Prediction.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Vorhersage und maschinelles Lernen"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Vorbereitungen"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Importe"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import seaborn as sns\n",
    "from matplotlib import pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "scrolled": false
   },
   "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>NUMIND</th>\n",
       "      <th>ID</th>\n",
       "      <th>ANFANG</th>\n",
       "      <th>VORNAME</th>\n",
       "      <th>NACHNAME</th>\n",
       "      <th>ALTER</th>\n",
       "      <th>GESCHLECHT</th>\n",
       "      <th>PARTEI_KURZ</th>\n",
       "      <th>BERUF</th>\n",
       "      <th>AUSLAND</th>\n",
       "      <th>...</th>\n",
       "      <th>WP10</th>\n",
       "      <th>WP11</th>\n",
       "      <th>WP12</th>\n",
       "      <th>WP13</th>\n",
       "      <th>WP14</th>\n",
       "      <th>WP15</th>\n",
       "      <th>WP16</th>\n",
       "      <th>WP17</th>\n",
       "      <th>WP18</th>\n",
       "      <th>WP19</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>WP-JAHRE</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1949-1953;01</th>\n",
       "      <td>0</td>\n",
       "      <td>11000009</td>\n",
       "      <td>1949</td>\n",
       "      <td>Konrad</td>\n",
       "      <td>Adenauer</td>\n",
       "      <td>73</td>\n",
       "      <td>1</td>\n",
       "      <td>Union</td>\n",
       "      <td>Bundeskanzler a. D.</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1949-1953;01</th>\n",
       "      <td>1</td>\n",
       "      <td>11000013</td>\n",
       "      <td>1949</td>\n",
       "      <td>Willi</td>\n",
       "      <td>Agatz</td>\n",
       "      <td>45</td>\n",
       "      <td>1</td>\n",
       "      <td>Andere</td>\n",
       "      <td>Bergmann, Gewerkschaftsfunktionär</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1949-1953;01</th>\n",
       "      <td>2</td>\n",
       "      <td>11000015</td>\n",
       "      <td>1949</td>\n",
       "      <td>Adolf</td>\n",
       "      <td>Ahrens</td>\n",
       "      <td>70</td>\n",
       "      <td>1</td>\n",
       "      <td>Andere</td>\n",
       "      <td>Kapitän a. D.</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1949-1953;01</th>\n",
       "      <td>3</td>\n",
       "      <td>11000020</td>\n",
       "      <td>1949</td>\n",
       "      <td>Johannes</td>\n",
       "      <td>Albers</td>\n",
       "      <td>59</td>\n",
       "      <td>1</td>\n",
       "      <td>Union</td>\n",
       "      <td>Geschäftsführer</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1949-1953;01</th>\n",
       "      <td>4</td>\n",
       "      <td>11000021</td>\n",
       "      <td>1949</td>\n",
       "      <td>Luise</td>\n",
       "      <td>Albertz</td>\n",
       "      <td>48</td>\n",
       "      <td>0</td>\n",
       "      <td>SPD</td>\n",
       "      <td>Buchhalterin</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 30 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "              NUMIND        ID  ANFANG   VORNAME  NACHNAME  ALTER  GESCHLECHT  \\\n",
       "WP-JAHRE                                                                        \n",
       "1949-1953;01       0  11000009    1949    Konrad  Adenauer     73           1   \n",
       "1949-1953;01       1  11000013    1949     Willi     Agatz     45           1   \n",
       "1949-1953;01       2  11000015    1949     Adolf    Ahrens     70           1   \n",
       "1949-1953;01       3  11000020    1949  Johannes    Albers     59           1   \n",
       "1949-1953;01       4  11000021    1949     Luise   Albertz     48           0   \n",
       "\n",
       "             PARTEI_KURZ                              BERUF  AUSLAND  ...   \\\n",
       "WP-JAHRE                                                              ...    \n",
       "1949-1953;01       Union                Bundeskanzler a. D.        0  ...    \n",
       "1949-1953;01      Andere  Bergmann, Gewerkschaftsfunktionär        0  ...    \n",
       "1949-1953;01      Andere                      Kapitän a. D.        0  ...    \n",
       "1949-1953;01       Union                    Geschäftsführer        0  ...    \n",
       "1949-1953;01         SPD                       Buchhalterin        0  ...    \n",
       "\n",
       "              WP10  WP11  WP12  WP13  WP14  WP15  WP16  WP17  WP18  WP19  \n",
       "WP-JAHRE                                                                  \n",
       "1949-1953;01     0     0     0     0     0     0     0     0     0     0  \n",
       "1949-1953;01     0     0     0     0     0     0     0     0     0     0  \n",
       "1949-1953;01     0     0     0     0     0     0     0     0     0     0  \n",
       "1949-1953;01     0     0     0     0     0     0     0     0     0     0  \n",
       "1949-1953;01     0     0     0     0     0     0     0     0     0     0  \n",
       "\n",
       "[5 rows x 30 columns]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv(\"MdB_Indexed.csv\")\n",
    "df = df.set_index(\"WP-JAHRE\")\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Speichern von nützlichen Spalten"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Einige Spalten können für die Anwendung der ML-Modelle zwar übergagagnen werden, jedoch sind sie nötig um die finale Vorhersage für den nächsten Bundestag dem entsprechenden MdB zuzuordnen."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1949 1953 1957 1961 1965 1969 1972 1976 1980 1983 1987 1990 1994 1998\n",
      " 2002 2005 2009 2013 2017]\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ANFANG</th>\n",
       "      <th>ALTER</th>\n",
       "      <th>GESCHLECHT</th>\n",
       "      <th>PARTEI_KURZ</th>\n",
       "      <th>AUSLAND</th>\n",
       "      <th>MANDATSART</th>\n",
       "      <th>WP1</th>\n",
       "      <th>WP2</th>\n",
       "      <th>WP3</th>\n",
       "      <th>WP4</th>\n",
       "      <th>...</th>\n",
       "      <th>WP10</th>\n",
       "      <th>WP11</th>\n",
       "      <th>WP12</th>\n",
       "      <th>WP13</th>\n",
       "      <th>WP14</th>\n",
       "      <th>WP15</th>\n",
       "      <th>WP16</th>\n",
       "      <th>WP17</th>\n",
       "      <th>WP18</th>\n",
       "      <th>WP19</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>WP-JAHRE</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1949-1953;01</th>\n",
