Predicting Bike count in Munich.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Predicting Bike count in Munich\n",
    "\n",
    "To get some pratical experience I chose my own project to practice the concepts introduced in Chapter 9 \"Tabular Modelling Deep Dive\" of the fastai_v2 course. I wanted to share my approach and would be happy if you could point me to errors, misunderstandings or possible improvements.\n",
    "\n",
    "1. [Goal](#introduction)\n",
    "2. [Data](#data)\n",
    "    1. [Get data](#getdata)\n",
    "    1. [Explore data](#exploredata)\n",
    "3. [Modelling](#model)\n",
    "    1. [Random Forest](#rf)\n",
    "    2. [Neuronal Net](#nn)    \n",
    "4. [Analyze results](#analyze)\n",
    "5. [Improvements](#improve)\n",
    "6. [Conclusion and next steps](#conclusion)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. My Goal <a name=\"introduction\"></a>\n",
    "\n",
    "The city of munich counts the number of bikers on six counting stations throughout munich. They provide those numbers to the public in their open data portal.\n",
    "\n",
    "Based on this data, can I create a model that predicts the number of bikers for a given day?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. The data <a name=\"data\"></a>\n",
    "\n",
    "At the momente there are 6 permanent bike counting stations in munich. They count bikes with sensors build into the street. The data can be accessed via the open data portal api. e.g. all currently available datasets can be retrieved with this call: https://www.opengov-muenchen.de/api/3/action/package_search?q=Raddauerz%C3%A4hlstellen&rows=1000\n",
    "\n",
    "In April 2020 the counting station Kreuther changed. The sensor size approximatly doubled in size. Take this into account for data after 1. april.\n",
    "\n",
    "More information:\n",
    "\n",
    "* OpenDataPortal Munich Website (german): https://www.opengov-muenchen.de/dataset?tags=Raddauerz%C3%A4hlstellen\n",
    "* Info about bike counting stations (Raddauerzählstellen) (german): https://www.opengov-muenchen.de/pages/raddauerzaehlstellen\n",
    "* API Doku: https://docs.ckan.org/en/ckan-2.5.3/api/\n",
    "* Analysis of data 2017 & 2018 in german - [„Aufs Radl – Fertig? – Los!“ Ergebnisse der Raddauerzählstellen in München 2017 und 2018 (PDF)](https://www.muenchen.de/rathaus/dam/jcr:9a65625e-952f-470f-b6a1-d4270f4526cb/mb190304.pdf)\n",
    "\n",
    "\n",
    "### 2.1 Get data <a name=\"getdata\"></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import requests\n",
    "import os\n",
    "import itertools"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "r = requests.get('https://www.opengov-muenchen.de/api/3/action/package_search?q=Raddauerz%C3%A4hlstellen&rows=1000')\n",
    "json = r.json()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "#json"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def download_csv(url):\n",
    "    filename = 'data/' + url.rsplit('/',1)[1]\n",
    "    if os.path.exists(filename):\n",
    "        print(f'{filename} already exists')\n",
    "    else:\n",
    "        r = requests.get(url)\n",
    "        os.makedirs(os.path.dirname(filename), exist_ok=True)\n",
    "        with open(filename, 'wb') as f:\n",
    "            f.write(r.content)\n",
    "        print(f'{filename} downloaded.')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "resources = [[{'name':res['name'], 'url':res['url']} for res in result['resources'] if res['name'] != 'Raddauerzählstellen in München'] for result in json['result']['results']]\n",
    "resources = list(itertools.chain.from_iterable(resources))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "data/rad20170615min.csv already exists\n",
      "data/rad201706tage.csv already exists\n",
      "data/rad20201015min.csv already exists\n",
      "data/rad202010tage.csv already exists\n",
      "data/rad20200915min.csv already exists\n",
      "data/rad202009tage.csv already exists\n",
      "data/rad20200815min.csv already exists\n",
      "data/rad202008tage.csv already exists\n",
      "data/rad20200715min.csv already exists\n",
      "data/rad202007tage.csv already exists\n",
      "data/rad20200615min.csv already exists\n",
      "data/rad202006tage.csv already exists\n",
      "data/rad20200115min.csv already exists\n",
      "data/rad202001tage.csv already exists\n",
      "data/rad20200315min.csv already exists\n",
      "data/rad202003tage.csv already exists\n",
      "data/rad20200215min.csv already exists\n",
      "data/rad202002tage.csv already exists\n",
      "data/rad20190415min.csv already exists\n",
      "data/rad201904tage.csv already exists\n",
      "data/rad20181115min.csv already exists\n",
      "data/rad201811tage.csv already exists\n",
      "data/rad20170715min.csv already exists\n",
      "data/rad201707tage.csv already exists\n",
      "data/rad20170115min.csv already exists\n",
      "data/rad201701tage.csv already exists\n",
      "data/rad20200415min.csv already exists\n",
      "data/rad202004tage.csv already exists\n",
      "data/rad20191215min.csv already exists\n",
      "data/rad201912tage.csv already exists\n",
      "data/rad20191115min.csv already exists\n",
      "data/rad201911tage.csv already exists\n",
      "data/rad20191015min.csv already exists\n",
      "data/rad201910tage.csv already exists\n",
      "data/rad20190915min.csv already exists\n",
      "data/rad201909tage.csv already exists\n",
      "data/rad20190815min.csv already exists\n",
      "data/rad201908tage.csv already exists\n",
      "data/rad20190715min.csv already exists\n",
      "data/rad201907tageneu.csv already exists\n",
      "data/rad20190515min.csv already exists\n",
      "data/rad201905tage.csv already exists\n",
      "data/rad20190315min.csv already exists\n",
      "data/rad201903tage.csv already exists\n",
      "data/rad20190215min.csv already exists\n",
      "data/rad201902tage.csv already exists\n",
      "data/rad20190115min.csv already exists\n",
      "data/rad201901tage.csv already exists\n",
      "data/rad20180315min.csv already exists\n",
      "data/rad201803tage.csv already exists\n",
      "data/rad20181215min.csv already exists\n",
      "data/rad201812tage.csv already exists\n",
      "data/rad20180915min.csv already exists\n",
      "data/rad201809tage.csv already exists\n",
      "data/rad20181015min.csv already exists\n",
      "data/rad201810tage.csv already exists\n",
      "data/rad20170315min.csv already exists\n",
      "data/rad201703tage.csv already exists\n",
      "data/rad20170815min.csv already exists\n",
      "data/rad201708tage.csv already exists\n",
      "data/rad20180715min.csv already exists\n",
      "data/rad201807tage.csv already exists\n",
      "data/rad20171215min.csv already exists\n",
      "data/rad201712tage.csv already exists\n",
      "data/rad20171115min.csv already exists\n",
      "data/rad201711tage.csv already exists\n",
      "data/rad20171015min.csv already exists\n",
      "data/rad201710tage.csv already exists\n",
      "data/rad20170915min.csv already exists\n",
      "data/rad201709tage.csv already exists\n",
      "data/rad201702tage.csv already exists\n",
      "data/rad20170215min.csv already exists\n",
      "data/rad20170415min.csv already exists\n",
      "data/rad201704tage.csv already exists\n",
      "data/rad20170515min.csv already exists\n",
      "data/rad201705tage.csv already exists\n",
      "data/rad20180215min.csv already exists\n",
      "data/rad201802tage.csv already exists\n",
      "data/rad20180115min.csv already exists\n",
      "data/rad201801tage.csv already exists\n",
      "data/rad20180415min.csv already exists\n",
      "data/rad201804tage.csv already exists\n",
      "data/rad20180515min.csv already exists\n",
      "data/rad201805tage.csv already exists\n",
      "data/rad20180615min.csv already exists\n",
      "data/rad201806tage.csv already exists\n",
      "data/rad20180815min.csv already exists\n",
      "data/rad201808tage.csv already exists\n",
      "data/rad20200515min.csv already exists\n",
      "data/rad202005tage.csv already exists\n",
      "data/rad20190615min.csv already exists\n",
      "data/rad201906tageneu.csv already exists\n"
     ]
    }
