DL_project_troesch_stein_tri_ranking_ClassificationModels_2020-04-11_V1.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "### Import Base Packages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import pandas as pd\n",
    "import os\n",
    "from pathlib import Path\n",
    "import chardet\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Define Working Directory and get Filesize"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2099333"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_folder = Path(r\"C:\\Users\\finns\\Documents\\CAS Machine Intelligence_Cdrive\\Deep Learning\\Projekt\")\n",
    "os.chdir(data_folder)\n",
    "os.path.getsize('tri_Data.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Determine Encoding of File"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ISO-8859-1\n"
     ]
    }
   ],
   "source": [
    "data_folder = Path(r\"C:\\Users\\finns\\Documents\\CAS Machine Intelligence_Cdrive\\Deep Learning\\Projekt\")\n",
    "os.chdir(data_folder)\n",
    "\n",
    "rawdata = open(\"tri_Data.csv\", 'rb').read(500000) #Judge format according to first 500k bytes\n",
    "result = chardet.detect(rawdata)\n",
    "charenc = result['encoding']\n",
    "print(charenc) #FileEncoding is ISO-8859-1 (Latin-1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Import File and Recode Variable Classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 30319 entries, 0 to 30318\n",
      "Data columns (total 12 columns):\n",
      " #   Column     Non-Null Count  Dtype  \n",
      "---  ------     --------------  -----  \n",
      " 0   subj_id    30319 non-null  int64  \n",
      " 1   yob        29168 non-null  float64\n",
      " 2   age_y      20926 non-null  float64\n",
      " 3   nat        30319 non-null  object \n",
      " 4   gender     30319 non-null  object \n",
      " 5   race_id    30319 non-null  int64  \n",
      " 6   race_inf   30319 non-null  object \n",
      " 7   race_date  21761 non-null  object \n",
      " 8   race_y     21761 non-null  float64\n",
      " 9   race_m     21761 non-null  float64\n",
      " 10  time       25975 non-null  object \n",
      " 11  rank       30319 non-null  object \n",
      "dtypes: float64(4), int64(2), object(6)\n",
      "memory usage: 2.8+ MB\n"
     ]
    }
   ],
   "source": [
    "df = pd.read_csv(\"tri_Data.csv\", sep = \";\", header = 0, encoding=charenc)\n",
    "df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['race_date'] = pd.to_datetime(df['race_date'])\n",
    "df['subj_id'] = df['subj_id'].astype(str)\n",
    "df['race_id'] = df['race_id'].astype(str)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Get Overview of Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "scrolled": true
   },
   "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>subj_id</th>\n",
       "      <th>yob</th>\n",
       "      <th>age_y</th>\n",
       "      <th>nat</th>\n",
       "      <th>gender</th>\n",
       "      <th>race_id</th>\n",
       "      <th>race_inf</th>\n",
       "      <th>race_date</th>\n",
       "      <th>race_y</th>\n",
       "      <th>race_m</th>\n",
       "      <th>time</th>\n",
       "      <th>rank</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>19916</th>\n",
       "      <td>1737</td>\n",
       "      <td>1981.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>ENG</td>\n",
       "      <td>Women</td>\n",
       "      <td>1141</td>\n",
       "      <td>Madrid 2009</td>\n",
       "      <td>NaT</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>DNF</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5710</th>\n",
       "      <td>853</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>ITA</td>\n",
       "      <td>Men</td>\n",
       "      <td>314</td>\n",
       "      <td>Karlovy Vary 2016</td>\n",
       "      <td>2016-08-28</td>\n",
       "      <td>2016.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>LAP</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23107</th>\n",
       "      <td>1038</td>\n",
       "      <td>1983.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>LUX</td>\n",
       "      <td>Women</td>\n",
       "      <td>1072</td>\n",
       "      <td>Kitzbühel 2007</td>\n",
       "      <td>2007-07-22</td>\n",
       "      <td>2007.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>01:58:16</td>\n",
       "      <td>33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22978</th>\n",
       "      <td>2249</td>\n",
       "      <td>1984.0</td>\n",
       "      <td>23.0</td>\n",
       "      <td>NED</td>\n",
       "      <td>Men</td>\n",
       "      <td>552</td>\n",
       "      <td>Salford 2007</td>\n",
       "      <td>2007-07-29</td>\n",
       "      <td>2007.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>DNS</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22808</th>\n",
       "      <td>1874</td>\n",
       "      <td>1979.0</td>\n",
       "      <td>28.0</td>\n",
       "      <td>BRA</td>\n",
       "      <td>Men</td>\n",
       "      <td>653</td>\n",
       "      <td>Tiszaújváros 2007</td>\n",
       "      <td>2007-11-08</td>\n",
       "      <td>2007.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>01:49:50</td>\n",
       "      <td>24</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      subj_id     yob  age_y  nat gender race_id            race_inf  \\\n",
       "19916    1737  1981.0    NaN  ENG  Women    1141         Madrid 2009   \n",
       "5710      853     NaN    NaN  ITA    Men     314   Karlovy Vary 2016   \n",
       "23107    1038  1983.0   24.0  LUX  Women    1072      Kitzbühel 2007   \n",
       "22978    2249  1984.0   23.0  NED    Men     552        Salford 2007   \n",
       "22808    1874  1979.0   28.0  BRA    Men     653   Tiszaújváros 2007   \n",
       "\n",
       "       race_date  race_y  race_m      time rank  \n",
       "19916        NaT     NaN     NaN       NaN  DNF  \n",
       "5710  2016-08-28  2016.0     8.0       NaN  LAP  \n",
       "23107 2007-07-22  2007.0     7.0  01:58:16   33  \n",
       "22978 2007-07-29  2007.0     7.0       NaN  DNS  \n",
       "22808 2007-11-08  2007.0     8.0  01:49:50   24  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.sample(5)"
   ]
  },
  {
   "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>AnyNAN</th>\n",
       "      <th>SumOfNAN</th>\n",
       "      <th>UniqueEntries</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>subj_id</th>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>3903</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>yob</th>\n",
       "      <td>True</td>\n",
       "      <td>1151</td>\n",
       "      <td>45</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>age_y</th>\n",
       "      <td>True</td>\n",
       "      <td>9393</td>\n",
       "      <td>33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>nat</th>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>114</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>gender</th>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>race_id</th>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>1489</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>race_inf</th>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>745</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>race_date</th>\n",
       "      <td>True</td>\n",
       "      <td>8558</td>\n",
       "      <td>504</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>race_y</th>\n",
       "      <td>True</td>\n",
       "      <td>8558</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>race_m</th>\n",
       "      <td>True</td>\n",