       "      <td>1949</td>\n",
       "      <td>73</td>\n",
       "      <td>1</td>\n",
       "      <td>Union</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1949-1953;01</th>\n",
       "      <td>1949</td>\n",
       "      <td>45</td>\n",
       "      <td>1</td>\n",
       "      <td>Andere</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1949-1953;01</th>\n",
       "      <td>1949</td>\n",
       "      <td>70</td>\n",
       "      <td>1</td>\n",
       "      <td>Andere</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1949-1953;01</th>\n",
       "      <td>1949</td>\n",
       "      <td>59</td>\n",
       "      <td>1</td>\n",
       "      <td>Union</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1949-1953;01</th>\n",
       "      <td>1949</td>\n",
       "      <td>48</td>\n",
       "      <td>0</td>\n",
       "      <td>SPD</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 25 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "              ANFANG  ALTER  GESCHLECHT PARTEI_KURZ  AUSLAND  MANDATSART  WP1  \\\n",
       "WP-JAHRE                                                                        \n",
       "1949-1953;01    1949     73           1       Union        0         1.0    1   \n",
       "1949-1953;01    1949     45           1      Andere        0         0.0    1   \n",
       "1949-1953;01    1949     70           1      Andere        0         0.0    1   \n",
       "1949-1953;01    1949     59           1       Union        0         1.0    1   \n",
       "1949-1953;01    1949     48           0         SPD        0         0.0    1   \n",
       "\n",
       "              WP2  WP3  WP4  ...   WP10  WP11  WP12  WP13  WP14  WP15  WP16  \\\n",
       "WP-JAHRE                     ...                                              \n",
       "1949-1953;01    1    1    1  ...      0     0     0     0     0     0     0   \n",
       "1949-1953;01    0    0    0  ...      0     0     0     0     0     0     0   \n",
       "1949-1953;01    0    0    0  ...      0     0     0     0     0     0     0   \n",
       "1949-1953;01    1    0    0  ...      0     0     0     0     0     0     0   \n",
       "1949-1953;01    1    1    1  ...      0     0     0     0     0     0     0   \n",
       "\n",
       "              WP17  WP18  WP19  \n",
       "WP-JAHRE                        \n",
       "1949-1953;01     0     0     0  \n",
       "1949-1953;01     0     0     0  \n",
       "1949-1953;01     0     0     0  \n",
       "1949-1953;01     0     0     0  \n",
       "1949-1953;01     0     0     0  \n",
       "\n",
       "[5 rows x 25 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "IDs = df[[\"ANFANG\",\"ID\", \"VORNAME\", \"NACHNAME\"]].copy()\n",
    "df = df.drop([\"ID\", \"NUMIND\", \"BERUF\", \"VORNAME\", \"NACHNAME\"], axis = 1)\n",
    "\n",
    "#für etwaige Visualisierungen von Zeitreihen\n",
    "timeline= df[\"ANFANG\"].unique()\n",
    "print(timeline)\n",
    "\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Extrahieren von einzelen DataFrames"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Die einzelnen Wahlperioden müssen wieder in eigene DataFrames überführt werden, um später einzeln Modelle anwenden zu können."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "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>ANFANG</th>\n",
       "      <th>ALTER</th>\n",
       "      <th>GESCHLECHT</th>\n",
       "      <th>PARTEI_KURZ</th>\n",
       "      <th>AUSLAND</th>\n",
       "      <th>MANDATSART</th>\n",
       "      <th>WP1</th>\n",
       "      <th>WP2</th>\n",
       "      <th>WP3</th>\n",
       "      <th>WP4</th>\n",
       "      <th>...</th>\n",
       "      <th>WP10</th>\n",
       "      <th>WP11</th>\n",
       "      <th>WP12</th>\n",
       "      <th>WP13</th>\n",
       "      <th>WP14</th>\n",
       "      <th>WP15</th>\n",
       "      <th>WP16</th>\n",
       "      <th>WP17</th>\n",
       "      <th>WP18</th>\n",
       "      <th>WP19</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>WP-JAHRE</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1965–1969;05</th>\n",
       "      <td>1965</td>\n",
       "      <td>35</td>\n",
       "      <td>1</td>\n",
       "      <td>Union</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1965–1969;05</th>\n",
       "      <td>1965</td>\n",
       "      <td>56</td>\n",
       "      <td>1</td>\n",
       "      <td>FDP</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1965–1969;05</th>\n",
       "      <td>1965</td>\n",
       "      <td>46</td>\n",
       "      <td>1</td>\n",
       "      <td>SPD</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1965–1969;05</th>\n",
       "      <td>1965</td>\n",
       "      <td>89</td>\n",
       "      <td>1</td>\n",
       "      <td>Union</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1965–1969;05</th>\n",
       "      <td>1965</td>\n",
       "      <td>45</td>\n",
       "      <td>1</td>\n",
       "      <td>Union</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 25 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "              ANFANG  ALTER  GESCHLECHT PARTEI_KURZ  AUSLAND  MANDATSART  WP1  \\\n",
       "WP-JAHRE                                                                        \n",
       "1965–1969;05    1965     35           1       Union        0         1.0    0   \n",
       "1965–1969;05    1965     56           1         FDP        0         0.0    0   \n",
       "1965–1969;05    1965     46           1         SPD        0         0.0    0   \n",
       "1965–1969;05    1965     89           1       Union        0         1.0    1   \n",
       "1965–1969;05    1965     45           1       Union        0         1.0    0   \n",
       "\n",
       "              WP2  WP3  WP4  ...   WP10  WP11  WP12  WP13  WP14  WP15  WP16  \\\n",
       "WP-JAHRE                     ...                                              \n",
       "1965–1969;05    0    0    0  ...      1     1     0     0     0     0     0   \n",
       "1965–1969;05    0    1    1  ...      0     0     0     0     0     0     0   \n",
       "1965–1969;05    0    0    0  ...      0     0     0     0     0     0     0   \n",
       "1965–1969;05    1    1    1  ...      0     0     0     0     0     0     0   \n",
       "1965–1969;05    0    0    1  ...      0     0     0     0     0     0     0   \n",
       "\n",
       "              WP17  WP18  WP19  \n",
       "WP-JAHRE                        \n",