   ],
   "source": [
    "for res in resources:\n",
    "    download_csv(res['url'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.2. Explore data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# pip install pandas"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "files = [f for f in os.listdir('data') if os.path.isfile(os.path.join('data', f))]\n",
    "files.sort()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Data complete?\n",
    "For each month there should be two csv files (15 min und day). Therefore there should 92 files `(3 years * 12 month + 10 (10 for october)) * 2`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(files) == (3 * 12 + 10) * 2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### How does the data look like?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "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>datum</th>\n",
       "      <th>uhrzeit_start</th>\n",
       "      <th>uhrzeit_ende</th>\n",
       "      <th>zaehlstelle</th>\n",
       "      <th>richtung_1</th>\n",
       "      <th>richtung_2</th>\n",
       "      <th>gesamt</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2020.10.01</td>\n",
       "      <td>00:00</td>\n",
       "      <td>00:15</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2020.10.01</td>\n",
       "      <td>00:15</td>\n",
       "      <td>00:30</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2020.10.01</td>\n",
       "      <td>00:30</td>\n",
       "      <td>00:45</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2020.10.01</td>\n",
       "      <td>00:45</td>\n",
       "      <td>01:00</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2020.10.01</td>\n",
       "      <td>01:00</td>\n",
       "      <td>01:15</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17851</th>\n",
       "      <td>2020.10.31</td>\n",
       "      <td>22:45</td>\n",
       "      <td>23:00</td>\n",
       "      <td>Erhardt</td>\n",
       "      <td>5</td>\n",
       "      <td>14</td>\n",
       "      <td>19</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17852</th>\n",
       "      <td>2020.10.31</td>\n",
       "      <td>23:00</td>\n",
       "      <td>23:15</td>\n",
       "      <td>Erhardt</td>\n",
       "      <td>8</td>\n",
       "      <td>11</td>\n",
       "      <td>19</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17853</th>\n",
       "      <td>2020.10.31</td>\n",
       "      <td>23:15</td>\n",
       "      <td>23:30</td>\n",
       "      <td>Erhardt</td>\n",
       "      <td>5</td>\n",
       "      <td>9</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17854</th>\n",
       "      <td>2020.10.31</td>\n",
       "      <td>23:30</td>\n",
       "      <td>23:45</td>\n",
       "      <td>Erhardt</td>\n",
       "      <td>5</td>\n",
       "      <td>9</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17855</th>\n",
       "      <td>2020.10.31</td>\n",
       "      <td>23:45</td>\n",
       "      <td>00:00</td>\n",
       "      <td>Erhardt</td>\n",
       "      <td>5</td>\n",
       "      <td>8</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>17856 rows × 7 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            datum uhrzeit_start uhrzeit_ende zaehlstelle  richtung_1  \\\n",
       "0      2020.10.01         00:00        00:15      Arnulf           2   \n",
       "1      2020.10.01         00:15        00:30      Arnulf           1   \n",
       "2      2020.10.01         00:30        00:45      Arnulf           0   \n",
       "3      2020.10.01         00:45        01:00      Arnulf           0   \n",
       "4      2020.10.01         01:00        01:15      Arnulf           1   \n",
       "...           ...           ...          ...         ...         ...   \n",
       "17851  2020.10.31         22:45        23:00     Erhardt           5   \n",
       "17852  2020.10.31         23:00        23:15     Erhardt           8   \n",
       "17853  2020.10.31         23:15        23:30     Erhardt           5   \n",
       "17854  2020.10.31         23:30        23:45     Erhardt           5   \n",
       "17855  2020.10.31         23:45        00:00     Erhardt           5   \n",
       "\n",
       "       richtung_2  gesamt  \n",
       "0               0       2  \n",
       "1               0       1  \n",
       "2               0       0  \n",
       "3               0       0  \n",
       "4               0       1  \n",
       "...           ...     ...  \n",
       "17851          14      19  \n",
       "17852          11      19  \n",
       "17853           9      14  \n",
       "17854           9      14  \n",
       "17855           8      13  \n",
       "\n",
       "[17856 rows x 7 columns]"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_min15 = pd.read_csv('data/rad20201015min.csv')\n",
    "df_min15"
   ]
  },
  {
   "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>datum</th>\n",
       "      <th>uhrzeit_start</th>\n",
       "      <th>uhrzeit_ende</th>\n",
       "      <th>zaehlstelle</th>\n",
       "      <th>richtung_1</th>\n",
       "      <th>richtung_2</th>\n",
       "      <th>gesamt</th>\n",
       "      <th>min-temp</th>\n",
       "      <th>max-temp</th>\n",
       "      <th>niederschlag</th>\n",
       "      <th>bewoelkung</th>\n",
       "      <th>sonnenstunden</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2020.10.01</td>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>1738</td>\n",
       "      <td>107</td>\n",
       "      <td>1845</td>\n",
       "      <td>6.5</td>\n",
       "      <td>19.4</td>\n",
       "      <td>0.0</td>\n",
       "      <td>65</td>\n",
       "      <td>8.2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2020.10.02</td>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>1469</td>\n",
       "      <td>89</td>\n",
       "      <td>1558</td>\n",
       "      <td>9.8</td>\n",
       "      <td>18.4</td>\n",
       "      <td>0.0</td>\n",
       "      <td>99</td>\n",
       "      <td>0.9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2020.10.03</td>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>548</td>\n",
       "      <td>37</td>\n",
       "      <td>585</td>\n",
       "      <td>5.1</td>\n",
       "      <td>22.1</td>\n",
       "      <td>2.9</td>\n",
       "      <td>75</td>\n",
       "      <td>2.9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2020.10.04</td>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>821</td>\n",
       "      <td>37</td>\n",
       "      <td>858</td>\n",
       "      <td>4.7</td>\n",
       "      <td>17.3</td>\n",
       "      <td>0.7</td>\n",
       "      <td>73</td>\n",
       "      <td>6.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2020.10.05</td>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>1048</td>\n",
       "      <td>63</td>\n",
       "      <td>1111</td>\n",
       "      <td>9.0</td>\n",
       "      <td>14.6</td>\n",
       "      <td>4.7</td>\n",
       "      <td>98</td>\n",
       "      <td>1.8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>181</th>\n",
       "      <td>2020.10.27</td>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Erhardt</td>\n",
       "      <td>2289</td>\n",
       "      <td>2213</td>\n",
       "      <td>4502</td>\n",
       "      <td>3.9</td>\n",
       "      <td>11.1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>75</td>\n",
       "      <td>1.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>182</th>\n",
       "      <td>2020.10.28</td>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Erhardt</td>\n",
       "      <td>2675</td>\n",
       "      <td>2592</td>\n",
       "      <td>5267</td>\n",
       "      <td>3.5</td>\n",
       "      <td>14.7</td>\n",
       "      <td>2.4</td>\n",
       "      <td>95</td>\n",
       "      <td>1.3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>183</th>\n",
       "      <td>2020.10.29</td>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Erhardt</td>\n",
       "      <td>1334</td>\n",
       "      <td>1342</td>\n",
       "      <td>2676</td>\n",
       "      <td>8.9</td>\n",
       "      <td>11.2</td>\n",