       "      <td>8558</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>time</th>\n",
       "      <td>True</td>\n",
       "      <td>4344</td>\n",
       "      <td>3013</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rank</th>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>75</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           AnyNAN  SumOfNAN  UniqueEntries\n",
       "subj_id     False         0           3903\n",
       "yob          True      1151             45\n",
       "age_y        True      9393             33\n",
       "nat         False         0            114\n",
       "gender      False         0              2\n",
       "race_id     False         0           1489\n",
       "race_inf    False         0            745\n",
       "race_date    True      8558            504\n",
       "race_y       True      8558             32\n",
       "race_m       True      8558              8\n",
       "time         True      4344           3013\n",
       "rank        False         0             75"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame(\n",
    "    {'AnyNAN':df.isnull().any(),\n",
    "     'SumOfNAN':df.isnull().sum(axis = 0),\n",
    "     'UniqueEntries': df.nunique()\n",
    "    })"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For lots of Races there are only the first 10 Ranks registered. Furthermore, there are lots of \"special rankings\" like \"DNF => Did not Finish\" registered. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df[\"rank\"].value_counts().plot.bar(figsize=(20,4))\n",
    "plt.show()\n",
    "#DNF = Did not Finish\n",
    "#DNS = Did not Start\n",
    "#DSQ = Disqualified\n",
    "#LAP = Ueberrundet"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Below plot shows, that there are only few Persons (subj_ids), that are registered for more than 10 Triathlon Contests."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2455fa8bac8>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "subjRaceNum = df[\"subj_id\"].value_counts().sort_values(ascending=True)\n",
    "subjRaceNum.plot.hist(bins=100)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Below Plot shows Age Distribution of Triathlon participants is right skewed. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x245600d64a8>"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "# df['age_y'].value_counts().plot.bar(figsize=(20,4))\n",
    "binCount = df['age_y'].max() - df['age_y'].min()\n",
    "df['age_y'].plot.hist(bins=int(binCount), alpha=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Below Histogram shows, that most races have 10 Entries or less. Only for few races there are more than 10 entries in the triathlon data. Probably there were only Top 10 Participants registered for many races."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x24560197588>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "#Checking number of participants per Race, there seem to be lots fo races with 10 or less participants\n",
    "#Probably the collection of results isn't complete\n",
    "entriesPerRace = df.groupby(['race_id'])['subj_id'].count()\n",
    "entriesPerRace.plot.hist(bins=50, alpha=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As there are some races, that have even less entries than 10, I want to check a couple of exmpales. Below table shows, that for some races indeed entries aren't complete. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "scrolled": true
   },
   "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>subj_id</th>\n",
       "      <th>yob</th>\n",
       "      <th>age_y</th>\n",
       "      <th>nat</th>\n",
       "      <th>gender</th>\n",
       "      <th>race_id</th>\n",
       "      <th>race_inf</th>\n",
       "      <th>race_date</th>\n",
       "      <th>race_y</th>\n",
       "      <th>race_m</th>\n",
       "      <th>time</th>\n",
       "      <th>rank</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>10382</th>\n",
       "      <td>966</td>\n",
       "      <td>1994.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>FRA</td>\n",
       "      <td>Men</td>\n",
       "      <td>35</td>\n",
       "      <td>Antalya 2014</td>\n",
       "      <td>NaT</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>01:46:52</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10383</th>\n",
       "      <td>1548</td>\n",
       "      <td>1986.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>RUS</td>\n",
       "      <td>Men</td>\n",
       "      <td>35</td>\n",
       "      <td>Antalya 2014</td>\n",
       "      <td>NaT</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>01:47:17</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10384</th>\n",
       "      <td>723</td>\n",
       "      <td>1986.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>GER</td>\n",
       "      <td>Men</td>\n",
       "      <td>35</td>\n",
       "      <td>Antalya 2014</td>\n",
       "      <td>NaT</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>01:48:50</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      subj_id     yob  age_y  nat gender race_id       race_inf race_date  \\\n",
       "10382     966  1994.0    NaN  FRA    Men      35   Antalya 2014       NaT   \n",
       "10383    1548  1986.0    NaN  RUS    Men      35   Antalya 2014       NaT   \n",
       "10384     723  1986.0    NaN  GER    Men      35   Antalya 2014       NaT   \n",
       "\n",
       "       race_y  race_m      time rank  \n",
       "10382     NaN     NaN  01:46:52    2  \n",
       "10383     NaN     NaN  01:47:17    3  \n",
       "10384     NaN     NaN  01:48:50   10  "
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[df['race_id']==\"35\"]\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Data Cleanup and Recoding for Variable \"rank\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Later on I want to try to forcast rank as categorical as well as a continuous variable. In case of rank as a continuous variable I will need to standardize the \"rank\" variable, because (as shown above) number of registered entries per race are vastly different. Furthermore, I want to recode all special rank entries to \"last rank\" of the respective number of entries per competition. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "#create variable with number of participants for each race_id\n",
    "entries = df.groupby([\"race_id\"])[\"subj_id\"].transform(\"count\")\n",
    "df['entries'] = entries\n",
    "\n",
    "# Create 2 Column Table that shows the lowest rank entry for each race Id\n",
    "idx = df['rank'].isin(['DNF','DNS','LAP','DSO','DNF/LAP','DSQ'])\n",
    "dfRaceAndRank = df.loc[-idx,['race_id','rank']]\n",
    "dfRaceAndRank['rank'] = pd.to_numeric(dfRaceAndRank['rank']) \n",
    "maxRank = dfRaceAndRank.groupby([\"race_id\"])[\"rank\"].max()\n",
    "\n",
    "#Create Dictionary for vlookup\n",
    "maxRankDict = dict(zip(maxRank.index,maxRank.values))\n",
    "\n",
    "#Create variable with maximum rank entry for each line of df\n",
    "df['maxRank']= df['race_id'].map(maxRankDict)\n",
    "\n",
    "# Some race_ids have less entries than there are ranks. In those cases fill in max rank as \"number of particpants\"\n",
    "df['participants'] = np.where(df['maxRank'] > df['entries'], df['maxRank'], df['entries'])\n",
    "\n",
    "#Replace non-numeric ranks with \"last rank\" according to number of participants\n",
    "df['rank'] = np.where(df['rank'].isin(['DNF','DNS','LAP','DSO','DNF/LAP','DSQ']), df['participants'], df['rank'])\n",
    "\n",
    "#Change Variable \"Rank\" to numeric Variable\n",
    "df['rank'] = df[['rank']].apply(pd.to_numeric)\n",