       "1965–1969;05     0     0     0  \n",
       "1965–1969;05     0     0     0  \n",
       "1965–1969;05     0     0     0  \n",
       "1965–1969;05     0     0     0  \n",
       "1965–1969;05     0     0     0  \n",
       "\n",
       "[5 rows x 25 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "uniq = list(df.index.unique())\n",
    "\n",
    "#Einzelne DFs\n",
    "df_WP01 = df.loc[uniq[0]]\n",
    "df_WP02 = df.loc[uniq[1]]\n",
    "df_WP03 = df.loc[uniq[2]]\n",
    "df_WP04 = df.loc[uniq[3]]\n",
    "df_WP05 = df.loc[uniq[4]]\n",
    "df_WP06 = df.loc[uniq[5]]\n",
    "df_WP07 = df.loc[uniq[6]]\n",
    "df_WP08 = df.loc[uniq[7]]\n",
    "df_WP09 = df.loc[uniq[8]]\n",
    "df_WP10 = df.loc[uniq[9]]\n",
    "df_WP11 = df.loc[uniq[10]]\n",
    "df_WP12 = df.loc[uniq[11]]\n",
    "df_WP13 = df.loc[uniq[12]]\n",
    "df_WP14 = df.loc[uniq[13]]\n",
    "df_WP15 = df.loc[uniq[14]]\n",
    "df_WP16 = df.loc[uniq[15]]\n",
    "df_WP17 = df.loc[uniq[16]]\n",
    "df_WP18 = df.loc[uniq[17]]\n",
    "df_WP19 = df.loc[uniq[18]]\n",
    "\n",
    "\n",
    "df_wp_list = [df_WP01, df_WP02, df_WP03, df_WP04, df_WP05, df_WP06, df_WP07, df_WP08, df_WP09, df_WP10, df_WP11, df_WP12, df_WP13,\n",
    "df_WP14,df_WP15,df_WP16,df_WP17,df_WP18, df_WP19]\n",
    "\n",
    "\n",
    "#Test\n",
    "df_wp_list[4].head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Beseitigung von Spalten mit überflüssiger Information"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Alle Spalten für Wahlperioden, die zeitlich nach der unmittelbar Folgenden (später das Target) liegen, da diese aus Gründen der Chronologie nicht relevant sind. Sowie die Spalte für die jetzige Wahlperiode (Target für das DataFrame der vorherigen Wahlperiode)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "scrolled": false
   },
   "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>ANFANG</th>\n",
       "      <th>ALTER</th>\n",
       "      <th>GESCHLECHT</th>\n",
       "      <th>PARTEI_KURZ</th>\n",
       "      <th>AUSLAND</th>\n",
       "      <th>MANDATSART</th>\n",
       "      <th>WP1</th>\n",
       "      <th>WP2</th>\n",
       "      <th>WP3</th>\n",
       "      <th>WP5</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>WP-JAHRE</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1961–1965;04</th>\n",
       "      <td>1961</td>\n",
       "      <td>52</td>\n",
       "      <td>1</td>\n",
       "      <td>FDP</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1961–1965;04</th>\n",
       "      <td>1961</td>\n",
       "      <td>48</td>\n",
       "      <td>0</td>\n",
       "      <td>Union</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1961–1965;04</th>\n",
       "      <td>1961</td>\n",
       "      <td>85</td>\n",
       "      <td>1</td>\n",
       "      <td>Union</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1961–1965;04</th>\n",
       "      <td>1961</td>\n",
       "      <td>41</td>\n",
       "      <td>1</td>\n",
       "      <td>Union</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1961–1965;04</th>\n",
       "      <td>1961</td>\n",
       "      <td>37</td>\n",
       "      <td>1</td>\n",
       "      <td>Union</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              ANFANG  ALTER  GESCHLECHT PARTEI_KURZ  AUSLAND  MANDATSART  WP1  \\\n",
       "WP-JAHRE                                                                        \n",
       "1961–1965;04    1961     52           1         FDP        0         0.0    0   \n",
       "1961–1965;04    1961     48           0       Union        1         0.0    0   \n",
       "1961–1965;04    1961     85           1       Union        0         1.0    1   \n",
       "1961–1965;04    1961     41           1       Union        0         1.0    0   \n",
       "1961–1965;04    1961     37           1       Union        0         1.0    0   \n",
       "\n",
       "              WP2  WP3  WP5  \n",
       "WP-JAHRE                     \n",
       "1961–1965;04    0    1    1  \n",
       "1961–1965;04    1    1    0  \n",
       "1961–1965;04    1    1    1  \n",
       "1961–1965;04    0    0    1  \n",
       "1961–1965;04    0    1    1  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Vorbereitung:\n",
    "wp_list = np.array([\"WP1\",\"WP2\",\"WP3\",\"WP4\",\"WP5\",\"WP6\",\"WP7\",\"WP8\",\"WP9\",\"WP10\",\"WP11\",\"WP12\",\"WP13\",\"WP14\",\"WP15\",\"WP16\",\"WP17\",\"WP18\",\"WP19\"])\t\n",
    "# Funktion\n",
    "def find_redun(dafr):\n",
    "    # Nummer der WP\n",
    "    num = int(dafr.index[0][-2:])\n",
    "    # Selektion von Indizes für wp_list\n",
    "    ind = np.array(range(0,19))\n",
    "    sel = list(ind[ind>num])\n",
    "    # Hinzufügen der DF-eigenen Wahlperiode\n",
    "    sel.append(ind[num-1])\n",
    "    # Zusammenstellung der redundanten Spalten\n",
    "    drop_cols = list(wp_list[sel])\n",
    "    return drop_cols\n",
    "\n",
    "# Anwendung\n",
    "for d in list(range(0,19)):\n",
    "    drop = list(find_redun(df_wp_list[d]))\n",
    "    df_wp_list[d] = df_wp_list[d].drop(drop, axis = 1)\n",
    "\n",
    "    \n",
    "# Test an einem Beispiel \n",
    "# Bundestag Nr.4 --> Spalten für WP4 und alle nach WP5 sollten verschwunden sein\n",
    "df_wp_list[3].head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Generierung einer Spalte für die Erfahrung eines MdBs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    " Für die Anwendung einer ML-Modells benötigen die einzelnen DataFrames die gleiche Anzahl an Features. Daher werden hier die einzelnen Spalten für alle Wahlperioden (binär) addiert und so ein Maß für die Erfahrung eines MdBs geschaffen. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Im Anschluss fallen alle Spalten für die Wahlperioden mit Ausnahme des Targets weg."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Iteration über die DataFrames in df_wp_list\n",
    "\n",
    "for d in list(range(0,19)):\n",
    "    col_list = list(df_wp_list[d].columns.values)\n",
    "    del col_list[0:6]\n",
    "    col_list.pop()\n",
    "    df_wp_list[d][\"ERFAHRUNG\"] = df_wp_list[d].loc[:,col_list].sum(axis=1)\n",
    "    df_wp_list[d] = df_wp_list[d].drop(col_list, axis = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Korrektur (df_WP19 hat kein Target, aber die Letzte Spalte WP18 wurde mit .pop() aus praktischen Gründen aus der Liste entfernt \n",
    "# und nicht mit dem \"Erfahrungsschatz\" verrechnet)\n",
    "\n",
    "df_wp_list[18][\"ERFAHRUNG\"] = df_wp_list[18][\"ERFAHRUNG\"] + df_wp_list[18][\"WP18\"]\n",
