2nvBm+H3PuPtV0g1r4T5otRNZAMBu6qLKYsjGQg2AdcJGURqS84oAiyspwSzPDdotk+/bz3nU7Dzp1VN28OmX9bJHRu7sDc29HO10gtYqYsgEzgLuBp974lcIGXzquymLIwEk2o454nnBC9sgjWV7jSkCKdx7mqoN90u1F11y5VDb96qOzKwo63Vg+8KgsvNotngd04sZ7AcbZ7IMa7YYaR8oTKg71+fWz6CkV7PqWQXM7kbW5iCj3WP12a9jTc3n9l7QLh8n4b1Q8vyuI4VX0E2AugqjuBMIf1DKNmEirE9u7dWWRlZcWkr0CupoCldCWDPeTxDs9wU5l3w0WODfesMjJaqPHqiRdlsUdE6gIKICLH4aw0DKNGEip8uIZwWcrOfomXXnrJU1++k0iRjojWYi9PMJwCrqGQXDqxnA/pEvRd//zaY8Y43/x940HZ3NteFUU4Ge14azUl0j4VcDbwDrAVKAA2AN287HFVZTGbhVEVRGtLCLeHH27PP9w4h/P/9G3yVEGf4GatxZ5KjxerfwuzWaQexPI0FJAFnA9cADT28k5VF1MWRlXgNXy4z1A8bZrqiBEjdMSIEZ778hmYg4UrP5kPdTPH6G/U1Wvl+bBhQKIdryLYaajUx6uyiBjuQ0Q6BqneDmxU1X1BniUEC/dhVAUVCR8eKkS515AWvnY38AxPMZRvOZpLeIXPJCeq0BcWQsMIRizDfUzASUA0BSeM+FJgOvC1iJxTKSkNI8mJRfjwaNsE1v+u6W4mMYhnuInFdCOXQj4lJ2rbQCjZzcZgeMGLstgAnKSquaraCSevxBdAD+CROMpmGAklWDiMHTugdu3w73k9furp6OqWLSypnccgpjCWEfRiLj+SFXXoi4ICR/ZAate2EBqGN7woixNUdaXvRlVX4SiPdfETyzASTzA/gj17oEGDssdN8/Mrdvw04tHVd9+FTp1o/P1K3v3TLCZnj0UlPeojrr657NlTvr5Bg+j6MWoutTy0WS0iE3G2ngCuxNmCOgjX98IwqiOh/AV+/BG2bfPeT5MmTUI+8/k7lEEV/vY3uOUWJ5z4229zRuvWbPg/72MGEm4uhuEFLyuLAcAaYDjwZ5zsdwNwFMWZ8RLMMBJNZfwI/G0d778/jfPOm+Zt0OJi+MMf4OaboVcv+OgjaN26XJ+NGzvFa2hxs1cYlcbLkalUKHZ01og1FfUjqLD/wbp1qjk5zjnU++9XLTmQ9jSSf0ck/4qMjPLv1K5tR10N70dnIzdwAgfOAlbhrCrWAeu8dF6VxZSFEQ8q4kdQ3p/hZoWbw/szzJ+v2rCh6qGHqr7xRrnxvfp1eJPHKVlZHv8RjGqNV2XhxWbxLDAaeAJn2+k6LDaUUUMIalOIQHn7wIoQ9Tif2w895Fig27Z10p7+7nfAgdNYXgMKRhuryewVRjR4sVnUVdVFgKjqRlW9BzgrvmIZRmpSUODYEYJRzj6wYwf06QN33glXXglLl5YqCogu8mzQ/itYbxjB8KIsdolIGvCNiAwTkUuAw+Msl2GkHL6VQElJ+WdpaQH+DKtXwymnwGuvwWOPwQsvQL16Zd6JJnprOL8LS1FqxAIvymI4TgKkPwGdgGuAP8RTKMNIRcKtBI4/3m87a/Zs6NwZtm6FBQvgL39xHC0CCOctnpXlFC++HZUNRW4Y4E1ZNFfVX1V1i6pep6qXAZ4WsCLSU0RWi8gaERkR5PkAEdkqIivccqPfs/4i8o1b+nufkmEkhtArgeM5/fTjnSXH3XfDRRc52mP5cjgz+OnzcB7X06Y5fh7btnkPLe4fpjzaUOSGAd6UxR0e68ogIunAU8B5QGvgKhFpHaTpDFXNccsz7ruNcIzqpwCdgdEi0tCDrIZRZQTGjQrYRSolO3sKZ+c+zNsHXwj338/Metfx7yHvlTEaBPZ1882RPa5D5cIwjHgQ8jSUiJwH9AKOEZEn/R41ALxEm+0MrFE3LIiITAcuwjmCG4lzgQWq+qP77gKgJ/BvD+8aRtwZMgQmTToQxXXjxuDtMjLgxlM+p+PgS2iqm8hnApN+G0zmMGF/hvPBH3jqKVRfcOAEU7B3Bg50rm3VYMSDcEdnvwUKgd7Acr/6HTie3JE4Btjsd78FZ6UQyGUicgbwNfBnVd0c4t1jPIxZYXxhpP254oorGDJkCMXFxfTq1avc8wEDBjBgwAC2bdtGnz59yj3Pz8/nyiuvZPPmzVx77bXlnt9yyy1ceOGFrF69mkGDBpV7PmrUKHr06MGKFSsYPnx4uecPPvggp556KkuWLOHOO+8s93z8+PHk5OSwcOFCHnigfNr0yZMn06pVK+bMmcNjjz1W7vnzzz9P06ZNmTFjBhMnTiz3fNasWTRu3JipU6cyderUcs/nzp1LZmYmEyZMYObMmeWe+0J2P/roo7z++utlntWtW