       "      <td>15.5</td>\n",
       "      <td>94</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>184</th>\n",
       "      <td>2020.10.30</td>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Erhardt</td>\n",
       "      <td>2167</td>\n",
       "      <td>2142</td>\n",
       "      <td>4309</td>\n",
       "      <td>9.0</td>\n",
       "      <td>14.7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>99</td>\n",
       "      <td>0.6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>185</th>\n",
       "      <td>2020.10.31</td>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Erhardt</td>\n",
       "      <td>2780</td>\n",
       "      <td>2621</td>\n",
       "      <td>5401</td>\n",
       "      <td>6.3</td>\n",
       "      <td>17.9</td>\n",
       "      <td>0.0</td>\n",
       "      <td>55</td>\n",
       "      <td>8.9</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>186 rows × 12 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          datum uhrzeit_start  uhrzeit_ende zaehlstelle  richtung_1  \\\n",
       "0    2020.10.01         00:00         23.59      Arnulf        1738   \n",
       "1    2020.10.02         00:00         23.59      Arnulf        1469   \n",
       "2    2020.10.03         00:00         23.59      Arnulf         548   \n",
       "3    2020.10.04         00:00         23.59      Arnulf         821   \n",
       "4    2020.10.05         00:00         23.59      Arnulf        1048   \n",
       "..          ...           ...           ...         ...         ...   \n",
       "181  2020.10.27         00:00         23.59     Erhardt        2289   \n",
       "182  2020.10.28         00:00         23.59     Erhardt        2675   \n",
       "183  2020.10.29         00:00         23.59     Erhardt        1334   \n",
       "184  2020.10.30         00:00         23.59     Erhardt        2167   \n",
       "185  2020.10.31         00:00         23.59     Erhardt        2780   \n",
       "\n",
       "     richtung_2  gesamt  min-temp  max-temp  niederschlag  bewoelkung  \\\n",
       "0           107    1845       6.5      19.4           0.0          65   \n",
       "1            89    1558       9.8      18.4           0.0          99   \n",
       "2            37     585       5.1      22.1           2.9          75   \n",
       "3            37     858       4.7      17.3           0.7          73   \n",
       "4            63    1111       9.0      14.6           4.7          98   \n",
       "..          ...     ...       ...       ...           ...         ...   \n",
       "181        2213    4502       3.9      11.1           0.0          75   \n",
       "182        2592    5267       3.5      14.7           2.4          95   \n",
       "183        1342    2676       8.9      11.2          15.5          94   \n",
       "184        2142    4309       9.0      14.7           0.0          99   \n",
       "185        2621    5401       6.3      17.9           0.0          55   \n",
       "\n",
       "     sonnenstunden  \n",
       "0              8.2  \n",
       "1              0.9  \n",
       "2              2.9  \n",
       "3              6.5  \n",
       "4              1.8  \n",
       "..             ...  \n",
       "181            1.5  \n",
       "182            1.3  \n",
       "183            0.0  \n",
       "184            0.6  \n",
       "185            8.9  \n",
       "\n",
       "[186 rows x 12 columns]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_day = pd.read_csv('data/rad202010tage.csv')\n",
    "df_day"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* Every file contains all counting_stations (zaehlstelle).\n",
    "* The counts are in three columns: direction_1 (richtung_1), direction_2 (richtung_2) and total (gesamt)\n",
    "* The day datasets also contain weather information - min/max temp, sun hours, rain and cloud"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['Arnulf', 'Kreuther', 'Olympia', 'Hirsch', 'Margareten', 'Erhardt'],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Available counting stations (zaehlstellen):\n",
    "df_day.zaehlstelle.unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# getDataframe of datafiles\n",
    "# - prefixes - filter used files by prefix e.g. rad2018 for year 2018\n",
    "# - zaehlstelle - filter by given counting station ()\n",
    "def getDataframe(prefixes, zaehlstelle: str):\n",
    "    files = [f for f in os.listdir('data') if os.path.isfile(os.path.join('data', f)) and f.endswith('tage.csv')]\n",
    "    files = [f for f in files if f.startswith(prefixes)]\n",
    "    df = pd.concat([pd.read_csv(os.path.join('data', f)) for f in files])\n",
    "    return df[df.zaehlstelle == zaehlstelle].sort_values(by=['datum'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_arnulf = getDataframe(zaehlstelle='Arnulf', prefixes = ('rad2017', 'rad2018', 'rad2019'))"
   ]
  },
  {
   "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>datum</th>\n",
       "      <th>uhrzeit_start</th>\n",
       "      <th>uhrzeit_ende</th>\n",
       "      <th>zaehlstelle</th>\n",
       "      <th>richtung_1</th>\n",
       "      <th>richtung_2</th>\n",
       "      <th>gesamt</th>\n",
       "      <th>min-temp</th>\n",
       "      <th>max-temp</th>\n",
       "      <th>niederschlag</th>\n",
       "      <th>bewoelkung</th>\n",
       "      <th>sonnenstunden</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2017.01.01</td>\n",
       "      <td>00:00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>121</td>\n",
       "      <td>14</td>\n",
       "      <td>135</td>\n",
       "      <td>-6.1</td>\n",
       "      <td>2.6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>36</td>\n",
       "      <td>5.9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2017.01.02</td>\n",
       "      <td>00:00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>362</td>\n",
       "      <td>38</td>\n",
       "      <td>400</td>\n",
       "      <td>-5.4</td>\n",
       "      <td>0.8</td>\n",
       "      <td>3.2</td>\n",
       "      <td>69</td>\n",
       "      <td>0.3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2017.01.03</td>\n",
       "      <td>00:00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>232</td>\n",
       "      <td>30</td>\n",
       "      <td>262</td>\n",
       "      <td>-2.1</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>85</td>\n",
       "      <td>3.6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2017.01.04</td>\n",
       "      <td>00:00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>182</td>\n",
       "      <td>25</td>\n",
       "      <td>207</td>\n",
       "      <td>-0.9</td>\n",
       "      <td>1.3</td>\n",
       "      <td>2.2</td>\n",
       "      <td>94</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2017.01.05</td>\n",
       "      <td>00:00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>193</td>\n",
       "      <td>31</td>\n",
       "      <td>224</td>\n",
       "      <td>-7.1</td>\n",
       "      <td>-0.2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>88</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>2019.12.27</td>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>312</td>\n",
       "      <td>30</td>\n",
       "      <td>342</td>\n",
       "      <td>1.7</td>\n",
       "      <td>4.9</td>\n",
       "      <td>4.9</td>\n",
       "      <td>98</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>2019.12.28</td>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>337</td>\n",
       "      <td>37</td>\n",
       "      <td>374</td>\n",
       "      <td>-2.1</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>69</td>\n",
       "      <td>3.2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>2019.12.29</td>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>253</td>\n",
       "      <td>32</td>\n",
       "      <td>285</td>\n",
       "      <td>-4.8</td>\n",
       "      <td>2.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>26</td>\n",
       "      <td>7.9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>2019.12.30</td>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>550</td>\n",