    "df_wp_list[18] = df_wp_list[18].drop(\"WP18\", axis = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "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>ANFANG</th>\n",
       "      <th>ALTER</th>\n",
       "      <th>GESCHLECHT</th>\n",
       "      <th>PARTEI_KURZ</th>\n",
       "      <th>AUSLAND</th>\n",
       "      <th>MANDATSART</th>\n",
       "      <th>WP15</th>\n",
       "      <th>ERFAHRUNG</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>WP-JAHRE</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1998–2002;14</th>\n",
       "      <td>1998</td>\n",
       "      <td>48</td>\n",
       "      <td>1</td>\n",
       "      <td>Union</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1998–2002;14</th>\n",
       "      <td>1998</td>\n",
       "      <td>54</td>\n",
       "      <td>0</td>\n",
       "      <td>SPD</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1998–2002;14</th>\n",
       "      <td>1998</td>\n",
       "      <td>55</td>\n",
       "      <td>0</td>\n",
       "      <td>FDP</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1998–2002;14</th>\n",
       "      <td>1998</td>\n",
       "      <td>47</td>\n",
       "      <td>1</td>\n",
       "      <td>SPD</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1998–2002;14</th>\n",
       "      <td>1998</td>\n",
       "      <td>57</td>\n",
       "      <td>1</td>\n",
       "      <td>Union</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              ANFANG  ALTER  GESCHLECHT PARTEI_KURZ  AUSLAND  MANDATSART  \\\n",
       "WP-JAHRE                                                                   \n",
       "1998–2002;14    1998     48           1       Union        0         1.0   \n",
       "1998–2002;14    1998     54           0         SPD        0         0.0   \n",
       "1998–2002;14    1998     55           0         FDP        0         0.0   \n",
       "1998–2002;14    1998     47           1         SPD        0         1.0   \n",
       "1998–2002;14    1998     57           1       Union        0         0.0   \n",
       "\n",
       "              WP15  ERFAHRUNG  \n",
       "WP-JAHRE                       \n",
       "1998–2002;14     1          2  \n",
       "1998–2002;14     0          3  \n",
       "1998–2002;14     0          2  \n",
       "1998–2002;14     1          3  \n",
       "1998–2002;14     1          5  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_wp_list[13].head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Z-Transformation des Alters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "for d in list(range(0,19)):\n",
    "     df_wp_list[d][\"ALTER\"] = (df_wp_list[d][\"ALTER\"] -  df_wp_list[d][\"ALTER\"].mean())/df_wp_list[d][\"ALTER\"].std()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Transformieren der Spalte für Parteikürzel in einzelne, binäre Spalten"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Auch hier muss darauf geachtet werden, dass die DataFrames die gleiche Anzahl an Spalten haben. Eine einheitliche Liste von allen möglichen Parteien wird universell auf alle DataFrames angewandt. Auch wenn es einige Parteien in manchen Perioden noch nicht gab."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "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>ANFANG</th>\n",
       "      <th>ALTER</th>\n",
       "      <th>GESCHLECHT</th>\n",
       "      <th>PARTEI_KURZ</th>\n",
       "      <th>AUSLAND</th>\n",
       "      <th>MANDATSART</th>\n",
       "      <th>WP11</th>\n",
       "      <th>ERFAHRUNG</th>\n",
       "      <th>Union</th>\n",
       "      <th>SPD</th>\n",
       "      <th>FDP</th>\n",
       "      <th>Bündnis 90/Die Grünen</th>\n",
       "      <th>Die Linke</th>\n",
       "      <th>Andere</th>\n",
       "      <th>AfD</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>WP-JAHRE</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1983–1987;10</th>\n",
       "      <td>1983</td>\n",
       "      <td>0.575459</td>\n",
       "      <td>1</td>\n",
       "      <td>Union</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1983–1987;10</th>\n",
       "      <td>1983</td>\n",
       "      <td>1.287521</td>\n",
       "      <td>1</td>\n",
       "      <td>SPD</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1983–1987;10</th>\n",
       "      <td>1983</td>\n",
       "      <td>0.812813</td>\n",
       "      <td>1</td>\n",
       "      <td>Union</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1983–1987;10</th>\n",
       "      <td>1983</td>\n",
       "      <td>0.100752</td>\n",
       "      <td>1</td>\n",
       "      <td>SPD</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1983–1987;10</th>\n",
       "      <td>1983</td>\n",
       "      <td>-0.492633</td>\n",
       "      <td>1</td>\n",
       "      <td>SPD</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              ANFANG     ALTER  GESCHLECHT PARTEI_KURZ  AUSLAND  MANDATSART  \\\n",
       "WP-JAHRE                                                                      \n",
       "1983–1987;10    1983  0.575459           1       Union        0         1.0   \n",
       "1983–1987;10    1983  1.287521           1         SPD        0         0.0   \n",
       "1983–1987;10    1983  0.812813           1       Union        0         1.0   \n",
       "1983–1987;10    1983  0.100752           1         SPD        0         0.0   \n",
       "1983–1987;10    1983 -0.492633           1         SPD        0         0.0   \n",
       "\n",
       "              WP11  ERFAHRUNG  Union  SPD  FDP  Bündnis 90/Die Grünen  \\\n",
       "WP-JAHRE                                                                \n",
       "1983–1987;10     1          5      1    0    0                      0   \n",
       "1983–1987;10     1          4      0    1    0                      0   \n",
       "1983–1987;10     0          6      1    0    0                      0   \n",
       "1983–1987;10     1          3      0    1    0                      0   \n",
       "1983–1987;10     1          1      0    1    0                      0   \n",
       "\n",
       "             Die Linke Andere AfD  \n",
       "WP-JAHRE                           \n",
       "1983–1987;10         0      0   0  \n",
       "1983–1987;10         0      0   0  \n",
       "1983–1987;10         0      0   0  \n",
       "1983–1987;10         0      0   0  \n",
       "1983–1987;10         0      0   0  "