5d58+YBcP/997No0aIyz7Oyskqzz91xxx0sXbq0zPMmTZowbZrjNT18+HBWrFhR5vnxxx/PlClTABg4cCBff/11mec5OTmMHz8egGuuuYYtW7aUPvv+e/jqq67AWLfmMqAoYHbdgbv4Q8Z0Js28ir+Rxi468AszgBkUF1/AyJG30q8f3HBDN3aXyz15BTAEKMb5zuaQkQHdusEXXwyguHgAsA1wfveKi+GGG+Dpp+13r7r+7gF07dqVsWOd373LLruMoqKicuHv40FIZaGqnwKfisgLqlqRGFDBfDECo+nPAf6tqrtFZDDwHM6xXC/vIiIDgYEAzewcoBGB77+H9eth927HXJCXV7Fv4evXR24j7OcRbuXWXx+jAen8Sj2UBmXa+Gwc5RVFcNLSoEUL57ooUDcRXV+GETWRvPaA04AFON/81wHr8eDBDXQF5vvd3wHcEaZ9OrDdvb4KmOz3bDJwVbjxzIPbCEcsU4CKhPekbswPuogzVUGn1BmmwhkKeSE9rsN5WIfyHo9H1jujZkIMw318hWOkPhwnvWoWkOXhvVqucmkBZACfAm0C2hzld30JsMy9buQqpYZuWQ80CjeeKQsjHLH8cA0XfiOXj3QjTbWYOnpTxlTNylJXUZRVFiIHPvwrosgs/7URK2KpLD700lGId3u5K5K1wEi37j6gt3s9FljpKpK3cXJn+N69Hifa7RrgukhjmbIwwhFqNSASfV/BPqhFVMed+A/dxUG6gWba68jlOm2ab9zyygLK9xltDCrLf23EAq/KwksO7ofcLaKXgdIdUVX9b9gXqxjLwW2Eo3nz4KeM0tPhuee82y6GDHEc2vy9tH/XbA9vtLyZ4xdNgu7dYfp0J3546bg+A/H4oOOXlDiOcmPG2Ekmo+rxmoPbSyBB3wkm/84Uiw9lpBBjxgQPyldS4v3I6ZAhEHgw5yi+Zf7uPhy7aCncfrszUK0Df1bOuONDenb7lI4dfTWSnYgri1TBVhZGJAoKoH//4LGbsrMdz+Zw1KpV9t3TeJ8XuZz67ODgmc/C5ZdHPW5F5DCMWOJ1ZRHRg1tEjhCRf4jIPPe+tYjcEAshDaMq6dfPCXcRDC9B+w582CtDeIq3OZNfOZguLAupKADmzbuGkpJrPMkYTfBAw6hKvIT7mArMB45277/GCS5oGClHZcJ1p6dDHXYylQE8xTDepCcn8zFfpbct066gwDFZiDhlxowt1Kq1JUSv0cthGInAi7JorKozgf0AqroP8LCgNozkozLhukdctZH3+T39+RejuYeLeI3tHFpqawBHUVx/fVmnuX37nJKREb5/CxtuJDNelMVvIpKF60EtIl2A7XGVyjDiRIXDdS9cyAPzOnFixlouSpvDfYwmLT2N/HyYMOFAs5EjgwcABKhf3xkPnFWK/08LG24kO15OQ/0FmA0cJyIfAIfhC0ZjGClIVKlSVeHRR2HECDjxRDKXvsJrLVuGbB7O5vDjj05YccNIRSKuLFx/ijzgVGAQjhf2Z/EWzDCqmsCQ3zP+8auT7vT223m9zmXUX7mMxl1b0rhx2bDg/u8FT6naFehq9ggjpYm4shCRoUCBqq507xuKyFWqOiHCq4aRMgSG/K698Rva3XQx+/mKu2o/woPFtwLCr362iI0b4brrnO0s39ZT8OOxY6ld2+wRRmrjxWZxk6r+7LtR1Z+Am+InkmFUPf4pUc/ndQrJ5XD9nvNkPg/uvY3ggZBh797gNgr/FUZWFjz7rNkjjNTGi7JIEzmQINjNgBfhXIdhpBabNjlhxUdzD69zIWv4HbkU8tb+HhXqzz8KVF7eZbz88mUxltgwqhYvyuItYKaIdBeRs3Cy1b0ZX7EMIzw+O4GI41kt4twPGVLW7uC7D2wXmIK0bZOfeY2LuId7mUp/fs/7bKR56WmlaPG3TxQVFVEUKgGFYaQIXk5D3Y6TYCgfZy3+FvBMPIUyjHAE2hf84yv5x24KvA8Zh2nlSt7fewl1Wc9Q/s4EhgBCZqYTpuO558rHlAqH+UsY1ZGwKwt3y+lfqjpJVfuo6mWqOllVzSnPSBj+9oWKUlzs9MOLL8Ipp9CAHbx919u8kT0UESn1e5gwoaxfRlaWU0KRnm7+Ekb1JOzKQlVLROQwEclQ1RCuRoZRtcQiflI6+xi68U64Yhx07QqzZnHO0Uez4b7ybYP5ZaSlOfaIQPbvN0VhVE+82Cw2AB+IyF0i8hdfibNchhGSyvorZLGNN+nJbYyD/HxYvBiOPjpo20DfC5+tI5oYU927d6d79+6VE9owEowXm8W3bkkD6sdXHMOITKjcFF44if/yMpdyJP+PpTf9k64TrgvZNtA24m/rCCZDKFvFXXfdFb2ghpFseEmn5+a8qOe1bSKKpVVNLSqbEtT3Pqimp2tpPu38/LL9+u5Btb88p8XU0S3pTXXufR9HHCNS3m5La2pUB4hhDu6uwCpgk3vfAZjgpfOqLKYsUodgOawzMyuuMEJ9WPueZ7Bbnz5oqCroIs7UI9J+KKdcApXOtGnBFUVF8nb37NlTe/bsGd1LhlFFxFJZfAjetmtSAAAgAElEQVQ0BT7xq/vCS+dVWUxZpA6RvrF7IZLC8T0/km/1XX6vCjqOWzSdvSGVgH/JyAj9LBo5VVXz8vI0Ly8vupcMo4rwqiy8GLhR1c0BVXZ01qgwoU4zRXPKKdjx2dLjsO7zDsVLWE4nOvJf+vJvbuNRSjyZ6UKHGRcxHwqjZuJFWWwWkVMBFZEMEbkV+DLOchnVmMpkq/MRVuGo0mvjRBbTjWIy6cpSZtA3ajmDoWpHY42aiRdlMRgYChwD/A/Ice8jIiI9RWS1iKwRkRFh2vURERWRXPe+uYjsFJEVbpnkZTwjMYQ6XhoKL9nq/Pts3JhyYcFDKZaWTXexpPUNTGAICzibk/mYz2lf8ckF4EteZBg1Di97VRUpQDqwFjgWJ/Dgp0DrIO3qA+8Cy4Bct645UdpFzGaRGCpqrA5nnA7WZ2D/+fnl2xxfZ6N+VT9XFfRe7lKhxJN9wmupiBFeVXXcuHE6bty46F80jCqAGBq4jwXmAFuBH4DXgGM9vNcVmO93fwdwR5B244ELgMWmLFKPWBirvfYZ2L+/wul7xH90Z/3Gup362ptXg76TlaUhj9p6Hc8wqhtelYUXa98LwFPAJe59X5zIs6dEeO8YwN8wviXwHRE5CWiqqq+7thB/WojIJ8AvwChVfc+DrEYVEwtjdUXe3bTJDcNxtcLjj8Ptt0OrVnT48hW+plXQdyKlNA0VwkMENmyILJNhVGe82CxEVZ9X1X1umQYE+ZMq/16QutL3RCQNeAK4JUi774BmqnoSTg7wF0SkQbkBRAaKSKGIFG7dutWDSEasiYWx2p+CglCpSYP0/9tvcPXVcOutzK1zCQ2+/DCkovASajzWc/HRrVs3unXrVrlODCPBeFEWb4vICNfonC0itwNviEgjEWkU5r0tOP4ZPprghA3xUR9oCywWkQ1AF2C2iOSq6m5VLQJQ1eU4to/jAwdQ1SmqmququYcddpiHqRixxoux2iu+8BrBU5OW7f///rQWunZl/4yZ3F17LOcXv8iOMNFofGE6whHLuRhGtSPSPhWwPkxZF+a9WsA6oAUHDNxtwrRfzAGbxWFAuh6wmfwPaBROTrNZJI5Yhb0IZTsQcewNvv7/c9tc1UMPVW3USK85fH5Ee0O9elU/F3/MKc9IZoiVzUJVW1RQCe0TkWHAfJyTUf9U1ZUicp8r3Owwr58B3Cci+3AcAAer6o8VkcOIP8FCeFeEcLaKbdtw4n+PGQOjR0OHDvDyyxQcF/nXM5qAg7Gai2FUNzx5cFcUVZ2rqser6nGqOsatuzuYolDVbqpa6F6/pKptVLWDqnZU1TnxlNOIPdH6XkAEm8H27XDppXD33c6n+QcfULCkhSf7hqp3GQIJN4+KzNEwUhVvsQ8MIwrChfYO9629V6+yaVB9XN9lFXS+BNatgyefhGHDKHhBPNk3fHiVwes8wPscr7jiCm8DGkYSIxrsrGAKkpubq4WFhYkWw8D5lr1xY/n67OzwR1CDvXcpL/GcDODgwzKdFKhnnBF2jLQ0aNgQioqCjxFJhkjy+PqAis3RMJINEVmuqrkR23lRFiJyDJCN30pEVd+tlIQxxpRF8hDOX2H/fm/vpVHCA4ziDh5iGafQZctLcMwxnseoqAxe5wHe+y92lx+ZgUetDCMJ8KosIu74isjDwAfAKOA2twQ60BlGKRX1V/A9b0QRc+nFHTzEZAZyVto7pDU9Jqq0pqGeNwp32NujvM2aRTfHXr160atXL+8DG0YS4sXAfTHQSlV7qeqFbukdb8GM1KWi/gpjxkDXOp9QSC7dWMyNPM1gJrNz/0GoHrALFBREHmPMGMjIKD/GL794N0SHG8N8MowaR6SztcA84GAv53ATWczPIrmokL/C88/r3tp19Nv0Y/QUlpXGcAoVdyrSGL5YUJWJWxUp4KGXOZqfhZHM4NHPIqLNQkRewkmlugjY7adk/hQ/FRY9ZrNIYfbuhVtvdU465eXBjBlwxBGVtjvEwm4RC3yhPhYvXlx1gxqGR2JmswBmA/cDS4DlfsWoYcTSr8DX15HyPR/W7+EoiuHDYcECOOIIIHrbR6B8oewTlY31ZBg1ES8e3M9VhSBGclNR34lwfbUrXsZLXEbD3T9xXUYBPXKvpl/tA+3GjCk7JoS2CwSTr3Ztx27hnyI1EXaFAQMGVO2AhhEPQu1PATPdn58DnwUWL3tcVVnMZhFfYpm3Ijtb9SYm6y4ydC0ttD0rQvbl1S4QSr6srNjHejKM6gSVTX4EHOX+zA5WvHRelcWURXREa4AWCf5hLBLlwLt26dPcqAo6j3O1IUXl+vMiT6D8oYIIRi1fiP4ro2S2bt2qW7durXgHhhFHYqEshgMnA7W8dJToYsrCOxVJhRqTlcWmTaqdO6uC3s9ITWNfyA/5cPIEkz+UMqvIyqeiqWJDYaehjGTGq7IIZ+BuAvwf8IOILBaRB0Xk/Ag5LIwUYOTI8pFYi4ud+lBU2q/gnXegUyf48kveufllxmY+wH5CZyQKJ08w+VUPeFZXSL4I/Uf69zGM6k5IZaGqt6rqqcCRwJ3Aj8D1wBcisqqK5DPiQEVSofbrB1OmOLGPRJyfU6Z4MG6rwvjx0L27czzpo4/IG39Jmb5iJadqBeSLwbiGURPwcnS2LtAAOMQt3wIfxlMoI7bE6khpv35OkLz9+52fgR/EgeNM/2cxXHMN/PnPcOGFzLjlI5r3PIG0NOdb+pgxTl++wHxe5QlV7wviF0o+r8QrvaphpDSh9qeAKTgxod4E7gXOAxp62dtKRDGbRXCC7b9nZKjWrh27Pflg47RgrX4qHXS/iOoDD+i0f5WEtANEayOItU0h3v2bzcJIZoiBgftNoBCYCgwE2uFGqU3GYsoiOFV1pNR/nHOZp0U01CIa6h8OnxdWDq+hOwKJR/rTePU/ffp0nT59eqxEM4yY4lVZhA33ISICtAFOdUtbHNvFUlUdHY+VTkWxcB/BqaqQF844yh2M5QFG8TntuIRX2CDHxixkuGEYsScm4T5cxfMFMBcnoOAHwHHAzTGR0og7kfbfYxXCo3WTX3iJy3iQkUynL6eyhPUcS1qa02es7ADRypsMqU83b97M5s2bq35gw4gloZYcwJ+A6cBmYB3wPDAEJ6hgmpdlS1UW24YKTrj995jtzX/5pf589Am6l3S9mScU9pfrMz+/8mMlm23DK2azMJIZYmCzeBzog+vJnezFlEVoQu2/x8TR7pVXVOvXVz3sMH3rzrfDhhWvrB0gWnljGaKkMpiyMJIZr8rCcnDXYCplRygpgdGjnfOvJ58ML70ETZvG1TYRbd/JYiexEOVGMhPLEOVGNaXCdoQff4QLLnAUxQ03wLvvQtOmlevTA9H2bf4ShhE74qosRKSniKwWkTUiMiJMuz4ioiKS61d3h/veahE5N55y1lQqFMLj00+dlcSiRTB5MjzzDNSpU7k+4ySvpT41jBjiZa+qIgVIB9YCxwIZwKdA6yDt6gPvAsuAXLeutdv+IKCF2096uPHMZlExorIjvPCCat26qkcfrbp0aWz6jKe8cZbFK7Nnz9bZs2dX/cCG4QESbbMQka7APap6rnt/h6ucxga0Gw8sBG4FblXVwsC2IjLf7WtpqPHMZhFH9u2D22+HJ56A00+HmTPhyCMTLZVhGDEgGWwWx+Acu/Wxxa0rRUROApqq6uvRvuu+P1BECkWkcOvWrbGR2ijLDz/A2Wc7iuKPf3S2nyIoCp9vgwjUquX8DPRxSAb/h6pi9erVrF69OtFiGEaliJhWtRIEiydauowRkTTgCWBAtO+WVqhOwYlhRW5ubvU41pVMfPQRXHYZbNsG//oXXHttxFcC05uWlDg//dOwQuxStKYCgwYNAuw0lJHaxFNZbAGa+t03wYlY66M+TviQxU5UEY4EZotIbw/vGvHmH/+AIUPg6KNhyRI46SRPrwXLBeHDPydEqHwR1VFZGEZ1IJ7bUB8DLUWkhYhkAH2B2b6HqrpdVRuranNVbY5j4O6tqoVuu74icpCItABaAh/FUVbDx+7dMHgw3Hgj5OVBYaFnRQGRcz5s2mT5IgwjFYmbslDVfcAwYD7wJTBTVVeKyH3u6iHcuyuBmcAqnOi3Q1W1JF6y1gQKCqBxY8d+IOJcl7MT/O9/0K2bcyR2xAiYNw+ysqIaJ5IPQ6NG5v9gGKlIXP0sVHWuqh6vqsep6hi37m5VnR2kbTd3VeG7H+O+10pV58VTzupOQQFcdx0UFR2oKyqC66/3UxjvveekPf38c5g1C8aOhfTQaU9DEcy3wZ9ffoFevcz/wTBSDfPgrgGMHAl795av37MHRt6p8Le/wVlnQYMGB4zaFcQ//Wow9u6FuXMrmKI1RRk1ahSjRo1KtBiGUSksNlQNIFSMpLoUM4nB/IHnoXdv58TTIYdUaqyCAkc5bdoUfEywHBaGkUwkg5+FkSQEswVks4EPOI1rmAb33QevvBITRTFwoHMUNtx3kJpmm1ixYgUrVqxItBiGUSlMWdQAxoyB2rUP3PdgAcvpRAvW886tr8NddznLj0oS7tisj5pomxg+fDjDhw9PtBiGUSlMWdQA+vWDZ5+FrEbK7TzMm/Tk+/SjeefRQs4c1ytm44Q7+loTbBOGUZ0xZVFD6Nd7B9vOvJyHGUH6FX1o/fNSLrrldxXuL1i4jlDbS9nZjo1iwwZTFIaRqpiyqAl8/TV06eLYJR59FKZPh