    "\n",
    "#Create Variable with normalized Rank\n",
    "df['rankNorm'] = df['rank']/df['participants']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x24560282cc0>"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAD4CAYAAAAdIcpQAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAATf0lEQVR4nO3df7Bc9Xnf8ffHAhs5MTFEgqgSjnBGdSw8wQaZauqktU1aZNxEuBMapWnQeEiUYtI4k87UwpOJ0+lohsy0jksTcBTXg3DqMEr8AyUOcRQlttsJNr4k2EJgimooqNIgmbSF2C5U8tM/9ihdi9U9R3DP3t2779fMzp7z7Dm7z1c/9rPnx55NVSFJ0nxestgNSJImn2EhSWplWEiSWhkWkqRWhoUkqdVZi91AX1asWFFr165d7DYkaarcd999X6uqlafWl2xYrF27lrm5ucVuQ5KmSpL/PqrubihJUivDQpLUyrCQJLUyLCRJrQwLSVIrw0KS1MqwkCS1MiwkSa0MC0lSqyX7DW5JWsrWbv/UyPpjN7+9l9dzy0KS1MqwkCS1MiwkSa0MC0lSK8NCktTKsJAktTIsJEmtDAtJUivDQpLUyrCQJLUyLCRJrQwLSVKrXsMiyWNJ9ie5P8lcUzs/yd4kjzT35w0tf1OSg0keTnLVUP3y5nkOJrklSfrsW5L07caxZfGWqnp9VW1o5rcD+6pqHbCvmSfJemALcAmwCbg1ybJmnduAbcC65rZpDH1LkhqLsRtqM7Crmd4FXDNUv7Oqnq2qR4GDwBVJVgHnVtU9VVXAHUPrSJLGoO+wKOCPk9yXZFtTu7CqjgA09xc09dXAE0PrHmpqq5vpU+vPk2Rbkrkkc8eOHVvAYUjSbOv7x4/eVFWHk1wA7E3ylXmWHXUcouapP79YtRPYCbBhw4aRy0iSzlyvWxZVdbi5Pwp8ArgCeLLZtURzf7RZ/BBw0dDqa4DDTX3NiLokaUx6C4sk35HkFSengX8IPADsAbY2i20F7mqm9wBbkrwsycUMDmTf2+yqeibJxuYsqOuG1pEkjUGfu6EuBD7RnOV6FvDRqvqjJF8Edie5HngcuBagqg4k2Q08CBwHbqyqE81z3QDcDiwH7m5ukqQx6S0squqrwKUj6k8BV55mnR3AjhH1OeB1C92jJKkbv8EtSWplWEiSWhkWkqRWhoUkqZVhIUlqZVhIkloZFpKkVoaFJKmVYSFJamVYSJJaGRaSpFaGhSSplWEhSWplWEiSWhkWkqRWhoUkqZVhIUlqZVhIkloZFpKkVoaFJKmVYSFJamVYSJJaGRaSpFaGhSSplWEhSWplWEiSWhkWkqRWhoUkqZVhIUlq1XtYJFmW5C+T/EEzf36SvUkeae7PG1r2piQHkzyc5Kqh+uVJ9jeP3ZIkffctSfr/xrFl8W7goaH57cC+qloH7GvmSbIe2AJcAmwCbk2yrFnnNmAbsK65bRpD35KkRq9hkWQN8HbgQ0PlzcCuZnoXcM1Q/c6qeraqHgUOAlckWQWcW1X3VFUBdwytI0kag763LD4A/CvgW0O1C6vqCEBzf0FTXw08MbTcoaa2upk+tf48SbYlmUsyd+zYsYUZgSSpv7BI8o+Ao1V1X9dVRtRqnvrzi1U7q2pDVW1YuXJlx5eVJLU5q8fnfhPwo0muBs4Bzk3y28CTSVZV1ZFmF9PRZvlDwEVD668BDjf1NSPqkqQx6W3Loqpuqqo1VbWWwYHrP62qfwbsAbY2i20F7mqm9wBbkrwsycUMDmTf2+yqeibJxuYsqOuG1pEkjUGfWxanczOwO8n1wOPAtQBVdSDJbuBB4DhwY1WdaNa5AbgdWA7c3dwkSWMylrCoqs8An2mmnwKuPM1yO4AdI+pzwOv661CSNB+/wS1JamVYSJJaGRaSpFaGhSSplWEhSWplWEiSWhkWkqRWhoUkqZVhIUlqZVhIkloZFpKkVoaFJKmVYSFJamVYSJJadQqLJF4eXJJmWNctiw8muTfJu5K8steOJEkTp1NYVNUPAj/J4Dey55J8NMk/6LUzSdLE6HzMoqoeAX4JeA/w94FbknwlyT/uqzlJ0mToesziB5L8GvAQ8FbgR6rqtc30r/XYnyRpAnT9De5fB34LeG9VffNksaoOJ/mlXjqTJE2MrmFxNfDNqjoBkOQlwDlV9Y2q+khv3UmSJkLXYxZ/Aiwfmn95U5MkzYCuYXFOVf31yZlm+uX9tCRJmjRdw+LrSS47OZPkcuCb8ywvSVpCuh6z+AXgd5McbuZXAT/eT0uSpEnTKSyq6otJvh94DRDgK1X1f3vtTJI0MbpuWQC8EVjbrPOGJFTVHb10JUmaKJ3CIslHgO8D7gdONOUCDAtJmgFdtyw2AOurqvpsRpI0mbqeDfUA8D1n8sRJzmmuVPulJAeS/Oumfn6SvUkeae7PG1rnpiQHkzyc5Kqh+uVJ9jeP3ZIkZ9KLJOnF6RoWK4AHk3w6yZ6Tt5Z1ngXeWlWXAq8HNiXZCGwH9lXVOmBfM0+S9cAW4BJgE3BrkmXNc90GbAPWNbdNnUcoSXrRuu6G+pUzfeJml9XJL/Kd3dwK2Ay8uanvAj7D4Eq2m4E7q+pZ4NEkB4ErkjwGnFtV9wAkuQO4Brj7THuSJL0wXX/P4rPAY8DZzfQXgb9oWy/JsiT3A0eBvVX1BeDCqjrSPO8R4IJm8dXAE0OrH2pqq5vpU+ujXm9bkrkkc8eOHesyNElSB10vUf4zwO8Bv9mUVgOfbFuvqk5U1euBNQy2Eub7edZRxyFqnvqo19tZVRuqasPKlSvb2pMkddT1mMWNwJuAp+FvfgjpgnnXGFJV/4vB7qZNwJNJVgE090ebxQ4x+CW+k9YAh5v6mhF1SdKYdA2LZ6vquZMzSc7iNJ/uh5ZZefL3upMsB34Y+AqwB9jaLLYVuKuZ3gNsSfKyJBczOJB9b7Or6pkkG5uzoK4bWkeSNAZdD3B/Nsl7geXNb2+/C/j9lnVWAbuaM5peAuyuqj9Icg+wO8n1wOPAtQBVdSDJbuBB4Dhw48nfzwBuAG5ncJn0u/HgtiSNVdew2A5cD+wHfhb4Q+BD861QVV8G3jCi/hRw5WnW2QHsGFGfA+Y73iFJ6lHXCwl+i8HPqv5Wv+1IkiZR12tDPcqIYxRV9eoF70iSNHHO5NpQJ53D4DjD+QvfjiRpEnX9Ut5TQ7f/UVUfAN7ac2+SpAnRdTfUZUOzL2GwpfGKXjqSJE2crruh/t3Q9HEGl/74JwvejSRpInU9G+otfTciSZpcXXdD/eJ8j1fV+xemHUnSJDqTs6HeyOCSHAA/AnyOb79KrCRpieoaFiuAy6rqGYAkvwL8blX9dF+NSZImR9cLCb4KeG5o/jlg7YJ3I0maSF23LD4C3JvkEwy+yf0O4I7eupIkTZSuZ0PtSHI38ENN6Z1V9Zf9tSVJmiRdd0MBvBx4uqr+PXCo+c0JSdIM6Pqzqu8D3gPc1JTOBn67r6YkSZOl65bFO4AfBb4OUFWH8XIfkjQzuobFc1VVNJcpT/Id/bUkSZo0XcNid5LfBF6Z5GeAP8EfQpKkmdH1bKh/2/z29tPAa4Bfrqq9vXYmSZoYrWGRZBnw6ar6YcCAkKQZ1LobqqpOAN9I8l1j6EeSNIG6foP7/wD7k+ylOSMKoKp+vpeuJEkTpWtYfKq5SZJm0LxhkeRVVfV4Ve0aV0OSpMnTdszikycnknys514kSROqLSwyNP3qPhuRJE2utrCo00xLkmZI2wHuS5M8zWALY3kzTTNfVXVur91JkibCvGFRVcvG1YgkaXKdye9ZnJEkFyX5syQPJTmQ5N1N/fwke5M80tyfN7TOTUkOJnk4yVVD9cuT7G8euyVJRr2mJKkfvYUFcBz4l1X1WmAjcGOS9cB2YF9VrQP2NfM0j20BLgE2Abc2lxoBuA3YBqxrbpt67FuSdIrewqKqjlTVXzTTzwAPAauBzcDJ723sAq5ppjcDd1bVs1X1KHAQuCLJKuDcqrqnuUz6HUPrSJLGoM8ti7+RZC3wBuALwIVVdQQGgQJc0Cy2GnhiaLVDTW11M31qXZI0Jr2HRZLvBD4G/EJVPT3foiNqNU991GttSzKXZO7YsWNn3qwkaaRewyLJ2QyC4j9V1ceb8pPNriWa+6NN/RBw0dDqa4DDTX3NiPrzVNXOqtpQVRtWrly5cAORpBnX59lQAf4j8FBVvX/ooT3A1mZ6K3DXUH1LkpcluZjBgex7m11VzyTZ2DzndUPrSJLGoOtVZ1+INwE/xeDS5vc3tfcCNzP4mdbrgceBawGq6kCS3cCDDM6kurH5LQ2AG4DbgeXA3c1NkjQmvYVFVf0XRh9vALjyNOvsAHaMqM8Br1u47iRJZ2IsZ0NJkqabYSFJamVYSJJaGRaSpFaGhSSplWEhSWplWEiSWhkWkqRWhoUkqZVhIUlqZVhIkloZFpKkVoaFJKmVYSFJamVYSJJaGRaSpFaGhSSplWEhSWplWEiSWhkWkqRWhoUkqZVhIUlqZVhIkloZFpKkVoaFJKmVYSFJamVYSJJaGRaSpFaGhSSpVW9hkeTDSY4meWCodn6SvUkeae7PG3rspiQHkzyc5Kqh+uVJ9jeP3ZIkffUsSRqtzy2L24FNp9S2A/uqah2wr5knyXpgC3BJs86tSZY169wGbAPWNbdTn1OS1LPewqKqPgf81SnlzcCuZnoXcM1Q/c6qeraqHgUOAlckWQWcW1X3VFUBdwytI0kak3Efs7iwqo4ANPcXNPXVwBNDyx1qaqub6VPrkqQxmpQD3KOOQ9Q89dFPkmxLMpdk7tixYwvWnCTNurPG/HpPJllVVUeaXUxHm/oh4KKh5dYAh5v6mhH1kapqJ7ATYMOGDacNlcW2dvunRtYfu/ntS/q1Z81i/Vmf7nXH8dpausYdFnuArcDNzf1dQ/WPJnk/8LcYHMi+t6pOJHkmyUbgC8B1wH8Yc89aAAakNN16C4skvwO8GViR5BDwPgYhsTvJ9cDjwLUAVXUgyW7gQeA4cGNVnWie6gYGZ1YtB+5ubpIm0KRtTfmBYOH0FhZV9ROneejK0yy/A9gxoj4HvG4BW5MknaFJOcAtSZpghoUkqdW4D3BL0pIyK8dL3LKQJLUyLCRJrQwLSVIrw0KS1MqwkCS1MiwkSa0MC0lSK8NCktTKsJAktTIsJEmtvNzHBJmm33zou9dp+rPo26T1A/P/wJKWJsNCC2qx3kR885L6ZVhMgRfyRrjULmKm51vIgJzErRdNFsNihKXwKXVaxjAtfcJ09TotFmr3p/pnWGiqjOPTdN9m8VP8mf5ZGwqTx7DQvPxPOz7j+LP2TVsvlKfOSpJauWUhaclayC2jpbybsAu3LCRJrdyykKQOZv34jVsWkqRWbllI0hhN66nThoUk9WCp7bZyN5QkqZVbFpI0ASZ9S8QtC0lSK8NCktRqasIiyaYkDyc5mGT7YvcjSbNkKsIiyTLgN4C3AeuBn0iyfnG7kqTZMRVhAVwBHKyqr1bVc8CdwOZF7kmSZsa0nA21GnhiaP4Q8HdOXSjJNmBbM/vXSR7u+PwrgK+9qA6nk+OeHbM4ZpjBcedXX/SYv3dUcVrCIiNq9bxC1U5g5xk/eTJXVRteSGPTzHHPjlkcM8zmuPsa87TshjoEXDQ0vwY4vEi9SNLMmZaw+CKwLsnFSV4KbAH2LHJPkjQzpmI3VFUdT/JzwKeBZcCHq+rAAr7EGe+6WiIc9+yYxTHDbI67lzGn6nm7/iVJ+jbTshtKkrSIDAtJUquZCou2S4Zk4Jbm8S8nuWwx+lxoHcb9k814v5zkz5Ncuhh9LqSul4dJ8sYkJ5L82Dj760uXcSd5c5L7kxxI8tlx97jQOvz7/q4kv5/kS82Y37kYfS6kJB9OcjTJA6d5fOHfy6pqJm4MDoz/N+DVwEuBLwHrT1nmauBuBt/r2Ah8YbH7HtO4/y5wXjP9tmkfd5cxDy33p8AfAj+22H2P6e/6lcCDwKua+QsWu+8xjPm9wK820yuBvwJeuti9v8hx/z3gMuCB0zy+4O9ls7Rl0eWSIZuBO2rg88Ark6wad6MLrHXcVfXnVfU/m9nPM/geyzTrenmYfwF8DDg6zuZ61GXc/xT4eFU9DlBV0z72LmMu4BVJAnwng7A4Pt42F1ZVfY7BOE5nwd/LZiksRl0yZPULWGbanOmYrmfwiWSatY45yWrgHcAHx9hX37r8Xf9t4Lwkn0lyX5LrxtZdP7qM+deB1zL4Iu9+4N1V9a3xtLdoFvy9bCq+Z7FAulwypNNlRaZM5zEleQuDsPjBXjvqX5cxfwB4T1WdGHzgXBK6jPss4HLgSmA5cE+Sz1fVf+27uZ50GfNVwP3AW4HvA/Ym+c9V9XTfzS2iBX8vm6Ww6HLJkKV4WZFOY0ryA8CHgLdV1VNj6q0vXca8AbizCYoVwNVJjlfVJ8fTYi+6/hv/WlV9Hfh6ks8BlwLTGhZdxvxO4OYa7Mw/mORR4PuBe8fT4qJY8PeyWdoN1eWSIXuA65ozCTYC/7uqjoy70QXWOu4krwI+DvzUFH/CHNY65qq6uKrWVtVa4PeAd015UEC3f+N3AT+U5KwkL2dw9eaHxtznQuoy5scZbEmR5ELgNcBXx9rl+C34e9nMbFnUaS4ZkuSfN49/kMFZMVcDB4FvMPhEMtU6jvuXge8Gbm0+aR+vKb5SZ8cxLzldxl1VDyX5I+DLwLeAD1XVyNMvp0HHv+t/A9yeZD+D3TPvqaqpvmx5kt8B3gysSHIIeB9wNvT3XublPiRJrWZpN5Qk6QUyLCRJrQwLSVIrw0KS1MqwkCS1MiwkSa0MC0lSq/8HINbpvVYHPfoAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df['rankNorm'].plot.hist(bins=50, alpha=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Data Cleanup and Recoding for Variables \"race_y\" and \"race_inf\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The variables \"race_y\" and \"age_y\" that I want to use for my models have lots of nan's"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "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>AnyNAN</th>\n",
       "      <th>SumOfNAN</th>\n",
       "      <th>UniqueEntries</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>subj_id</th>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>3903</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>yob</th>\n",