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Funktion\n",
    "def binary_transformer(dafr):\n",
    "    uni_lst = ['Union', 'SPD', 'FDP', 'Bündnis 90/Die Grünen', 'Die Linke', 'Andere', 'AfD']\n",
    "    for e in uni_lst:\n",
    "        dafr[e]= dafr[\"PARTEI_KURZ\"] == e\n",
    "        dafr[e].replace([True, False], [1,0], inplace = True)\n",
    "        \n",
    "# Anwendung\n",
    "for e in df_wp_list:\n",
    "    binary_transformer(e)\n",
    "    \n",
    "# Test an einem Beispiel\n",
    "df_wp_list[9].head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Entfernen der letzten überflüssigen Spalten"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "for e in list(range(0,19)):\n",
    "    drop = [\"ANFANG\", \"PARTEI_KURZ\"]\n",
    "    df_wp_list[e] = df_wp_list[e].drop(drop, axis = 1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Umindizieren"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "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>ALTER</th>\n",
       "      <th>GESCHLECHT</th>\n",
       "      <th>AUSLAND</th>\n",
       "      <th>MANDATSART</th>\n",
       "      <th>WP8</th>\n",
       "      <th>ERFAHRUNG</th>\n",
       "      <th>Union</th>\n",
       "      <th>SPD</th>\n",
       "      <th>FDP</th>\n",
       "      <th>Bündnis 90/Die Grünen</th>\n",
       "      <th>Die Linke</th>\n",
       "      <th>Andere</th>\n",
       "      <th>AfD</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-0.485003</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1.848721</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.737424</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.404034</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.181775</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      ALTER  GESCHLECHT  AUSLAND  MANDATSART  WP8  ERFAHRUNG  Union  SPD  FDP  \\\n",
       "0 -0.485003           1        0         1.0    1          2      1    0    0   \n",
       "1  1.848721           1        0         0.0    0          4      0    0    1   \n",
       "2  0.737424           1        0         1.0    1          2      0    1    0   \n",
       "3  0.404034           1        0         1.0    1          0      0    1    0   \n",
       "4  0.181775           1        0         1.0    1          1      0    1    0   \n",
       "\n",
       "  Bündnis 90/Die Grünen Die Linke Andere AfD  \n",
       "0                     0         0      0   0  \n",
       "1                     0         0      0   0  \n",
       "2                     0         0      0   0  \n",
       "3                     0         0      0   0  \n",
       "4                     0         0      0   0  "
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Nur der Ästhetik halber \n",
    "for e in df_wp_list:\n",
    "     e.index = range(len(list(e.index)))\n",
    "\n",
    "# Test an einem Beispiel\n",
    "df_wp_list[6].head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Machine Learning"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Importe"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Laszlo Kuehl\\Anaconda3\\lib\\site-packages\\sklearn\\cross_validation.py:41: DeprecationWarning: This module was deprecated in version 0.18 in favor of the model_selection module into which all the refactored classes and functions are moved. Also note that the interface of the new CV iterators are different from that of this module. This module will be removed in 0.20.\n",
      "  \"This module will be removed in 0.20.\", DeprecationWarning)\n",
      "C:\\Users\\Laszlo Kuehl\\Anaconda3\\lib\\site-packages\\sklearn\\ensemble\\weight_boosting.py:29: DeprecationWarning: numpy.core.umath_tests is an internal NumPy module and should not be imported. It will be removed in a future NumPy release.\n",
      "  from numpy.core.umath_tests import inner1d\n"
     ]
    }
   ],
   "source": [
    "\n",
    "from sklearn.cross_validation import cross_val_score\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "\n",
    "\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "\n",
    "\n",
    "from sklearn.gaussian_process import GaussianProcessClassifier\n",
    "from sklearn.gaussian_process.kernels import RBF\n",
    "\n",
    "\n",
    "from sklearn import neighbors\n",
    "\n",
    "from sklearn.svm import SVC\n",
    "\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "from sklearn.ensemble import VotingClassifier"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Modelle"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Decision Tree Classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[62.67, 3.31]\n",
      "[60.55, 5.56]\n",
      "[63.7, 3.19]\n",
      "[62.59, 2.54]\n",
      "[62.64, 4.45]\n",
      "[59.91, 3.76]\n",
      "[68.48, 3.68]\n",
      "[62.4, 4.29]\n",
      "[69.96, 4.0]\n",
      "[67.7, 2.82]\n",
      "[63.67, 2.66]\n",
      "[61.1, 4.41]\n",
      "[66.8, 3.52]\n",
      "[63.25, 6.91]\n",
      "[57.99, 8.93]\n",
      "[59.65, 4.43]\n",
      "[69.82, 12.73]\n",
      "[63.1, 4.79]\n"
     ]
    }
   ],
   "source": [
    "for i in list(range(0,18)):\n",
    "    df = df_wp_list[i]\n",
    "    wpnum = \"WP\" + str(int(i+2))\n",
    "    X, y = df.loc[:, df.columns != wpnum], df[wpnum]\n",
    "    dtc = DecisionTreeClassifier()\n",
    "    scores = cross_val_score(dtc, X,y, cv = 5, scoring =\"accuracy\")\n",
    "    print([round(np.mean(scores)*100, 2), round(np.std(scores)*100,2)])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Logistic Regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[72.15, 3.59]\n",
      "[71.85, 2.33]\n",
      "[73.13, 2.72]\n",
      "[74.48, 3.65]\n",
      "[76.03, 1.99]\n",
      "[73.21, 3.22]\n",
      "[73.58, 1.99]\n",
      "[76.14, 1.49]\n",
      "[77.79, 1.76]\n",
      "[77.96, 1.98]\n",
      "[68.24, 2.91]\n",
      "[69.4, 4.54]\n",
      "[75.76, 2.4]\n",
      "[69.68, 8.0]\n",
      "[72.93, 1.61]\n",
      "[68.23, 3.17]\n",
      "[77.17, 7.84]\n",
      "[69.18, 5.57]\n"
     ]
    }
   ],
   "source": [
    "for i in list(range(0,18)):\n",
    "    df = df_wp_list[i]\n",
    "    wpnum = \"WP\" + str(int(i+2))\n",
    "    X, y = df.loc[:, df.columns != wpnum], df[wpnum]\n",
    "    log = LogisticRegression()\n",
    "    scores = cross_val_score(log, X,y, cv = 5, scoring =\"accuracy\")\n",
    "    print([round(np.mean(scores)*100, 2), round(np.std(scores)*100,2)])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Random Forest Classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[66.87, 1.79]\n",