4MPrnB3gbYJX7gOOxJrGNUXUxbVndmznfwTP/wACxbALbc4e0KVIJhtori45h2JNYyahB2dra7s3w/33uucdOrUCV5+OWbHkJIlXWmqsGTJEgBOPfXUBEtiGOXxenQ2noEEjUTx88/O1/m5c2HAAJgwAerWjVn3zZo5W0/B6o3ymJIwqgO2DVXd+OILyM11tpwmTIB//jOmigIsXWm0LFmypHR1YRipiq0sqhMzZjgBnw45BBYvhjh9o/XZIHye2s2aOYrCbBPBufPOOwHLZ2GkNqYsqgP79jlRYh97DE47DV58EY46Kq5D9utnysEwahKmLFKdrVuhb1/4z39g6FB4/HHIyEi0VIZhVDNMWaQyy5fDpZfC99/D1KnQv3+iJTIMo5piBu5UZepUZ8sJ4IMPTFEYhhFXbGWRauzZA3/+s3PS6ayzHG/sww5LtFRGGMaPH59oEaqcvXv3smXLFnbt2pVoUQyXOnXq0KRJE2r7RxWNAlMWqcS338Lll8OSJXDbbfDgg1DL/guTnZycnESLUOVs2bKF+vXr07x5c6SSEQOMyqOqFBUVsWXLFlq0aFGhPmwbKlX44APHE/vTT50jso88YooiRVi4cCELFy5MtBhVyq5du8jKyjJFkSSICFlZWZVa6dmnTbKjChMnws03O+FdFyyAtm0TLZURBQ888AAAPXr0SLAkVYspiuSisv8ftrJIZnbudJzshg6Fc8+Fjz82RWEYRkIwZZGsbNwIp5/unHoaPdqJHnvooYmWyjDiQrD8KEZyEVdlISI9RWS1iKwRkRFBng8Wkc9FZIWIvC8ird365iKy061fISKT4iln0rFokRPf6ZtvHCVxzz0xSXtqGMlIqPwolVEYP//8MxMmTIj6vV69evHzzz+HbXP33XfH1QZ1zz338Oijj4Zt8+qrr7Jq1aq4yRCMuH0CiUg68BRwHtAauMqnDPx4QVXbqWoO8AjwuN+ztaqa45bB8ZIzqVCFcePgnHPg8MOhsBAuvDDRUhlGXAmVH2XkyIr3GUpZlJSUhH1v7ty5HBphBX/fffcl3P5UrZQF0BlYo6rrVHUPMB24yL+Bqv7id1sPqB7JNSrCr786YTtuv93xyv7wQ2jZMubD2HK/6pk8eTKTJ09OtBhJS6jc7eFyukdixIgRrF27lpycHE4++WTOPPNMrr76atq1awfAxRdfTKdOnWjTpg1Tpkwpfa958+Zs27aNDRs2cOKJJ3LTTTfRpk0bzjnnHHbu3AnAgAEDmDVrVmn70aNH07FjR9q1a8dXX30FwNatWzn77LPp2LEjgwYNIjs7m23btoWUd8yYMbRq1YoePXqwevXq0vqnn36ak08+mQ4dOnDZZZdRXFzMkiVLmD17Nrfddhs5OTmsXbs2aLuYo6pxKUAf4Bm/+2uBvwdpNxRYC2wGWrp1zYHfgE+Ad4DTQ4wxECgECps1a6YpyzffqLZtq5qWpvrww6r798dlmGnTVDMzVZ0ljFMyM516w4glq1at8tw2O7vs76SvZGdXfPz169drmzZtVFX17bff1szMTF23bl3p86KiIlVVLS4u1jZt2ui2bdtcWbJ169atun79ek1PT9dPPvlEVVUvv/xyff7551VVtX///vriiy+Wtn/yySdVVfWpp57SG264QVVVhw4dqg8++KCqqs6bN08B3bp1a1BZCwsLtW3btvrbb7/p9u3b9bjjjtNx48apqpbKpao6cuTI0rH8ZQjXLpBg/y9AoXr4TI/nyiLYOa1yKwdVfUpVjwP+Coxyq78DmqnqScBfgBdEpEGQd6eoaq6q5h6Wql7Mb7zh2Ce+/RbefNNZWcTpyGE8lvtGZObMmcOcOXMSLUbSUhX5UTp37lzGGe3JJ5+kQ4cOdOnShc2bN/PNN9+Ue6dFixalDpWdOnViw4YNQfu+9NJLy7V5//336du3LwA9e/akYcOGIWV77733uOSSS8jMzKRBgwb07t279NkXX3zB6aefTrt27SgoKGDlypVB+/DarjLE089iC9DU774J8G2Y9tOBiQCquhvY7V4vF5G1wPE4q4jqxRtvwLHHOmlPmzeP61DxWO4bkXnssccAuNDsT0Gpivwo9erVK71evHgxCxcuZOnSpWRmZtKtW7egzmoHHXRQ6XV6enrpNlSodunp6ezbtw/At/PhmVA+EAMGDODVV1+lQ4cOTJ06NWROFK/tKkM8VxYfAy1FpIWIZAB9gdn+DUTEf1P+fOAbt/4w10COiBwLtATWxVHWxPHEE453dhSKoqJ2h1BpTy0dqpFo+vWDDRucHO4bNlReUdSvX58dO3YEfbZ9+3YaNmxIZmYmX331FcuWLavcYEH4/e9/z8yZMwF46623+Omnn0K2PeOMM3jllVfYuXMnO3bsKLMK3bFjB0cddRR79+6lwO8PPXB+odrFkritLFR1n4gMA+YD6cA/VXWliNyHs0c2GxgmIj2AvcBPgC906hnAfSKyDygBBqvqj/GSNaH4fXvxgu+YoW87yXfMECL/gY0ZU/ZdsHSoRvUkKyuL0047jbZt21K3bl2OOOKI0mc9e/Zk0qRJtG/fnlatWtGlS5eYjz969GiuuuoqZsyYQV5eHkcddRT169cP2rZjx45ceeWV5OTkkJ2dzemnn1767P777+eUU04hOzubdu3alSqIvn37ctNNN/Hkk08ya9askO1iiUS7XEpWcnNztbCw+u1SBdK8uaMgAsnOdr6RRaKgwNKhVjXdunUDalZa1S+//JITTzwx0WIkjN27d5Oenk6tWrVYunQp+fn5rFixItFiBf1/EZHlqpob6V2LDZViVNbuYOlQDSP+bNq0iSuuuIL9+/eTkZHB008/nWiRKo0pixSjWbPgKwuzOyQvzz//fKJFMKqYli1b8sknn5SpKyoqonv37uXaLlq0iKysrKoSrcKYskgxzO6QejRt2jRyI6Pak5WVlRRbURXFAg6lGP36wZQpjo1CxPk5ZYptLSUzM2bMYMaMGYkWwzAqha0sUhCzO6QWEydOBODKK69MsCSGUXFsZWEYhmFExJSFYRiGERFTFoZhGBHwRaOtTJsnn3ySE088kX79+jF79mweeughIDHhxiuC2SwMw4gvw4dDrE8B5eTA+PGx7TPOTJgwgXnz5pUGNPQFDHz11Ve54IILaN06MN1PcmErC8OIM7NmzSrNf2BUHRs2bOCEE07gxhtvpG3btvTr14+FCxdy2mmn0bJlSz766CN+/PFHLr74Ytq3b0+XLl347LPPAMcn4pxzzuGkk05i0KBBZQIDTps2jc6dO5OTk8OgQYMiJlQCGDx4MOvWraN379488cQTTJ06lWHDhgXNTZG0eIljngqlU6dOQeO3G4ZR9USTzyJe+HJSfPbZZ1pSUqIdO3bU6667Tvfv36+vvvqqXnTRRTps2DC95557VFV10aJF2qFDB1VV/eMf/6j33nuvqqq+/vrrpfkoVq1apRdccIHu2bNHVVXz8/P1ueeeU9UDuTBC4f/82Wef1aFDh6pq+dwU8aQy+SxsG8ow4szUqVMBJ4y0UbW0aNGiNDtemzZt6N69OyJCu3bt2LBhAxs3buSll14C4KyzzqKoqIjt27fz7rvv8vLLLwNw/vnnl+ajWLRoEcuXL+fkk08GYOfOnRx++OEJmFnVY8rCMOKMKYvE4Z+TIi0trfQ+LS2Nffv2UatW+Y9AX26JYDkmVJX+/fszduzYOEmcvJjNIoFYPmzDSCxnnHFGaf6HxYsX07hxYxo0aFCmft68eaX5KLp3786sWbP44YcfAPjxxx/ZGCxYWxSEy72RTJiySBC+vBQbNzoZh315KUxhGEbVcc8991BYWEj79u0ZMWIEzz33HODko3j33Xfp2LEjb731Fs3cSJ2tW7fmgQce4JxzzqF9+/acffbZfPfdd5WSoW/fvowbN46TTjopqQ3cls8iQVQ2L4WROlg+CyNZqEw+C1tZJAjLh20YRiphBu4EYXkpag5z585NtAhGFZHqOSvCYcoiQVheippDZmZmokVICKoa9ERRdSaZc1ZU1uRg21AJwvJS1BwmTJjAhAkTEi1GlVKnTh2Kiooq/QFlxAZVpaioiDp16lS4DzNwG0acqYkG7r1797JlyxZ27dqVaFEMlzp16tCkSRNq165dpt6rgdu2oQzDiDm1a9cuDZhnVA/iug0lIj1FZLWIrBGREUGeDxaRz0VkhYi8LyKt/Z7d4b63WkTOjaechmEYRnjipixEJB14CjgPaA1c5a8MXF5Q1XaqmgM8Ajzuvtsa6Au0AXoCE9z+DMMwjAQQz5VFZ2CNqq5T1T3AdOAi/waq+ovfbT3AZ0C5CJiuqrtVdT2wxu3PMAzDSADxtFkcA2z2u98CnBLYSESGAn8BMoCz/N5dFvDuMUHeHQgMdG9/FZHVlZC3MRA+FVZqUF3mAdVsLiJSbeZC9fh/qS7zgMrNJdtLo3gqi2AHrMsdvVLVp4CnRORqYBTQP4p3pwBTKiknACJS6OVEQLJTXeYBNpdkpbrMpbrMA6pmLvHchtoCNPW7bwJ8G6b9dODiCr5rGIZhxJF4KouPgZYi0kJEMnAM1rP9G4hIS7/b84Fv3OvZQF8ROUhEWgAtgY/iKKthGIYRhrhtQ6nqPhEZBswH0oF/qupKEbkPJ43fbGCYiPQA9gI/4WxB4babCawC9gFDVTVyotvKEZPtrCSguswDbC7JSnWZS3WZB1TBXKqNB7dhGIYRPyw2lGEYhhERUxaGYRhGRGqcshCR+0XkMzfEyFsicrRbLyLypBti5DMR6ej3Tn8R+cYt/RMnfVlEZJyIfOXK+4qIHOr3LGi4lEghWBKFiFwuIitFZL+I5AY8S6m5+