       "      <td>52</td>\n",
       "      <td>602</td>\n",
       "      <td>-4.4</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>9</td>\n",
       "      <td>8.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>2019.12.31</td>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>329</td>\n",
       "      <td>34</td>\n",
       "      <td>363</td>\n",
       "      <td>-0.5</td>\n",
       "      <td>8.6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>60</td>\n",
       "      <td>6.8</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1034 rows × 12 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         datum uhrzeit_start uhrzeit_ende zaehlstelle  richtung_1  richtung_2  \\\n",
       "0   2017.01.01      00:00:00        23.59      Arnulf         121          14   \n",
       "1   2017.01.02      00:00:00        23.59      Arnulf         362          38   \n",
       "2   2017.01.03      00:00:00        23.59      Arnulf         232          30   \n",
       "3   2017.01.04      00:00:00        23.59      Arnulf         182          25   \n",
       "4   2017.01.05      00:00:00        23.59      Arnulf         193          31   \n",
       "..         ...           ...          ...         ...         ...         ...   \n",
       "26  2019.12.27         00:00        23.59      Arnulf         312          30   \n",
       "27  2019.12.28         00:00        23.59      Arnulf         337          37   \n",
       "28  2019.12.29         00:00        23.59      Arnulf         253          32   \n",
       "29  2019.12.30         00:00        23.59      Arnulf         550          52   \n",
       "30  2019.12.31         00:00        23.59      Arnulf         329          34   \n",
       "\n",
       "    gesamt  min-temp  max-temp  niederschlag  bewoelkung  sonnenstunden  \n",
       "0      135      -6.1       2.6           0.0          36            5.9  \n",
       "1      400      -5.4       0.8           3.2          69            0.3  \n",
       "2      262      -2.1       0.6           0.0          85            3.6  \n",
       "3      207      -0.9       1.3           2.2          94            0.0  \n",
       "4      224      -7.1      -0.2           0.0          88            0.0  \n",
       "..     ...       ...       ...           ...         ...            ...  \n",
       "26     342       1.7       4.9           4.9          98            0.0  \n",
       "27     374      -2.1       2.0           0.0          69            3.2  \n",
       "28     285      -4.8       2.5           0.0          26            7.9  \n",
       "29     602      -4.4       6.0           0.0           9            8.0  \n",
       "30     363      -0.5       8.6           0.0          60            6.8  \n",
       "\n",
       "[1034 rows x 12 columns]"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_arnulf"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Are there missing days?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "from datetime import date, timedelta, datetime\n",
    "\n",
    "def findMissingDays(df):\n",
    "    dates = df.datum.map(lambda x: datetime.strptime(x, '%Y.%m.%d')).tolist()\n",
    "    dates.sort()\n",
    "    d = set(dates[0] + timedelta(x) for x in range((dates[-1] - dates[0]).days))\n",
    "    return sorted(d - set(dates))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(61, Timestamp('2019-06-01 00:00:00'), Timestamp('2019-07-31 00:00:00'))"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "missing = findMissingDays(df_arnulf)\n",
    "len(missing), missing[0], missing[-1]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There are missing days from June to July 2019"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Plot counts for a month"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# pip install matplotlib"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.rcParams['figure.figsize'] = [20, 15]\n",
    "\n",
    "df = getDataframe(prefixes=('rad2018'), zaehlstelle='Arnulf')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "<ipython-input-22-f5cfea7d9034>:2: MatplotlibDeprecationWarning: Adding an axes using the same arguments as a previous axes currently reuses the earlier instance.  In a future version, a new instance will always be created and returned.  Meanwhile, this warning can be suppressed, and the future behavior ensured, by passing a unique label to each axes instance.\n",
      "  ax1 = plt.axes()\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1080 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df.plot.bar(x='datum', y='gesamt', rot=45)\n",
    "ax1 = plt.axes()\n",
    "x_axis = ax1.axes.get_xaxis()\n",
    "for index, label in enumerate(x_axis.get_ticklabels()):\n",
    "    if not label.get_text().endswith('01'):\n",
    "        label.set_visible(False)\n",
    "plt.show()\n",
    "plt.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. Modelling <a name=\"model\"></a>\n",
    "\n",
    "I will work with the daily numbers because they already contain the weather information.\n",
    "I will use the year 2017 & 2018 as training data and 2019 for validation."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.1. Random Forest <a name=\"rf\"></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Prepare data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "# conda install -c fastai -c pytorch fastai"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.8/site-packages/torch/cuda/__init__.py:52: UserWarning: CUDA initialization: Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx (Triggered internally at  /opt/conda/conda-bld/pytorch_1603729096996/work/c10/cuda/CUDAFunctions.cpp:100.)\n",
      "  return torch._C._cuda_getDeviceCount() > 0\n"
     ]
    }
   ],
   "source": [
    "from fastai.tabular.all import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "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>uhrzeit_start</th>\n",
       "      <th>uhrzeit_ende</th>\n",
       "      <th>datumWeek</th>\n",
       "      <th>zaehlstelle</th>\n",
       "      <th>richtung_1</th>\n",
       "      <th>richtung_2</th>\n",
       "      <th>gesamt</th>\n",
       "      <th>min-temp</th>\n",
       "      <th>max-temp</th>\n",
       "      <th>niederschlag</th>\n",
       "      <th>...</th>\n",
       "      <th>datumDay</th>\n",
       "      <th>datumDayofweek</th>\n",
       "      <th>datumDayofyear</th>\n",
       "      <th>datumIs_month_end</th>\n",
       "      <th>datumIs_month_start</th>\n",
       "      <th>datumIs_quarter_end</th>\n",
       "      <th>datumIs_quarter_start</th>\n",
       "      <th>datumIs_year_end</th>\n",
       "      <th>datumIs_year_start</th>\n",
       "      <th>datumElapsed</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>00:00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>52</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>121</td>\n",
       "      <td>14</td>\n",
       "      <td>135</td>\n",
       "      <td>-6.1</td>\n",
       "      <td>2.6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>1483228800</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>00:00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>1</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>362</td>\n",
       "      <td>38</td>\n",
       "      <td>400</td>\n",
       "      <td>-5.4</td>\n",
       "      <td>0.8</td>\n",
       "      <td>3.2</td>\n",
       "      <td>...</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>1483315200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>00:00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>1</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>232</td>\n",
       "      <td>30</td>\n",
       "      <td>262</td>\n",
       "      <td>-2.1</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>1483401600</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>00:00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>1</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>182</td>\n",
       "      <td>25</td>\n",
       "      <td>207</td>\n",
       "      <td>-0.9</td>\n",
       "      <td>1.3</td>\n",
       "      <td>2.2</td>\n",