      "[64.32, 3.34]\n",
      "[66.36, 2.74]\n",
      "[66.91, 3.1]\n",
      "[67.8, 3.06]\n",
      "[67.45, 1.49]\n",
      "[67.57, 2.56]\n",
      "[70.16, 4.16]\n",
      "[73.79, 3.42]\n",
      "[74.13, 1.44]\n",
      "[67.94, 2.05]\n",
      "[65.82, 4.31]\n",
      "[71.56, 5.28]\n",
      "[65.54, 6.68]\n",
      "[61.65, 7.46]\n",
      "[60.59, 2.79]\n",
      "[71.96, 11.56]\n",
      "[62.96, 7.77]\n"
     ]
    }
   ],
   "source": [
    "for i in list(range(0,18)):\n",
    "    df = df_wp_list[i]\n",
    "    wpnum = \"WP\" + str(int(i+2))\n",
    "    X, y = df.loc[:, df.columns != wpnum], df[wpnum]\n",
    "    rfc = RandomForestClassifier()\n",
    "    scores = cross_val_score(rfc, X,y, cv = 5, scoring =\"accuracy\")\n",
    "    print([round(np.mean(scores)*100, 2), round(np.std(scores)*100,2)])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### KNeighbors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[70.25, 4.94]\n",
      "[66.99, 5.09]\n",
      "[68.51, 2.46]\n",
      "[68.8, 2.77]\n",
      "[73.7, 2.04]\n",
      "[72.11, 3.02]\n",
      "[70.67, 3.91]\n",
      "[72.68, 2.5]\n",
      "[78.33, 2.51]\n",
      "[75.01, 3.7]\n",
      "[69.23, 3.57]\n",
      "[65.11, 4.24]\n",
      "[76.34, 2.34]\n",
      "[63.82, 5.03]\n",
      "[61.82, 10.78]\n",
      "[62.31, 2.0]\n",
      "[73.96, 10.82]\n",
      "[67.51, 6.17]\n"
     ]
    }
   ],
   "source": [
    "for i in list(range(0,18)):\n",
    "    df = df_wp_list[i]\n",
    "    wpnum = \"WP\" + str(int(i+2))\n",
    "    X, y = df.loc[:, df.columns != wpnum], df[wpnum]\n",
    "    knn = neighbors.KNeighborsClassifier(n_neighbors=5)\n",
    "    scores = cross_val_score(knn, X,y, cv = 5, scoring =\"accuracy\")\n",
    "    print([round(np.mean(scores)*100, 2), round(np.std(scores)*100,2)])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### GaussianProcessClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[71.95, 2.18]\n",
      "[71.32, 1.62]\n",
      "[73.31, 0.49]\n",
      "[76.04, 3.12]\n",
      "[74.06, 0.8]\n",
      "[73.92, 3.53]\n",
      "[74.86, 1.29]\n",
      "[74.68, 1.67]\n",
      "[79.42, 0.59]\n",
      "[77.61, 1.06]\n",
      "[71.65, 1.69]\n",
      "[68.39, 0.92]\n",
      "[76.77, 0.66]\n",
      "[69.25, 4.09]\n",
      "[56.55, 12.11]\n",
      "[59.54, 12.86]\n",
      "[71.5, 12.84]\n",
      "[61.88, 11.32]\n"
     ]
    }
   ],
   "source": [
    "for i in list(range(0,18)):\n",
    "    df = df_wp_list[i]\n",
    "    wpnum = \"WP\" + str(int(i+2))\n",
    "    X, y = df.loc[:, df.columns != wpnum], df[wpnum]\n",
    "    gpc = GaussianProcessClassifier(random_state=0, kernel= 1.0*RBF(2.0))\n",
    "    scores = cross_val_score(gpc, X,y, cv = 3, scoring =\"accuracy\")\n",
    "    print([round(np.mean(scores)*100, 2), round(np.std(scores)*100,2)])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Entgegen der anfänglichen Feststellung, dass ein Gaussian Process Classifier die besten Ergebnisse bei der niedrigsten Standardabweichung liefert, hat sich schlussendlich die Logistic Regression als Modell der Wahl herausgestellt. Gerade die Einfachheit und Schnelligkeit machen es simpler dieses Modell zu benutzen. Vermutlich (hoffentlich) ist das Overfitting auch nicht allzu groß, doch dazu weiter unten mehr. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Der größte Faktor bei der Verbesserung dieser Modelle (allesamt) ist die Vereinfachung der Spalten gewesen. Die Spalte \"Erfahrung\" fasst die Wahlperioden zusammen und reduziert somit gerade in späteren Perioden die Dimension der DataFrames. Am erstaunlichsten ist sicherlich die Verbesserung der Standardabweichung (Reduktion ~ bessere Reproduktion der Ergebnisse bei der Validierung).    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Einzig der 14. Bundesatg (1998-2002) scheint das Modell vor Probleme zu stellen."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Plotting"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Eine durchschnittliche Rate für das Überstehen der Wahl als jeweils derzeitiges MdB wird als Vergleich genommen "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Hier als Überlebensrate geframet\n",
    "surv_rate_lst = [] \n",
    "\n",
    "for d in list(range(0,18)):\n",
    "    wpnum = \"WP\" + str(d+2)\n",
    "    surv_rate = df_wp_list[d][wpnum].sum() / len(df_wp_list[d])\n",
    "    surv_rate = round(surv_rate*100, 2)\n",
    "    surv_rate_lst.append(surv_rate)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "scrolled": true
   },
   "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": [
    "# Veranschaulichung der Ergebnisse mittels eines Plots \n",
    "\n",
    "log_score_list = [[72.15, 3.59],\n",
    "[71.85, 2.33],\n",
    "[73.13, 2.72],\n",
    "[74.48, 3.65],\n",
    "[76.03, 1.99],\n",
    "[73.21, 3.22],\n",
    "[73.58, 1.99],\n",
    "[76.14, 1.49],\n",
    "[77.79, 1.76],\n",
    "[77.96, 1.98],\n",
    "[68.24, 2.91],\n",
    "[69.4, 4.54],\n",
    "[75.76, 2.4],\n",
    "[69.68, 8.0],\n",
    "[72.93, 1.61],\n",
    "[68.23, 3.17],\n",
    "[77.17, 7.84],[69.18, 5.57]]\n",
    "\n",
    "means = []\n",
    "stds = []\n",
    "for e in log_score_list:\n",
    "    means.append(e[0])\n",
    "    stds.append(e[1])\n",
    "\n",
    "#Blau\n",
    "plt.plot(timeline[:18], surv_rate_lst)\n",
    "#Orange\n",
    "plt.plot(timeline[:18], means)\n",
    "#Grün\n",
    "plt.plot(timeline[:18], stds)\n",
    "plt.title('Durchschnitt und Standardabweichung des Modells bei der Validierung')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Fast konstant befindet sich das Modell im Punkt Akkuratheit (accuracy) über dem simplen Durchschnitt. Dies ist zunächst einmal positiv zu bewerten."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Modell bei Anwendung auf ungesehene Daten"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Hier wird jeweils immer auf eine Periode trainiert und auf die Nächste vorhergesagt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "acc_score = []\n",
    "\n",
    "for i in list(range(0,17)):\n",
    "    df = df_wp_list[i]\n",
    "    df2 = df_wp_list[i+1]\n",
    "    wpnum = \"WP\" + str(int(i+2))\n",
    "    wpnum2 = \"WP\" + str(int(i+3))\n",
    "    X_train, y_train, X_test, y_test = df.loc[:, df.columns != wpnum], df[wpnum], df2.loc[:, df2.columns != wpnum2], df2[wpnum2]  \n",
    "    log = LogisticRegression()\n",
    "    log.fit(X_train,y_train)\n",
    "    y_pred = log.predict(X_test)\n",
    "    acc_score.append(round(accuracy_score(y_test, y_pred)*100,2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "scrolled": false