JMKMvojIv8UkR9E5Au/ukYissD9/V8gIg3d+pB/M8mAiDQVkbdF5Ev3d+tmtz6l5iMidUTkIxH51J3HvW59CxH50J3HDPcgEe7BoBnuPD4UkeYxEURVa1QBGvhd/wmY5F73Aubh+Hh0AT506xsB69yfDd3rhomehyvbOUAt9/ph4GH3ujXwKXAQ0AJYi3PIIN29PhbHCfJToHWi5+HKfCLQClgM5PrVp9xc/GRPehmDyHwG0BH4wq/uEWCEez3C7/cs6N9MshTgKKCje10f+Nr9fUqp+bjyHOxe1wY+dOWbCfR16ycB+e71EL/Ptb7AjFjIUeNWFho+xMi/1GEZcKiIHAWcCyxQ1R9V9SdgAU68qoSjqm+p6j73dhmOPwqEDpcSMQRLolDVL1U1mAd+ys3Fj1SQsQyq+i7wY0D1RcBz7vVzHPCHCvU3kxSo6neq+l/3egfwJU4kiJSajyvPr+5tbbcoTsSLWW594Dx885sFdBepfBaqGqcsAERkjIhsBvoBd7vVwcKTHBOmPtm4HudbEaT+XPxJ5bmkgoxeOEJVvwPnAxg43K1Pmfm5WzEn4XwrT7n5iEi6iKwAfsD5wroW+Nnvy6K/rKXzcJ9vByqd07VaKgsRWSgiXwQpFwGo6khVbQoUAMN8rwXpSsPUVwmR5uK2GYnjj1LgqwrSVUrMJdhrQeoSPhePpIKMlSEl5iciBwMvAcMDdhbKNQ1SlxTzUdUSdaJzN8FZsZ4YrJn7My7zqJbJj1S1h8emLwBvAKMJHWJkC9AtoH5xpYX0SKS5uAb3C4Du6m5SEj5cSsLCqETx/+JPUs7FI9UlbM33InKUqn7nbsv84NYn/fxEpDaOoihQ1Zfd6pSdj6r+LCKLcWwWh4pILXf14C+rbx5bRKQWcAjltxajplquLMIhZUOM9Aa+cq9nA39wT0R0Aba7S9T5wDki0tA9NXGOW5dwRKQn8Fegt6oW+z0KFS4lYgiWJCSV55IKMnphNm50Bffna371wf5mkgJ3n/4fwJeq+rjfo5Saj4gcJu5JRxGpC/TAsb+8DfRxmwXOwze/PsB//L5IVpxEW/qruuB8y/gC+AyYAxyjB04cPIWzF/g5ZU/kXI9jWF0DXJfoOfjJtQZnb3KFWyb5PRvpzmU1cJ5ffS+cUyFrgZGJnoOfXJfgfCPaDXwPzE/VuQTMK+llDJD338B3OCF4tgA34Ox3L8KJ3bYIaOS2Dfk3kwwF+D3O9stnfn8jvVJtPkB74BN3Hl8Ad7v1x+J8cVoDvAgc5NbXce/XuM+PjYUcFu7DMAzDiEiN24YyDMMwoseUhWEYhhERUxaGYRhGRExZGIZhGBExZWEYhmFExJSFUa0QkRJxIgp/ISJzxHnp4ooAAAMYSURBVC8Sb4zHWSxudFwRmet3Dv5PbpTTAtc3ZKErz5XxkMMd81ARGeJ3f7SIzIrwTnPxiyxrGJEwZWFUN3aqao6qtsXxWh0a7wFVtZeq/uzeDgF6qWo/nFhEtV15Znjpy/W4jZZD3XF98nyrqn3CtDeMqDFlYVRnluIGVxORbiLyuu+BiPxdRAa41xtE5F4R+a+IfC4iJwR2JCJ1RWS6OHkOZgB1/Z5tEJHGIjIJx1Fqtoj8FZgG5Lgri+NEpJOIvCMiy0Vkvi+iqbtKeVBE3gFudj12XxKRj91ymtvuHnHyTSwWkXUi8idXhIeA49xxxvmvGtzr99y5/VdETo31P7JRM6iWsaEMQ0TSge444R68sE1VO7rbObcCNwY8zweKVbW9iLQH/hvYgaoOdkOwnKmq20TkQ+BWVb3AjVH0PHCRqm51t6XG4EQHADhUVfNc2V8AnlDV90WkGU54GV/guBOAM3HyM6wWkYk4ORnaqhNozhdh1ccPwNmqussNdfNvoExyKcPwgikLo7pRV5xQzs2B5TjhnL3gCzK3HLg0yPMzgCcBVPUzEfksSrlaAW2BBU7IItJxwmr48N+m6gG0lgMpCBqISH33+g1V3Q3sFpEfgCMijFsb+LuI5AAlwPFRym0YgCkLo/qxU1VzROQQ4HUcm8WTOCHc/bdd6wS8t9v9WULov4vKxMYRYKWqdg3x/De/6zSgq6ruLNOBozx2+1WFk9XHn3FibXVw+90VhcyGUYrZLIxqiapux0mbe6u7BbQR59v6Qa4i6R5ll+/iJMtCRNriBHeLhtXAYSLS1e2jtoi0CdH2LQ7kWcFdFYRjB862VDAOAb5T1f3AtTgrGsOIGlMWRrVFVT/ByXvdV1U34+Qs/gwnSdQnUXY3ETjY3X66HSeaZzSy7MEJF/2wiHyKEwE1lLH5T0Cua0xfBQyO0HcR8IF7XHhcwOMJQH8RWYazBfVbuQ4MwwMWddYwDMOIiK0sDMMwjIiYsjAMwzAiYsrCMAzDiIgpC8MwDCMipiwMwzCMiJiyMAzDMCJiysIwDMOIyP8Hl0TBDT3ed8IAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visualize\n",