       "      <td>...</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>1483488000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>00:00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>1</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>193</td>\n",
       "      <td>31</td>\n",
       "      <td>224</td>\n",
       "      <td>-7.1</td>\n",
       "      <td>-0.2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>1483574400</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 24 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "  uhrzeit_start uhrzeit_ende  datumWeek zaehlstelle  richtung_1  richtung_2  \\\n",
       "0      00:00:00        23.59         52      Arnulf         121          14   \n",
       "1      00:00:00        23.59          1      Arnulf         362          38   \n",
       "2      00:00:00        23.59          1      Arnulf         232          30   \n",
       "3      00:00:00        23.59          1      Arnulf         182          25   \n",
       "4      00:00:00        23.59          1      Arnulf         193          31   \n",
       "\n",
       "   gesamt  min-temp  max-temp  niederschlag  ...  datumDay  datumDayofweek  \\\n",
       "0     135      -6.1       2.6           0.0  ...         1               6   \n",
       "1     400      -5.4       0.8           3.2  ...         2               0   \n",
       "2     262      -2.1       0.6           0.0  ...         3               1   \n",
       "3     207      -0.9       1.3           2.2  ...         4               2   \n",
       "4     224      -7.1      -0.2           0.0  ...         5               3   \n",
       "\n",
       "   datumDayofyear  datumIs_month_end  datumIs_month_start  \\\n",
       "0               1              False                 True   \n",
       "1               2              False                False   \n",
       "2               3              False                False   \n",
       "3               4              False                False   \n",
       "4               5              False                False   \n",
       "\n",
       "   datumIs_quarter_end  datumIs_quarter_start  datumIs_year_end  \\\n",
       "0                False                   True             False   \n",
       "1                False                  False             False   \n",
       "2                False                  False             False   \n",
       "3                False                  False             False   \n",
       "4                False                  False             False   \n",
       "\n",
       "   datumIs_year_start  datumElapsed  \n",
       "0                True    1483228800  \n",
       "1               False    1483315200  \n",
       "2               False    1483401600  \n",
       "3               False    1483488000  \n",
       "4               False    1483574400  \n",
       "\n",
       "[5 rows x 24 columns]"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = getDataframe(prefixes=('rad2017', 'rad2018', 'rad2019'), zaehlstelle='Arnulf')\n",
    "# add date columns like day of week etc. \n",
    "df = add_datepart(df, 'datum')\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# drop richtung columns because they sum up to gesamt value, which we want to predict\n",
    "df = df.drop(['richtung_1', 'richtung_2'], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create Tabular Pandas\n",
    "procs = [Categorify, FillMissing] # processings or transformations\n",
    "cond = (df.datumYear<2019)\n",
    "train_idx = np.where(cond)[0]\n",
    "valid_idx = np.where(~cond)[0]\n",
    "\n",
    "splits = (list(train_idx),list(valid_idx))\n",
    "\n",
    "dep_var = 'gesamt' # dependent variable, the number we want to predict\n",
    "\n",
    "cont, cat = cont_cat_split(df, 1, dep_var=dep_var)\n",
    "to = TabularPandas(df, procs, cat, cont, y_names=dep_var, splits=splits)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>uhrzeit_start</th>\n",
       "      <th>uhrzeit_ende</th>\n",
       "      <th>datumWeek</th>\n",
       "      <th>zaehlstelle</th>\n",
       "      <th>datumIs_month_end</th>\n",
       "      <th>datumIs_month_start</th>\n",
       "      <th>datumIs_quarter_end</th>\n",
       "      <th>datumIs_quarter_start</th>\n",
       "      <th>datumIs_year_end</th>\n",
       "      <th>datumIs_year_start</th>\n",
       "      <th>datumElapsed</th>\n",
       "      <th>min-temp</th>\n",
       "      <th>max-temp</th>\n",
       "      <th>niederschlag</th>\n",
       "      <th>bewoelkung</th>\n",
       "      <th>sonnenstunden</th>\n",
       "      <th>datumYear</th>\n",
       "      <th>datumMonth</th>\n",
       "      <th>datumDay</th>\n",
       "      <th>datumDayofweek</th>\n",
       "      <th>datumDayofyear</th>\n",
       "      <th>gesamt</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>00:00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>52</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>1483228800</td>\n",
       "      <td>-6.1</td>\n",
       "      <td>2.6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>36</td>\n",
       "      <td>5.9</td>\n",
       "      <td>2017</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>1</td>\n",
       "      <td>135</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>00:00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>1</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>1483315200</td>\n",
       "      <td>-5.4</td>\n",
       "      <td>0.8</td>\n",
       "      <td>3.2</td>\n",
       "      <td>69</td>\n",
       "      <td>0.3</td>\n",
       "      <td>2017</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>400</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>00:00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>1</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>1483401600</td>\n",
       "      <td>-2.1</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>85</td>\n",
       "      <td>3.6</td>\n",
       "      <td>2017</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>262</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "to.show(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# save\n",
    "save_pickle((URLs.path()/'../to.pkl'), to)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Train Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestRegressor\n",
    "from sklearn.metrics import mean_absolute_error"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "xs,y = to.train.xs, to.train.y\n",
    "valid_xs, valid_y = to.valid.xs, to.valid.y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "def rf(xs, y, n_estimators=40, max_features=0.5, min_samples_leaf=5, **kwargs):\n",
    "    return RandomForestRegressor(n_jobs=-1, n_estimators=n_estimators, min_samples_leaf=min_samples_leaf, max_features=max_features, oob_score=True).fit(xs,y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "# root mean square error - penalizes large differences. Would MAE be better as metric?\n",
    "def r_mse(pred,y): return round(math.sqrt(((pred-y)**2).mean()), 6)\n",
    "def m_rmse(m, xs, y): return r_mse(m.predict(xs), y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "m = rf(xs, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(153.477832, 293.671934)"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "m_rmse(m, xs, y), m_rmse(m, valid_xs, valid_y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(109.65702609142153, 220.2170140588274)"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mean_absolute_error(m.predict(xs), y),mean_absolute_error(m.predict(valid_xs), valid_y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "293.671934"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preds = np.stack([t.predict(valid_xs) for t in m.estimators_])\n",
    "r_mse(preds.mean(0), valid_y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1080 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot([r_mse(preds[:i+1].mean(0), valid_y) for i in range(40)]);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "😳 Question: Best result in a previous run seem to be arround 20 not 40 what does that mean?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Out of bag error\n",