   },
   "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": [
    "#Blau\n",
    "plt.plot(timeline[:18], surv_rate_lst)\n",
    "#Orange\n",
    "plt.plot(timeline[1:18], acc_score)\n",
    "\n",
    "plt.title('Akkuratheit des Modells bei ungesehenen Daten')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Obwohl die unterliegenden Daten ein hohes Maß an Zufall implizieren, kann man hier sehen, dass ein ML-Modell (hier LogReg) doch gut performt und adäquate Ergebnisse liefert. (Der Vergleich mit dem Durchschnitt hinkt hier natürlich etwas, aber man sieht, dass ein Modell in dem alle \"weiterkommen\" oft nicht besser ist als die LogReg und weniger konsistent sein würde)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Das Modell könnte wahrscheinlich noch bessere Ergebnisse liefern, wenn alle MdBs für die jeweilige Wahlperiode aussortiert werden, die vorzeitig ausscheiden. Diese Rücktritte ziehen vermutlich die Performance des Modells herunter, da sie oft auch aus Gründen vollzogen werden, die aus Daten alleine (z.B. Alter) nicht allein abzulesen sind (bspw. private Gründe). Diese etwas mühselige Handarbeit haben wir jedoch gescheut.  Möglicherweise lässt sich hier mit WebScraping schneller arbeiten(?). "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Unsere Vorhersage für Bundestag Nr. 20"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "X_train, y_train, X_test = df_wp_list[17].loc[:, df_wp_list[17].columns != \"WP19\"], df_wp_list[17].loc[:,\"WP19\"], df_wp_list[18]  \n",
    "log = LogisticRegression()\n",
    "\n",
    "log.fit(X_train, y_train)\n",
    "y_pred = log.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "prediction = IDs.loc[IDs[\"ANFANG\"] == 2017, [\"VORNAME\", \"NACHNAME\"]]\n",
    "prediction[\"WP20\"] = y_pred\n",
    "prediction.index = list(range(0,711))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "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>VORNAME</th>\n",
       "      <th>NACHNAME</th>\n",
       "      <th>WP20</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Peter</td>\n",
       "      <td>Bleser</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Jörg-Diether</td>\n",
       "      <td>Dehm-Desoi</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Hans-Joachim</td>\n",
       "      <td>Fuchtel</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Gregor</td>\n",
       "      <td>Gysi</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Ulla</td>\n",
       "      <td>Jelpke</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Volker</td>\n",
       "      <td>Kauder</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Wolfgang</td>\n",
       "      <td>Kubicki</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Katharina</td>\n",
       "      <td>Landgraf</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Christoph</td>\n",
       "      <td>Matschie</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>Angela</td>\n",
       "      <td>Merkel</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Peter</td>\n",
       "      <td>Ramsauer</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>Wolfgang</td>\n",
       "      <td>Schäuble</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>Christian</td>\n",
       "      <td>Schmidt</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>Ulla</td>\n",
       "      <td>Schmidt</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>Hermann Otto</td>\n",
       "      <td>Solms</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>Michael</td>\n",
       "      <td>Stübgen</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>Peter</td>\n",
       "      <td>Altmaier</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>Doris</td>\n",
       "      <td>Barnett</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>Dagmar</td>\n",
       "      <td>Freitag</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>Hermann</td>\n",
       "      <td>Gröhe</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>Manfred</td>\n",
       "      <td>Grund</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>Barbara</td>\n",
       "      <td>Hendricks</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>Karl A.</td>\n",
       "      <td>Lamers</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>Steffi</td>\n",
       "      <td>Lemke</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>Michael</td>\n",
       "      <td>Meister</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>Hans</td>\n",
       "      <td>Michelbach</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>Gerd</td>\n",
       "      <td>Müller</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>Cem</td>\n",
       "      <td>Özdemir</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>Thomas</td>\n",
       "      <td>Rachel</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>Norbert</td>\n",
       "      <td>Röttgen</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>681</th>\n",
       "      <td>Marie-Agnes</td>\n",
       "      <td>Strack-Zimmermann</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>682</th>\n",
       "      <td>Friedrich</td>\n",
       "      <td>Straetmanns</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>683</th>\n",
       "      <td>Benjamin</td>\n",
       "      <td>Strasser</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>684</th>\n",
       "      <td>Margit</td>\n",
       "      <td>Stumpp</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>685</th>\n",
       "      <td>Katja</td>\n",
       "      <td>Suding</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>686</th>\n",
       "      <td>Jessica</td>\n",
       "      <td>Tatti</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>687</th>\n",
       "      <td>Hermann-Josef</td>\n",
       "      <td>Tebroke</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>688</th>\n",
       "      <td>Linda</td>\n",
       "      <td>Teuteberg</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>689</th>\n",
       "      <td>Michael</td>\n",
       "      <td>Theurer</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>690</th>\n",
       "      <td>Hans-Jürgen</td>\n",
       "      <td>Thies</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>691</th>\n",
       "      <td>Alexander</td>\n",
       "      <td>Throm</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>692</th>\n",
       "      <td>Dietlind</td>\n",