       "      <td>True</td>\n",
       "      <td>1151</td>\n",
       "      <td>45</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>age_y</th>\n",
       "      <td>True</td>\n",
       "      <td>9393</td>\n",
       "      <td>33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>nat</th>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>114</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>gender</th>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>race_id</th>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>1489</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>race_inf</th>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>745</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>race_date</th>\n",
       "      <td>True</td>\n",
       "      <td>8558</td>\n",
       "      <td>504</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>race_y</th>\n",
       "      <td>True</td>\n",
       "      <td>8558</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>race_m</th>\n",
       "      <td>True</td>\n",
       "      <td>8558</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>time</th>\n",
       "      <td>True</td>\n",
       "      <td>4344</td>\n",
       "      <td>3013</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rank</th>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>82</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>entries</th>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>82</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>maxRank</th>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>70</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>participants</th>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>82</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rankNorm</th>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>1920</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              AnyNAN  SumOfNAN  UniqueEntries\n",
       "subj_id        False         0           3903\n",
       "yob             True      1151             45\n",
       "age_y           True      9393             33\n",
       "nat            False         0            114\n",
       "gender         False         0              2\n",
       "race_id        False         0           1489\n",
       "race_inf       False         0            745\n",
       "race_date       True      8558            504\n",
       "race_y          True      8558             32\n",
       "race_m          True      8558              8\n",
       "time            True      4344           3013\n",
       "rank           False         0             82\n",
       "entries        False         0             82\n",
       "maxRank         True         1             70\n",
       "participants   False         0             82\n",
       "rankNorm       False         0           1920"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame(\n",
    "    {'AnyNAN':df.isnull().any(),\n",
    "     'SumOfNAN':df.isnull().sum(axis = 0),\n",
    "     'UniqueEntries': df.nunique()\n",
    "    })"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Looking at some examples where \"race_y\" is missing, you can see that the missing race dates can often be parsed from the race name (race_inf). Which I will do in the second chunk below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "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>subj_id</th>\n",
       "      <th>yob</th>\n",
       "      <th>age_y</th>\n",
       "      <th>nat</th>\n",
       "      <th>gender</th>\n",
       "      <th>race_id</th>\n",
       "      <th>race_inf</th>\n",
       "      <th>race_date</th>\n",
       "      <th>race_y</th>\n",
       "      <th>race_m</th>\n",
       "      <th>time</th>\n",
       "      <th>rank</th>\n",
       "      <th>entries</th>\n",
       "      <th>maxRank</th>\n",
       "      <th>participants</th>\n",
       "      <th>rankNorm</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3815</th>\n",
       "      <td>3492</td>\n",
       "      <td>1996.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>USA</td>\n",
       "      <td>Women</td>\n",
       "      <td>1327</td>\n",
       "      <td>Sarasota, FL 2017</td>\n",
       "      <td>NaT</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>02:16:00</td>\n",
       "      <td>16.0</td>\n",
       "      <td>30</td>\n",
       "      <td>20.0</td>\n",
       "      <td>30.0</td>\n",
       "      <td>0.533333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>357</th>\n",
       "      <td>64</td>\n",
       "      <td>1983.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>JPN</td>\n",
       "      <td>Women</td>\n",
       "      <td>1180</td>\n",
       "      <td>Miyazaki 2019</td>\n",
       "      <td>NaT</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>01:59:32</td>\n",
       "      <td>1.0</td>\n",
       "      <td>49</td>\n",
       "      <td>41.0</td>\n",
       "      <td>49.0</td>\n",
       "      <td>0.020408</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3593</th>\n",
       "      <td>2878</td>\n",
       "      <td>1992.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>CRC</td>\n",
       "      <td>Men</td>\n",
       "      <td>507</td>\n",
       "      <td>Playa Hermosa 2018</td>\n",
       "      <td>NaT</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>02:14:42</td>\n",
       "      <td>9.0</td>\n",
       "      <td>17</td>\n",
       "      <td>13.0</td>\n",
       "      <td>17.0</td>\n",
       "      <td>0.529412</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22298</th>\n",
       "      <td>246</td>\n",
       "      <td>1971.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>ENG</td>\n",
       "      <td>Women</td>\n",
       "      <td>895</td>\n",
       "      <td>Eilat 2007</td>\n",
       "      <td>NaT</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>48.0</td>\n",
       "      <td>34</td>\n",
       "      <td>48.0</td>\n",
       "      <td>48.0</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5288</th>\n",
       "      <td>1736</td>\n",
       "      <td>1989.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>ENG</td>\n",
       "      <td>Women</td>\n",
       "      <td>747</td>\n",
       "      <td>Abu Dhabi 2017</td>\n",
       "      <td>NaT</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>02:03:46</td>\n",
       "      <td>2.0</td>\n",
       "      <td>30</td>\n",
       "      <td>27.0</td>\n",
       "      <td>30.0</td>\n",
       "      <td>0.066667</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      subj_id     yob  age_y  nat gender race_id             race_inf  \\\n",
       "3815     3492  1996.0    NaN  USA  Women    1327    Sarasota, FL 2017   \n",
       "357        64  1983.0    NaN  JPN  Women    1180        Miyazaki 2019   \n",
       "3593     2878  1992.0    NaN  CRC    Men     507   Playa Hermosa 2018   \n",
       "22298     246  1971.0    NaN  ENG  Women     895           Eilat 2007   \n",
       "5288     1736  1989.0    NaN  ENG  Women     747       Abu Dhabi 2017   \n",
       "\n",
       "      race_date  race_y  race_m      time  rank  entries  maxRank  \\\n",
       "3815        NaT     NaN     NaN  02:16:00  16.0       30     20.0   \n",
       "357         NaT     NaN     NaN  01:59:32   1.0       49     41.0   \n",
       "3593        NaT     NaN     NaN  02:14:42   9.0       17     13.0   \n",
       "22298       NaT     NaN     NaN       NaN  48.0       34     48.0   \n",
       "5288        NaT     NaN     NaN  02:03:46   2.0       30     27.0   \n",