    "Good metric to check why our predictions in our training set or better then in our validation set. 108.36805410446944 vs 212.6982400192359"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "223.275392"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "r_mse(m.oob_prediction_, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "😳 Our error in the training set is higher than in the validation set. The example in the fastai book p.302 has an example where the error is lower.\n",
    "\n",
    "What does this indicate?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Prediction confidence\n",
    "\n",
    "how confident can I be in those predictions? One way is to look at the variance between the trees in the forest. Smaller is better."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(40, 304)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([274.7184305 , 233.65713255, 241.40697189, 203.38162658,\n",
       "        57.55733204,  55.90400176, 269.04874212, 189.78659779,\n",
       "       175.57909927, 199.08264523, 205.3682141 ,  86.69497514,\n",
       "        81.36701804, 190.08011092, 249.5200124 , 276.81426458,\n",
       "       210.1592155 , 245.63738955, 121.95687933, 120.68516076])"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preds = np.stack([t.predict(valid_xs) for t in m.estimators_])\n",
    "print(preds.shape) # 40 trees, with 304 predictions for each day\n",
    "preds_std = preds.std(0) # standard deviation\n",
    "preds_std[:20]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Feature importance\n",
    "\n",
    "Which feature are most important?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "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>cols</th>\n",
       "      <th>imp</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>max-temp</td>\n",
       "      <td>0.378081</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>datumDayofweek</td>\n",
       "      <td>0.198147</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>sonnenstunden</td>\n",
       "      <td>0.139527</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>min-temp</td>\n",
       "      <td>0.137272</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>datumDayofyear</td>\n",
       "      <td>0.046788</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>niederschlag</td>\n",
       "      <td>0.027064</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>datumWeek</td>\n",
       "      <td>0.022901</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>bewoelkung</td>\n",
       "      <td>0.020216</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>datumElapsed</td>\n",
       "      <td>0.016275</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>datumDay</td>\n",
       "      <td>0.007746</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              cols       imp\n",
       "12        max-temp  0.378081\n",
       "19  datumDayofweek  0.198147\n",
       "15   sonnenstunden  0.139527\n",
       "11        min-temp  0.137272\n",
       "20  datumDayofyear  0.046788\n",
       "13    niederschlag  0.027064\n",
       "2        datumWeek  0.022901\n",
       "14      bewoelkung  0.020216\n",
       "10    datumElapsed  0.016275\n",
       "18        datumDay  0.007746"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def rf_feat_importance(m, df):\n",
    "    return pd.DataFrame({'cols':df.columns, 'imp':m.feature_importances_}).sort_values('imp', ascending=False)\n",
    "fi = rf_feat_importance(m, xs)\n",
    "fi[:10]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Most important features are max-temp followed by day of week. Seems reasonable to me. e.g. workdays vs weekend and cold days vs warm days."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def plot_fi(fi):\n",
    "    return fi.plot('cols', 'imp', 'barh', figsize=(12,7), legend=False)\n",
    "plot_fi(fi[:20]);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Remove unimportant features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "10"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "to_keep =fi[fi.imp > 0.005].cols\n",
    "len(to_keep)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "xs_imp = xs[to_keep]\n",
    "valid_xs_imp = valid_xs[to_keep]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(155.023982, 292.665735)"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "m = rf(xs_imp, y)\n",
    "m_rmse(m, xs_imp, y), m_rmse(m, valid_xs_imp, valid_y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "😳 Got worse in comparison to prev run: 151.42499, 285.167373. But the removed columns shouldn't be important e.g. uhrzeit_start/end is always the same."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Redundant features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from scipy.cluster import hierarchy as hc\n",
    "\n",
    "def cluster_columns(df, figsize=(10,6), font_size=12):\n",
    "    corr = np.round(scipy.stats.spearmanr(df).correlation, 4)\n",
    "    corr_condensed = hc.distance.squareform(1-corr)\n",
    "    z = hc.linkage(corr_condensed, method='average')\n",
    "    fig = plt.figure(figsize=figsize)\n",
    "    hc.dendrogram(z, labels=df.columns, orientation='left', leaf_font_size=font_size)\n",
    "    plt.show()\n",
    "    \n",
    "cluster_columns(xs_imp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "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>max-temp</th>\n",
       "      <th>datumDayofweek</th>\n",
       "      <th>sonnenstunden</th>\n",
       "      <th>min-temp</th>\n",
       "      <th>datumDayofyear</th>\n",
       "      <th>niederschlag</th>\n",
       "      <th>datumWeek</th>\n",
       "      <th>bewoelkung</th>\n",
       "      <th>datumElapsed</th>\n",
       "      <th>datumDay</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2.6</td>\n",
       "      <td>6</td>\n",
       "      <td>5.9</td>\n",
       "      <td>-6.1</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>52</td>\n",
       "      <td>36</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.8</td>\n",
       "      <td>0</td>\n",
       "      <td>0.3</td>\n",
       "      <td>-5.4</td>\n",
       "      <td>2</td>\n",
       "      <td>3.2</td>\n",
       "      <td>1</td>\n",
       "      <td>69</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.6</td>\n",
       "      <td>1</td>\n",
       "      <td>3.6</td>\n",
       "      <td>-2.1</td>\n",
       "      <td>3</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>85</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1.3</td>\n",
       "      <td>2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-0.9</td>\n",
       "      <td>4</td>\n",
       "      <td>2.2</td>\n",
       "      <td>1</td>\n",
       "      <td>94</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-0.2</td>\n",
       "      <td>3</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-7.1</td>\n",
       "      <td>5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>88</td>\n",
       "      <td>5</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   max-temp  datumDayofweek  sonnenstunden  min-temp  datumDayofyear  \\\n",
       "0       2.6               6            5.9      -6.1               1   \n",
       "1       0.8               0            0.3      -5.4               2   \n",
       "2       0.6               1            3.6      -2.1               3   \n",
       "3       1.3               2            0.0      -0.9               4   \n",
       "4      -0.2               3            0.0      -7.1               5   \n",
       "\n",
       "   niederschlag  datumWeek  bewoelkung  datumElapsed  datumDay  \n",
       "0           0.0         52          36             1         1  \n",
       "1           3.2          1          69             2         2  \n",
       "2           0.0          1          85             3         3  \n",
       "3           2.2          1          94             4         4  \n",
       "4           0.0          1          88             5         5  "
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "xs_imp.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "month and day of year are close, makes sense both indicate the \"season\" or the progress into the year.\n",