       "      <td>Tiemann</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>693</th>\n",
       "      <td>Markus</td>\n",
       "      <td>Töns</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>694</th>\n",
       "      <td>Andrew</td>\n",
       "      <td>Ullmann</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>695</th>\n",
       "      <td>Gerald</td>\n",
       "      <td>Ullrich</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>696</th>\n",
       "      <td>Kerstin</td>\n",
       "      <td>Vieregge</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>697</th>\n",
       "      <td>Christoph</td>\n",
       "      <td>Vries</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>698</th>\n",
       "      <td>Andreas</td>\n",
       "      <td>Wagner</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>699</th>\n",
       "      <td>Sandra</td>\n",
       "      <td>Weeser</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>700</th>\n",
       "      <td>Alice</td>\n",
       "      <td>Weidel</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>701</th>\n",
       "      <td>Nicole</td>\n",
       "      <td>Westig</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>702</th>\n",
       "      <td>Harald</td>\n",
       "      <td>Weyel</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>703</th>\n",
       "      <td>Wolfgang</td>\n",
       "      <td>Wiehle</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>704</th>\n",
       "      <td>Bettina Margarethe</td>\n",
       "      <td>Wiesmann</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>705</th>\n",
       "      <td>Heiko</td>\n",
       "      <td>Wildberg</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>706</th>\n",
       "      <td>Christian</td>\n",
       "      <td>Wirth</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>707</th>\n",
       "      <td>Uwe</td>\n",
       "      <td>Witt</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>708</th>\n",
       "      <td>Paul</td>\n",
       "      <td>Ziemiak</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>709</th>\n",
       "      <td>Gyde</td>\n",
       "      <td>Jensen</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>710</th>\n",
       "      <td>Marja-Liisa</td>\n",
       "      <td>Völlers</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>711 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                VORNAME           NACHNAME  WP20\n",
       "0                 Peter             Bleser     0\n",
       "1          Jörg-Diether         Dehm-Desoi     0\n",
       "2          Hans-Joachim            Fuchtel     0\n",
       "3                Gregor               Gysi     1\n",
       "4                  Ulla             Jelpke     0\n",
       "5                Volker             Kauder     0\n",
       "6              Wolfgang            Kubicki     0\n",
       "7             Katharina           Landgraf     0\n",
       "8             Christoph           Matschie     0\n",
       "9                Angela             Merkel     0\n",
       "10                Peter           Ramsauer     0\n",
       "11             Wolfgang           Schäuble     0\n",
       "12            Christian            Schmidt     0\n",
       "13                 Ulla            Schmidt     0\n",
       "14         Hermann Otto              Solms     0\n",
       "15              Michael            Stübgen     1\n",
       "16                Peter           Altmaier     1\n",
       "17                Doris            Barnett     0\n",
       "18               Dagmar            Freitag     1\n",
       "19              Hermann              Gröhe     1\n",
       "20              Manfred              Grund     0\n",
       "21              Barbara          Hendricks     0\n",
       "22              Karl A.             Lamers     0\n",
       "23               Steffi              Lemke     1\n",
       "24              Michael            Meister     1\n",
       "25                 Hans         Michelbach     0\n",
       "26                 Gerd             Müller     0\n",
       "27                  Cem            Özdemir     1\n",
       "28               Thomas             Rachel     1\n",
       "29              Norbert            Röttgen     1\n",
       "..                  ...                ...   ...\n",
       "681         Marie-Agnes  Strack-Zimmermann     0\n",
       "682           Friedrich        Straetmanns     1\n",
       "683            Benjamin           Strasser     1\n",
       "684              Margit             Stumpp     1\n",
       "685               Katja             Suding     1\n",
       "686             Jessica              Tatti     1\n",
       "687       Hermann-Josef            Tebroke     1\n",
       "688               Linda          Teuteberg     1\n",
       "689             Michael            Theurer     1\n",
       "690         Hans-Jürgen              Thies     1\n",
       "691           Alexander              Throm     1\n",
       "692            Dietlind            Tiemann     0\n",
       "693              Markus               Töns     1\n",
       "694              Andrew            Ullmann     1\n",
       "695              Gerald            Ullrich     0\n",
       "696             Kerstin           Vieregge     1\n",
       "697           Christoph              Vries     1\n",
       "698             Andreas             Wagner     1\n",
       "699              Sandra             Weeser     1\n",
       "700               Alice             Weidel     1\n",
       "701              Nicole             Westig     1\n",
       "702              Harald              Weyel     0\n",
       "703            Wolfgang             Wiehle     1\n",
       "704  Bettina Margarethe           Wiesmann     1\n",
       "705               Heiko           Wildberg     0\n",
       "706           Christian              Wirth     1\n",
       "707                 Uwe               Witt     0\n",
       "708                Paul            Ziemiak     1\n",
       "709                Gyde             Jensen     1\n",
       "710         Marja-Liisa            Völlers     1\n",
       "\n",
       "[711 rows x 3 columns]"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "prediction"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Et voilà"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "prediction.to_csv(\"20_Bundestag.csv\")"
   ]
  }
 ],
 "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": "3.7.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}