       "\n",
       "       participants  rankNorm  \n",
       "3815           30.0  0.533333  \n",
       "357            49.0  0.020408  \n",
       "3593           17.0  0.529412  \n",
       "22298          48.0  1.000000  \n",
       "5288           30.0  0.066667  "
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[df['race_y'].isnull()].sample(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['race_y1'] = df['race_inf'].str.extract(r'(\\d{4})', expand = True) #Extract Year from Event Name to variable race_y1\n",
    "df['race_y'] = df['race_y'].fillna(df['race_y1'].astype(float)) #Where race_y is NA, fill in value from race_y1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Furthermore, I want to store the races themselves as a variable, but without the information for the year. Miyazaki 2019 should be the same as Miyazaki 2018. I will also delete some additional comments in the names that for example mention the US-State behind a comma. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['race_inf'] = df['race_inf'].str.split(',').str[0] #Delete state abbreviations etc in event name (behind comma)\n",
    "df['race_inf'] = df['race_inf'].str.replace('\\d+', '') #Delete Event Year from Event Name\n",
    "df['race_inf'] = df['race_inf'].str.strip() #Delete Lead and Tail Whitespace from Event Name\n",
    "\n",
    "# df.loc[~df['rank'].isin([\"1\",\"2\",\"3\",\"4\",\"5\",\"6\",\"7\",\"8\",\"9\",\"10\"]),'rank'] = \"11\" #Recode not Top10 Ranks to Rank 11"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I fill in the missing age values by subtracting the year of birth from the year in which the races took place. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['age_y'] = df['age_y'].fillna(df['race_y'] - df['yob']) #Where Age is missing, subtract birth year from event year to get age\n",
    "df = df[df['age_y'].notnull()] # Only keep lines where we have Age "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create Dataset for Forcasting 'rank' as categorical Variable"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "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>subj_id</th>\n",
       "      <th>age_y</th>\n",
       "      <th>nat</th>\n",
       "      <th>gender</th>\n",
       "      <th>race_inf</th>\n",
       "      <th>race_y</th>\n",
       "      <th>participants</th>\n",
       "      <th>rank</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2833</td>\n",
       "      <td>30.0</td>\n",
       "      <td>SRB</td>\n",
       "      <td>Men</td>\n",
       "      <td>Chennai</td>\n",
       "      <td>2020.0</td>\n",
       "      <td>52.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2869</td>\n",
       "      <td>21.0</td>\n",
       "      <td>HKG</td>\n",
       "      <td>Men</td>\n",
       "      <td>Chennai</td>\n",
       "      <td>2020.0</td>\n",
       "      <td>52.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>831</td>\n",
       "      <td>36.0</td>\n",
       "      <td>POL</td>\n",
       "      <td>Men</td>\n",
       "      <td>Chennai</td>\n",
       "      <td>2020.0</td>\n",
       "      <td>52.0</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>869</td>\n",
       "      <td>21.0</td>\n",
       "      <td>KAZ</td>\n",
       "      <td>Men</td>\n",
       "      <td>Chennai</td>\n",
       "      <td>2020.0</td>\n",
       "      <td>52.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1597</td>\n",
       "      <td>23.0</td>\n",
       "      <td>HKG</td>\n",
       "      <td>Men</td>\n",
       "      <td>Chennai</td>\n",
       "      <td>2020.0</td>\n",
       "      <td>52.0</td>\n",
       "      <td>7.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  subj_id  age_y  nat gender race_inf  race_y  participants  rank\n",
       "0    2833   30.0  SRB    Men  Chennai  2020.0          52.0   1.0\n",
       "1    2869   21.0  HKG    Men  Chennai  2020.0          52.0   2.0\n",
       "2     831   36.0  POL    Men  Chennai  2020.0          52.0   4.0\n",
       "3     869   21.0  KAZ    Men  Chennai  2020.0          52.0   5.0\n",
       "4    1597   23.0  HKG    Men  Chennai  2020.0          52.0   7.0"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df1 = df[['subj_id','age_y','nat','gender','race_inf','race_y','participants','rank']]\n",
    "df1.head()\n",
    "# dfMen = df1[df1['gender']==\"Men\"]\n",
    "#df1 = df1[df1['gender']==\"Women\"]\n",
    "#df1.describe(include='all')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Number Encoding of Categorical Variables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import LabelEncoder\n",
    "le = LabelEncoder()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'\\ndfleMen = dfle[dfle[\\'gender\\'] == \"Men\"].copy()\\ndfleMen = dfleMen.drop([\\'gender\\'],axis=\\'columns\\')\\ndfleWomen = dfle[dfle[\\'gender\\'] == \"Women\"].copy()\\ndfleWomen = dfleWomen.drop([\\'gender\\'],axis=\\'columns\\')\\n'"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dfle = df1.copy()\n",
    "dfle['subj_id'] = le.fit_transform(dfle['subj_id'])\n",
    "dfle['nat'] = le.fit_transform(dfle['nat'])\n",
    "dfle['race_inf'] = le.fit_transform(dfle['race_inf'])\n",
    "dfle['gender'] = le.fit_transform(dfle['gender'])\n",
    "#dfle['race_y'] = le.fit_transform(dfle['race_y'])\n",
    "\n",
    "\n",
    "'''\n",
    "dfleMen = dfle[dfle['gender'] == \"Men\"].copy()\n",
    "dfleMen = dfleMen.drop(['gender'],axis='columns')\n",
    "dfleWomen = dfle[dfle['gender'] == \"Women\"].copy()\n",
    "dfleWomen = dfleWomen.drop(['gender'],axis='columns')\n",
    "'''"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train/Test Split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "X_train, X_test, y_train, y_test = train_test_split(\n",
    "    dfle.drop(['rank'],axis='columns'),\n",
    "    dfle['rank'],\n",
    "    random_state=42,\n",
    "    test_size=0.2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Regular Decision Tree"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn import tree"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = tree.DecisionTreeClassifier()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "DecisionTreeClassifier(ccp_alpha=0.0, class_weight=None, criterion='gini',\n",
       "                       max_depth=None, max_features=None, max_leaf_nodes=None,\n",
       "                       min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "                       min_samples_leaf=1, min_samples_split=2,\n",
       "                       min_weight_fraction_leaf=0.0, presort='deprecated',\n",
       "                       random_state=None, splitter='best')"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(X_train,y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.06191950464396285"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.score(X_test,y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_predicted = model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[38, 33, 24, ...,  0,  0,  0],\n",
       "       [33, 20, 29, ...,  1,  1,  0],\n",
       "       [27, 20, 15, ...,  0,  0,  0],\n",
       "       ...,\n",
       "       [ 0,  0,  0, ...,  0,  0,  0],\n",
       "       [ 0,  0,  0, ...,  0,  1,  0],\n",
       "       [ 0,  0,  0, ...,  0,  0,  1]], dtype=int64)"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "labels = np.unique(y_test)\n",
    "cm = confusion_matrix(y_test, y_predicted,labels=labels)\n",
    "cm\n",
    "#cm = confusion_matrix(y_test, y_predicted, rownames=['True'], colnames=['Predicted'], labels=labels)\n",
    "\n",
    "#pd.set_option('display.max_rows',74)\n",
    "#pd.set_option('display.max_columns',74)\n",