    "Will keep it for now."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Waterfall Chart"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "#!pip install treeinterpreter\n",
    "#!pip install waterfallcharts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.8/site-packages/sklearn/utils/deprecation.py:143: FutureWarning: The sklearn.ensemble.forest module is  deprecated in version 0.22 and will be removed in version 0.24. The corresponding classes / functions should instead be imported from sklearn.ensemble. Anything that cannot be imported from sklearn.ensemble is now part of the private API.\n",
      "  warnings.warn(message, FutureWarning)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(array([669.61353289]), 1176.2661986301368, -506.65266574349744)"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from treeinterpreter import treeinterpreter\n",
    "row = valid_xs_imp.iloc[:1]\n",
    "prediction,bias,contributions = treeinterpreter.predict(m, row.values)\n",
    "prediction[0], bias[0], contributions[0].sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<module 'matplotlib.pyplot' from '/opt/conda/lib/python3.8/site-packages/matplotlib/pyplot.py'>"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1080 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import waterfall_chart\n",
    "waterfall_chart.plot(valid_xs_imp.columns, contributions[0], threshold=0.08, rotation_value=45)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Indicates the importance of different features for the prediction."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Out of domain data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_dom = pd.concat([xs_imp, valid_xs_imp])\n",
    "is_valid = np.array([0]*len(xs_imp) + [1]*len(valid_xs_imp))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "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>cols</th>\n",
       "      <th>imp</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>datumElapsed</td>\n",
       "      <td>0.958879</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>min-temp</td>\n",
       "      <td>0.012749</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>max-temp</td>\n",
       "      <td>0.012634</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>bewoelkung</td>\n",
       "      <td>0.003755</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>datumWeek</td>\n",
       "      <td>0.002873</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>datumDayofyear</td>\n",
       "      <td>0.002811</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             cols       imp\n",
       "8    datumElapsed  0.958879\n",
       "3        min-temp  0.012749\n",
       "0        max-temp  0.012634\n",
       "7      bewoelkung  0.003755\n",
       "6       datumWeek  0.002873\n",
       "4  datumDayofyear  0.002811"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# RandomForest to decide if value is in train oder valid set.\n",
    "m = rf(df_dom, is_valid)\n",
    "rf_feat_importance(m, df_dom)[:6]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "By extracting the most important features which seperate train from validation set we find feature which we could drop."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "without out of dom: 284.683368\n"
     ]
    }
   ],
   "source": [
    "xs_final = xs_imp.drop('datumElapsed', axis=1)\n",
    "valid_xs_final = valid_xs_imp.drop('datumElapsed', axis=1)\n",
    "m = rf(xs_final, y)\n",
    "print(f'without out of dom: {m_rmse(m, valid_xs_final, valid_y)}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "😳 Got worse again."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Summary of RandomForest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "orig: 287.91052\n",
      "imp: 290.178617\n",
      "without out of dom: 286.275386\n"
     ]
    }
   ],
   "source": [
    "m = rf(xs, y)\n",
    "print(f'orig: {m_rmse(m, valid_xs, valid_y)}')\n",
    "m = rf(xs_imp, y)\n",
    "print(f'imp: {m_rmse(m, valid_xs_imp, valid_y)}')\n",
    "m = rf(xs_imp.drop('datumElapsed',axis=1), y)\n",
    "print(f'without out of dom: {m_rmse(m, valid_xs_imp.drop(\"datumElapsed\",axis=1), valid_y)}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### less trees"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "orig: 289.908937\n",
      "imp: 283.236989\n",
      "without out of dom: 293.985873\n"
     ]
    }
   ],
   "source": [
    "m = rf(xs, y, 25)\n",
    "print(f'orig: {m_rmse(m, valid_xs, valid_y)}')\n",
    "m = rf(xs_imp, y, 25)\n",
    "print(f'imp: {m_rmse(m, valid_xs_imp, valid_y)}')\n",
    "m = rf(xs_imp.drop('datumElapsed',axis=1), y, 25)\n",
    "print(f'without out of dom: {m_rmse(m, valid_xs_imp.drop(\"datumElapsed\",axis=1), valid_y)}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Open questions"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "😳 Results change on every rerun. Therfore: How comparable are the results with the different changes then? \n",
    "\n",
    "Should I fix a seed for the RandomForest, is this possible?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.1. Neural Net <a name=\"nn\"></a>\n",
    "\n",
    "Let's try it with a neural net"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "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>max-temp</th>\n",
       "      <th>datumDayofweek</th>\n",
       "      <th>sonnenstunden</th>\n",
       "      <th>min-temp</th>\n",
       "      <th>datumDayofyear</th>\n",
       "      <th>niederschlag</th>\n",
       "      <th>datumWeek</th>\n",
       "      <th>bewoelkung</th>\n",
       "      <th>datumDay</th>\n",
       "      <th>gesamt</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2.6</td>\n",
       "      <td>6</td>\n",
       "      <td>5.9</td>\n",
       "      <td>-6.1</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>52</td>\n",
       "      <td>36</td>\n",
       "      <td>1</td>\n",
       "      <td>135</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.8</td>\n",
       "      <td>0</td>\n",
       "      <td>0.3</td>\n",
       "      <td>-5.4</td>\n",
       "      <td>2</td>\n",
       "      <td>3.2</td>\n",
       "      <td>1</td>\n",
       "      <td>69</td>\n",
       "      <td>2</td>\n",
       "      <td>400</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.6</td>\n",
       "      <td>1</td>\n",
       "      <td>3.6</td>\n",
       "      <td>-2.1</td>\n",
       "      <td>3</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>85</td>\n",
       "      <td>3</td>\n",
       "      <td>262</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1.3</td>\n",
       "      <td>2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-0.9</td>\n",
       "      <td>4</td>\n",
       "      <td>2.2</td>\n",
       "      <td>1</td>\n",
       "      <td>94</td>\n",
       "      <td>4</td>\n",
       "      <td>207</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-0.2</td>\n",
       "      <td>3</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-7.1</td>\n",
       "      <td>5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>88</td>\n",
       "      <td>5</td>\n",
       "      <td>224</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>4.9</td>\n",
       "      <td>4</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.7</td>\n",
       "      <td>361</td>\n",
       "      <td>4.9</td>\n",
       "      <td>52</td>\n",
       "      <td>98</td>\n",
       "      <td>27</td>\n",
       "      <td>342</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>2.0</td>\n",
       "      <td>5</td>\n",
       "      <td>3.2</td>\n",
       "      <td>-2.1</td>\n",
       "      <td>362</td>\n",
       "      <td>0.0</td>\n",
       "      <td>52</td>\n",
       "      <td>69</td>\n",
       "      <td>28</td>\n",
       "      <td>374</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>2.5</td>\n",
       "      <td>6</td>\n",
       "      <td>7.9</td>\n",
       "      <td>-4.8</td>\n",
       "      <td>363</td>\n",
       "      <td>0.0</td>\n",
       "      <td>52</td>\n",
       "      <td>26</td>\n",
       "      <td>29</td>\n",