    "x_ax = np.array(range(int(np.min(train_run_diff)), int(np.max(train_run_diff)))).reshape(-1, 1)\n",
    "y_ax = linear_reg.coef_ * x_ax + linear_reg.intercept_\n",
    "plt.plot(train_run_diff, train_win_pct, 'bo', label=\"training_data\")\n",
    "plt.plot(x_ax, y_ax, 'r', label=\"model_fit\")\n",
    "plt.plot([-300, 300], [0.5, 0.5], \"k--\")\n",
    "plt.plot([0, 0], [0.30, 0.65], \"k--\")\n",
    "plt.ylim([0.30, 0.65])\n",
    "plt.xlabel(\"Run differential\")\n",
    "plt.ylabel(\"Win percentage\")\n",
    "plt.legend(loc=\"lower right\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Construct test dataset\n",
    "log_file = \"GL2016.TXT\"\n",
    "df = pd.read_table(log_file, sep=\",\", header=None)\n",
    "df_proc = extract_linear_reg_inputs(df)\n",
    "test_run_diff = df_proc['RD'].values.reshape([-1, 1])\n",
    "test_win_pct = df_proc['Win Pct'].values.reshape([-1, 1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Predict outcomes using regression model\n",
    "predict_win_pct = linear_reg.predict(test_run_diff)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Percentage error on test set =  2.096802895798953 %\n",
      "Percentage error on training set  2.0550910528614867 %\n"
     ]
    }
   ],
   "source": [
    "# Compute percentage error for linear regression model on test set\n",
    "mean_abs_error_test = np.mean(np.abs(predict_win_pct - test_win_pct))\n",
    "print(\"Percentage error on test set = \", 100. * mean_abs_error_test, \"%\")\n",
    "\n",
    "# Compute percentage error for linear regression model on training set\n",
    "model_fit_train = linear_reg.predict(train_run_diff)\n",
    "mean_abs_error_training = np.mean(np.abs(model_fit_train - train_win_pct))\n",
    "print(\"Percentage error on training set \", 100. * mean_abs_error_training, \"%\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visualize\n",
    "plt.plot(test_win_pct, predict_win_pct, 'bo')\n",
    "plt.plot([0.35, 0.7], [0.35, 0.7], 'r')\n",
    "plt.xlabel(\"Actual win percentage\")\n",
    "plt.ylabel(\"Predicted win percentage\")\n",
    "plt.title(\"MLB 2016 season\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Tutorial: https://learning.oreilly.com/oriole/analyzing-baseball-statistics-with-python-part-1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}