    "\n",
    "#pd.DataFrame(cm, index=labels, columns=labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x245719347b8>"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x864 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sumOfTruePositive = []\n",
    "sumOfTrue = []\n",
    "sumOfPredicted = []\n",
    "\n",
    "for i in range(cm.shape[0]):\n",
    "    x = cm[i,i]\n",
    "    sumOfTruePositive.append(x)\n",
    "    y = sum(cm[i,:])\n",
    "    sumOfTrue.append(y)\n",
    "    z = sum(cm[:,i])\n",
    "    sumOfPredicted.append(z)\n",
    "    \n",
    "recall = np.array(sumOfTruePositive)/np.array(sumOfTrue)\n",
    "precision = np.array(sumOfTruePositive)/np.array(sumOfPredicted)\n",
    "\n",
    "resultDict = {'Rank':labels,'Recall':recall, 'Precision':precision, 'TotalTrue':sumOfTrue, 'TotalPredicted':sumOfPredicted}\n",
    "resultDf = pd.DataFrame(resultDict)\n",
    "\n",
    "fig = plt.figure(figsize=(20,12))\n",
    "ax1 = fig.add_subplot(411)\n",
    "ax2 = fig.add_subplot(412)\n",
    "ax3 = fig.add_subplot(413)\n",
    "ax4 = fig.add_subplot(414)\n",
    "\n",
    "resultDf.plot.bar(x='Rank', y='TotalTrue',ax=ax1)\n",
    "resultDf.plot.bar(x='Rank', y='TotalPredicted',ax=ax2)\n",
    "resultDf.plot.bar(x='Rank', y='Recall',ax=ax3)\n",
    "resultDf.plot.bar(x='Rank', y='Precision',ax=ax4)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap=True, ccp_alpha=0.0, class_weight=None,\n",
       "                       criterion='entropy', max_depth=None, max_features='auto',\n",
       "                       max_leaf_nodes=None, max_samples=None,\n",
       "                       min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "                       min_samples_leaf=1, min_samples_split=2,\n",
       "                       min_weight_fraction_leaf=0.0, n_estimators=30,\n",
       "                       n_jobs=None, oob_score=False, random_state=None,\n",
       "                       verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "model = RandomForestClassifier(criterion='entropy', n_estimators=30)\n",
    "model.fit(X_train,y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.06656346749226007"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.score(X_test,y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_predicted = model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "labels = np.unique(y_test)\n",
    "cm = confusion_matrix(y_test, y_predicted, labels=labels)\n",
    "\n",
    "#pd.set_option('display.max_rows',74)\n",
    "#pd.set_option('display.max_columns',74)\n",
    "#pd.DataFrame(cm, index=labels, columns=labels)\n",
    "#fig = plt.figure()\n",
    "#ax = fig.add_subplot(111)\n",
    "#ax.matshow(cm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2450ebd3860>"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x864 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sumOfTruePositive = []\n",
    "sumOfTrue = []\n",
    "sumOfPredicted = []\n",
    "\n",
    "for i in range(cm.shape[0]):\n",
    "    x = cm[i,i]\n",
    "    sumOfTruePositive.append(x)\n",
    "    y = sum(cm[i,:])\n",
    "    sumOfTrue.append(y)\n",
    "    z = sum(cm[:,i])\n",
    "    sumOfPredicted.append(z)\n",
    "    \n",
    "recall = np.array(sumOfTruePositive)/np.array(sumOfTrue)\n",
    "precision = np.array(sumOfTruePositive)/np.array(sumOfPredicted)\n",
    "\n",
    "resultDict = {'Rank':labels,'Recall':recall, 'Precision':precision, 'TotalTrue':sumOfTrue, 'TotalPredicted':sumOfPredicted}\n",
    "resultDf = pd.DataFrame(resultDict)\n",
    "\n",
    "fig = plt.figure(figsize=(20,12))\n",
    "ax1 = fig.add_subplot(411)\n",
    "ax2 = fig.add_subplot(412)\n",
    "ax3 = fig.add_subplot(413)\n",
    "ax4 = fig.add_subplot(414)\n",
    "\n",
    "resultDf.plot.bar(x='Rank', y='TotalTrue',ax=ax1)\n",
    "resultDf.plot.bar(x='Rank', y='TotalPredicted',ax=ax2)\n",
    "resultDf.plot.bar(x='Rank', y='Recall',ax=ax3)\n",
    "resultDf.plot.bar(x='Rank', y='Precision',ax=ax4)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Tryout: Calculation Pseudo NDCG Score for comparison with RankNet Results"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I tried to calculate a Pseudo NDCG Score for comparing the classificatin results to the RankNet Algorithm. Unfortunately this didn't work out nicely. I probably would need to do the sampling based on race_id in order to get a nice \"Result-Set\" structure. Like this I receive a lot of NDCG Score = 1.0 Results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "ndcgDf = X_test[['race_inf','race_y','gender']].copy()\n",
    "ndcgDf['rank'] = y_test\n",
    "ndcgDf['predicted'] = y_predicted"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "ndcgDf1 = ndcgDf.groupby(['race_inf','race_y','gender'],group_keys=False).apply(pd.DataFrame.sort_values,'rank')\n",
    "ndcgDf1['realRank'] = ndcgDf1.groupby(['race_inf','race_y','gender'])['rank'].rank(ascending=True)\n",
    "ndcgDf1['predRank'] = ndcgDf1.groupby(['race_inf','race_y','gender'])['predicted'].rank(ascending=True)\n",
    "ndcgDf1['realRelevance'] = ndcgDf1.groupby(['race_inf','race_y','gender'])['rank'].rank(ascending=False)\n",
    "ndcgDf1['realRelevance_V2'] = 1/ndcgDf1['realRank']\n",
    "ndcgDf1['ndcgBaseIdeal'] = ndcgDf1['realRelevance_V2']/np.log2(ndcgDf1['realRank']+1)\n",
    "ndcgDf1['ndcgBasePred'] = ndcgDf1['realRelevance_V2']/np.log2(ndcgDf1['predRank']+1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "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>race_inf</th>\n",
       "      <th>race_y</th>\n",
       "      <th>gender</th>\n",
       "      <th>rank</th>\n",
       "      <th>predicted</th>\n",
       "      <th>realRank</th>\n",
       "      <th>predRank</th>\n",
       "      <th>realRelevance</th>\n",
       "      <th>realRelevance_V2</th>\n",
       "      <th>ndcgBaseIdeal</th>\n",
       "      <th>ndcgBasePred</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>6835</th>\n",
       "      <td>0</td>\n",
       "      <td>2016.0</td>\n",
       "      <td>0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>35.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>6.5</td>\n",
       "      <td>11.0</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.344010</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6845</th>\n",
       "      <td>0</td>\n",
       "      <td>2016.0</td>\n",
       "      <td>0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>0.315465</td>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6846</th>\n",
       "      <td>0</td>\n",
       "      <td>2016.0</td>\n",
       "      <td>0</td>\n",
       "      <td>16.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.166667</td>\n",
       "      <td>0.210310</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6851</th>\n",
       "      <td>0</td>\n",
       "      <td>2016.0</td>\n",
       "      <td>0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>56.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>10.5</td>\n",
       "      <td>8.0</td>\n",
       "      <td>0.250000</td>\n",
       "      <td>0.107669</td>\n",
       "      <td>0.070951</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6852</th>\n",
       "      <td>0</td>\n",
       "      <td>2016.0</td>\n",
       "      <td>0</td>\n",
       "      <td>22.0</td>\n",
       "      <td>38.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>0.200000</td>\n",
       "      <td>0.077371</td>\n",
       "      <td>0.063093</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6861</th>\n",
       "      <td>0</td>\n",
       "      <td>2016.0</td>\n",
       "      <td>0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>34.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.166667</td>\n",
       "      <td>0.059368</td>\n",
       "      <td>0.064475</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6865</th>\n",
       "      <td>0</td>\n",
       "      <td>2016.0</td>\n",
       "      <td>0</td>\n",
       "      <td>36.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>0.142857</td>\n",
       "      <td>0.047619</td>\n",
       "      <td>0.071429</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6877</th>\n",
       "      <td>0</td>\n",
       "      <td>2016.0</td>\n",
       "      <td>0</td>\n",
       "      <td>50.0</td>\n",
       "      <td>56.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>10.5</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.125000</td>\n",
       "      <td>0.039433</td>\n",
       "      <td>0.035475</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6878</th>\n",
       "      <td>0</td>\n",
       "      <td>2016.0</td>\n",
       "      <td>0</td>\n",
       "      <td>51.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.111111</td>\n",
       "      <td>0.033448</td>\n",
       "      <td>0.047853</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6887</th>\n",
       "      <td>0</td>\n",
       "      <td>2016.0</td>\n",
       "      <td>0</td>\n",
       "      <td>56.0</td>\n",
       "      <td>44.0</td>\n",
       "      <td>10.5</td>\n",
       "      <td>9.0</td>\n",
       "      <td>1.5</td>\n",
       "      <td>0.095238</td>\n",
       "      <td>0.027029</td>\n",
       "      <td>0.028670</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6888</th>\n",
       "      <td>0</td>\n",
       "      <td>2016.0</td>\n",
       "      <td>0</td>\n",
       "      <td>56.0</td>\n",
       "      <td>35.0</td>\n",
       "      <td>10.5</td>\n",
       "      <td>6.5</td>\n",
       "      <td>1.5</td>\n",
       "      <td>0.095238</td>\n",
       "      <td>0.027029</td>\n",
       "      <td>0.032763</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6892</th>\n",
       "      <td>0</td>\n",
       "      <td>2016.0</td>\n",
       "      <td>1</td>\n",
       "      <td>7.0</td>\n",
       "      <td>27.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6895</th>\n",
       "      <td>0</td>\n",
       "      <td>2016.0</td>\n",
       "      <td>1</td>\n",
       "      <td>13.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.5</td>\n",
       "      <td>8.0</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>0.315465</td>\n",
       "      <td>0.378235</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6900</th>\n",
       "      <td>0</td>\n",
       "      <td>2016.0</td>\n",
       "      <td>1</td>\n",
       "      <td>19.0</td>\n",
       "      <td>50.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.166667</td>\n",
       "      <td>0.105155</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6901</th>\n",
       "      <td>0</td>\n",
       "      <td>2016.0</td>\n",
       "      <td>1</td>\n",
       "      <td>20.0</td>\n",
       "      <td>50.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.250000</td>\n",
       "      <td>0.107669</td>\n",
       "      <td>0.078866</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      race_inf  race_y  gender  rank  predicted  realRank  predRank  \\\n",
       "6835         0  2016.0       0   3.0       35.0       1.0       6.5   \n",
       "6845         0  2016.0       0  15.0        2.0       2.0       1.0   \n",
       "6846         0  2016.0       0  16.0        9.0       3.0       2.0   \n",
       "6851         0  2016.0       0  21.0       56.0       4.0      10.5   \n",
       "6852         0  2016.0       0  22.0       38.0       5.0       8.0   \n",
       "6861         0  2016.0       0  32.0       34.0       6.0       5.0   \n",
       "6865         0  2016.0       0  36.0       11.0       7.0       3.0   \n",
       "6877         0  2016.0       0  50.0       56.0       8.0      10.5   \n",
       "6878         0  2016.0       0  51.0       13.0       9.0       4.0   \n",
       "6887         0  2016.0       0  56.0       44.0      10.5       9.0   \n",
       "6888         0  2016.0       0  56.0       35.0      10.5       6.5   \n",
       "6892         0  2016.0       1   7.0       27.0       1.0       3.0   \n",
       "6895         0  2016.0       1  13.0       12.0       2.0       1.5   \n",
       "6900         0  2016.0       1  19.0       50.0       3.0       8.0   \n",
       "6901         0  2016.0       1  20.0       50.0       4.0       8.0   \n",
       "\n",
       "      realRelevance  realRelevance_V2  ndcgBaseIdeal  ndcgBasePred  \n",
       "6835           11.0          1.000000       1.000000      0.344010  \n",
       "6845           10.0          0.500000       0.315465      0.500000  \n",
       "6846            9.0          0.333333       0.166667      0.210310  \n",
       "6851            8.0          0.250000       0.107669      0.070951  \n",
       "6852            7.0          0.200000       0.077371      0.063093  \n",
       "6861            6.0          0.166667       0.059368      0.064475  \n",
       "6865            5.0          0.142857       0.047619      0.071429  \n",
       "6877            4.0          0.125000       0.039433      0.035475  \n",
       "6878            3.0          0.111111       0.033448      0.047853  \n",
       "6887            1.5          0.095238       0.027029      0.028670  \n",
       "6888            1.5          0.095238       0.027029      0.032763  \n",
       "6892            9.0          1.000000       1.000000      0.500000  \n",
       "6895            8.0          0.500000       0.315465      0.378235  \n",
       "6900            7.0          0.333333       0.166667      0.105155  \n",
       "6901            6.0          0.250000       0.107669      0.078866  "
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ndcgDf1.head(15)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x245128eba20>"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "ndcgDf2 = ndcgDf1.groupby(['race_inf','race_y','gender'],group_keys=False)[['ndcgBaseIdeal','ndcgBasePred']].sum()\n",
    "ndcgScores = ndcgDf2['ndcgBasePred'] / ndcgDf2['ndcgBaseIdeal']\n",
    "ndcgScores.plot.hist(bins=20, alpha=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Logistic Regression"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Logistic Regression doesn't work, because there aren't sufficient cases for each combination of categorical variables (?). Some subj_id s only exist once. Also some test-dataset subj_id aren't contained in the train-dataset. Correct(?)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "model = LogisticRegression()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\finns\\Anaconda3\\envs\\Deep Learning CAS 2020\\lib\\site-packages\\sklearn\\linear_model\\_logistic.py:940: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  extra_warning_msg=_LOGISTIC_SOLVER_CONVERGENCE_MSG)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,\n",
       "                   intercept_scaling=1, l1_ratio=None, max_iter=1500,\n",
       "                   multi_class='auto', n_jobs=None, penalty='l2',\n",
       "                   random_state=None, solver='lbfgs', tol=0.0001, verbose=0,\n",
       "                   warm_start=False)"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(X_train,y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In order to be able to do a logistic regression I delete the variable \"subj_id\", \"race_y\" and \"race_inf\" from the test/train datasets in order for the model to converge. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "dfLog = df[['age_y','nat','gender','rank']]\n",
    "#'race_inf',\n",
    "dfLogLe = dfLog.copy()\n",
    "dfLogLe['nat'] = le.fit_transform(dfLogLe['nat'])\n",
    "#dfLogLe['race_inf'] = le.fit_transform(dfLogLe['race_inf'])\n",
    "dfLogLe['gender'] = le.fit_transform(dfLogLe['gender'])\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "X_train1, X_test1, y_train1, y_test1 = train_test_split(\n",
    "    dfLogLe.drop(['rank'],axis='columns'),\n",
    "    dfLogLe['rank'],\n",
    "    random_state=42,\n",
    "    test_size=0.2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SVC(C=1.0, break_ties=False, cache_size=200, class_weight=None, coef0=0.0,\n",
       "    decision_function_shape='ovr', degree=3, gamma=0.01, kernel='rbf',\n",
       "    max_iter=-1, probability=False, random_state=None, shrinking=True,\n",
       "    tol=0.001, verbose=False)"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(X_train1,y_train1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.048331613347093225"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.score(X_test1, y_test1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_predicted1 = model.predict(X_test1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "labels = np.unique(y_test1)\n",
    "cm = confusion_matrix(y_test1, y_predicted1, labels=labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\finns\\Anaconda3\\envs\\Deep Learning CAS 2020\\lib\\site-packages\\ipykernel_launcher.py:14: RuntimeWarning: invalid value encountered in true_divide\n",
      "  \n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2452350c6d8>"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x864 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sumOfTruePositive = []\n",
    "sumOfTrue = []\n",
    "sumOfPredicted = []\n",
    "\n",
    "for i in range(cm.shape[0]):\n",
    "    x = cm[i,i]\n",
    "    sumOfTruePositive.append(x)\n",
    "    y = sum(cm[i,:])\n",
    "    sumOfTrue.append(y)\n",
    "    z = sum(cm[:,i])\n",
    "    sumOfPredicted.append(z)\n",
    "    \n",
    "recall = np.array(sumOfTruePositive)/np.array(sumOfTrue)\n",
    "precision = np.array(sumOfTruePositive)/np.array(sumOfPredicted)\n",
    "\n",
    "resultDict = {'Rank':labels,'Recall':recall, 'Precision':precision, 'TotalTrue':sumOfTrue, 'TotalPredicted':sumOfPredicted}\n",
    "resultDf = pd.DataFrame(resultDict)\n",
    "\n",
    "fig = plt.figure(figsize=(20,12))\n",
    "ax1 = fig.add_subplot(411)\n",
    "ax2 = fig.add_subplot(412)\n",
    "ax3 = fig.add_subplot(413)\n",
    "ax4 = fig.add_subplot(414)\n",
    "\n",
    "resultDf.plot.bar(x='Rank', y='TotalTrue',ax=ax1)\n",
    "resultDf.plot.bar(x='Rank', y='TotalPredicted',ax=ax2)\n",
    "resultDf.plot.bar(x='Rank', y='Recall',ax=ax3)\n",
    "resultDf.plot.bar(x='Rank', y='Precision',ax=ax4)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Support Vector Machine"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.svm import SVC\n",
    "model = SVC(gamma = 0.01)\n",
    "# gamma = 0.01\n",
    "# C=1.0, gamma = 'auto', kernel = 'rbf'\n",
    "# Support Vector machine tries to maximize the \"margin\" between data points. \n",
    "# Gamma determines how many points are considered for the Margin. \n",
    "# High Gamma means only points close to the partitioning vecotr. \n",
    "# Low Gamma means also points that are farther away are considered. High gamma tends to do more overfitting. \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SVC(C=1.0, break_ties=False, cache_size=200, class_weight=None, coef0=0.0,\n",
       "    decision_function_shape='ovr', degree=3, gamma=0.01, kernel='rbf',\n",
       "    max_iter=-1, probability=False, random_state=None, shrinking=True,\n",
       "    tol=0.001, verbose=False)"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.048331613347093225"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_predicted = model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "labels = np.unique(y_test)\n",
    "cm = confusion_matrix(y_test, y_predicted, labels=labels)\n",
    "\n",
    "#pd.set_option('display.max_rows',74)\n",
    "#pd.set_option('display.max_columns',74)\n",
    "#pd.DataFrame(cm, index=labels, columns=labels)\n",
    "#fig = plt.figure()\n",
    "#ax = fig.add_subplot(111)\n",
    "#ax.matshow(cm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\finns\\Anaconda3\\envs\\Deep Learning CAS 2020\\lib\\site-packages\\ipykernel_launcher.py:14: RuntimeWarning: invalid value encountered in true_divide\n",
      "  \n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x24519a27630>"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x864 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sumOfTruePositive = []\n",
    "sumOfTrue = []\n",
    "sumOfPredicted = []\n",
    "\n",
    "for i in range(cm.shape[0]):\n",
    "    x = cm[i,i]\n",
    "    sumOfTruePositive.append(x)\n",
    "    y = sum(cm[i,:])\n",
    "    sumOfTrue.append(y)\n",
    "    z = sum(cm[:,i])\n",
    "    sumOfPredicted.append(z)\n",
    "    \n",
    "recall = np.array(sumOfTruePositive)/np.array(sumOfTrue)\n",
    "precision = np.array(sumOfTruePositive)/np.array(sumOfPredicted)\n",
    "\n",
    "resultDict = {'Rank':labels,'Recall':recall, 'Precision':precision, 'TotalTrue':sumOfTrue, 'TotalPredicted':sumOfPredicted}\n",
    "resultDf = pd.DataFrame(resultDict)\n",
    "\n",
    "fig = plt.figure(figsize=(20,12))\n",
    "ax1 = fig.add_subplot(411)\n",
    "ax2 = fig.add_subplot(412)\n",
    "ax3 = fig.add_subplot(413)\n",
    "ax4 = fig.add_subplot(414)\n",
    "\n",
    "resultDf.plot.bar(x='Rank', y='TotalTrue',ax=ax1)\n",
    "resultDf.plot.bar(x='Rank', y='TotalPredicted',ax=ax2)\n",
    "resultDf.plot.bar(x='Rank', y='Recall',ax=ax3)\n",
    "resultDf.plot.bar(x='Rank', y='Precision',ax=ax4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Following Code chunks work a lot better to calculate precision and recall from the confusion matrix. Above I used a loop to do the same thing.\n",
    "\n",
    "#np.diagonal(cm)\n",
    "#np.diagonal(cm)/cm.sum(axis=1)\n",
    "#cm.sum(axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Explanation for SVM results: \n",
    "# https://stats.stackexchange.com/questions/125353/output-of-scikit-svm-in-multiclass-classification-always-gives-same-label"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Tutorials\n",
    "https://www.youtube.com/watch?v=PHxYNGo8NcI => Decision Tree <br>\n",
    "https://www.youtube.com/watch?v=ok2s1vV9XW0 => Random Forest <br>\n",
    "https://www.youtube.com/watch?v=9yl6-HEY7_s => Dummy Variables and One Hot Encoding <br>\n",
    "https://www.youtube.com/watch?v=J5bXOOmkopc => Logistic Regression <br>\n",
    "https://www.youtube.com/watch?v=FB5EdxAGxQg => Support Vector Machine <br>\n",
    "https://www.datacamp.com/community/tutorials/deep-learning-python?utm_source=adwords_ppc&utm_campaignid=1655852085&utm_adgroupid=61045434462&utm_device=c&utm_keyword=%2Bkeras%20%2Btutorial&utm_matchtype=b&utm_network=g&utm_adpostion=&utm_creative=318880582449&utm_targetid=kwd-321066923947&utm_loc_interest_ms=&utm_loc_physical_ms=1003297&gclid=EAIaIQobChMIgZSiw_vQ6AIVibHtCh29xwhXEAAYASAAEgJfQvD_BwE => Deep Learning <br>\n",
    "https://mlexplained.com/2019/05/27/learning-to-rank-explained-with-code/ => LambdaRank  <br>\n",
    "https://towardsdatascience.com/evaluate-your-recommendation-engine-using-ndcg-759a851452d1 => ndcg Score"
   ]
  },
  {
   "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": "3.6.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}