       "      <td>285</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>6.0</td>\n",
       "      <td>0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>-4.4</td>\n",
       "      <td>364</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>9</td>\n",
       "      <td>30</td>\n",
       "      <td>602</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>8.6</td>\n",
       "      <td>1</td>\n",
       "      <td>6.8</td>\n",
       "      <td>-0.5</td>\n",
       "      <td>365</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>60</td>\n",
       "      <td>31</td>\n",
       "      <td>363</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1034 rows × 10 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    max-temp  datumDayofweek  sonnenstunden  min-temp  datumDayofyear  \\\n",
       "0        2.6               6            5.9      -6.1               1   \n",
       "1        0.8               0            0.3      -5.4               2   \n",
       "2        0.6               1            3.6      -2.1               3   \n",
       "3        1.3               2            0.0      -0.9               4   \n",
       "4       -0.2               3            0.0      -7.1               5   \n",
       "..       ...             ...            ...       ...             ...   \n",
       "26       4.9               4            0.0       1.7             361   \n",
       "27       2.0               5            3.2      -2.1             362   \n",
       "28       2.5               6            7.9      -4.8             363   \n",
       "29       6.0               0            8.0      -4.4             364   \n",
       "30       8.6               1            6.8      -0.5             365   \n",
       "\n",
       "    niederschlag  datumWeek  bewoelkung  datumDay  gesamt  \n",
       "0            0.0         52          36         1     135  \n",
       "1            3.2          1          69         2     400  \n",
       "2            0.0          1          85         3     262  \n",
       "3            2.2          1          94         4     207  \n",
       "4            0.0          1          88         5     224  \n",
       "..           ...        ...         ...       ...     ...  \n",
       "26           4.9         52          98        27     342  \n",
       "27           0.0         52          69        28     374  \n",
       "28           0.0         52          26        29     285  \n",
       "29           0.0          1           9        30     602  \n",
       "30           0.0          1          60        31     363  \n",
       "\n",
       "[1034 rows x 10 columns]"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_nn = getDataframe(zaehlstelle='Arnulf', prefixes = ('rad2017', 'rad2018', 'rad2019'))\n",
    "df_nn = add_datepart(df_nn, 'datum')\n",
    "df_nn = df_nn[list(xs_final.columns) + [dep_var]]\n",
    "df_nn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "cont_nn, cat_nn = cont_cat_split(df_nn, max_card=9000, dep_var=dep_var)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(['max-temp', 'sonnenstunden', 'min-temp', 'niederschlag'],\n",
       " ['datumDayofweek', 'datumDayofyear', 'datumWeek', 'bewoelkung', 'datumDay'])"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cont_nn, cat_nn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "datumDayofweek      7\n",
       "datumDayofyear    365\n",
       "datumWeek          52\n",
       "bewoelkung         81\n",
       "datumDay           31\n",
       "dtype: int64"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_nn[cat_nn].nunique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_nn=df_nn.astype('float')\n",
    "\n",
    "procs_nn = [Categorify, FillMissing, Normalize]\n",
    "to_nn = TabularPandas(df_nn, procs_nn, cat_nn, cont_nn, splits=splits, y_names=dep_var)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "dls = to_nn.dataloaders()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn = tabular_learner(dls, y_range=(29,2707), layers=[500,250], n_out=1, loss_func=F.mse_loss)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "SuggestedLRs(lr_min=0.0033113110810518267, lr_steep=7.585775847473997e-07)"
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1080 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "learn.lr_find()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>234669.062500</td>\n",
       "      <td>246372.031250</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>169412.312500</td>\n",
       "      <td>176401.015625</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>142154.109375</td>\n",
       "      <td>183510.671875</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>122397.789062</td>\n",
       "      <td>130206.195312</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>101110.593750</td>\n",
       "      <td>130590.023438</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>89784.742188</td>\n",
       "      <td>119201.250000</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>78525.718750</td>\n",
       "      <td>103552.343750</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>69051.562500</td>\n",
       "      <td>111986.250000</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>63434.929688</td>\n",
       "      <td>113892.578125</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>55306.207031</td>\n",
       "      <td>94354.257812</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>47302.691406</td>\n",
       "      <td>92095.726562</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>40780.195312</td>\n",
       "      <td>90601.148438</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>34885.820312</td>\n",
       "      <td>84908.648438</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>29407.443359</td>\n",
       "      <td>86845.812500</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>24780.492188</td>\n",
       "      <td>90764.921875</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>21015.712891</td>\n",
       "      <td>87902.289062</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>18285.087891</td>\n",
       "      <td>84684.562500</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>15470.899414</td>\n",
       "      <td>84034.000000</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>13186.029297</td>\n",
       "      <td>84104.734375</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>11543.091797</td>\n",
       "      <td>84225.312500</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "learn.fit_one_cycle(20, 1e-2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "290.215976"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preds,targs = learn.get_preds()\n",
    "r_mse(preds,targs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Is better than the random forest"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Analyizing results <a name=\"analyze\"></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.1 Other metrics for evaluation?\n",
    "\n",
    "Above I used the root mean square error to get a metric for the models. It is helpful to compare both models with each other. Are there other metrics that would help me to get a better understanding of the performance of the models?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# results\n",
    "rf_m = rf(xs_final, y)\n",
    "\n",
    "rf_preds = rf_m.predict(valid_xs_final)\n",
    "nn_preds, nn_y = learn.get_preds()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "from fastai import metrics"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Root Mean Squared Error (RMSE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "290.346405\n",
      "290.215976\n"
     ]
    }
   ],
   "source": [
    "# Root Mean Squared Error (RMSE)\n",
    "print(r_mse(rf_preds, valid_y))\n",
    "print(r_mse(nn_preds, nn_y))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Penalizes bigger errors more than MAE."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Mean absolute error (MAE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "221.92851825213108\n",
      "216.30911\n"
     ]
    }
   ],
   "source": [
    "# Mean absolute error (MAE)\n",
    "def mae(preds, y): return np.abs(preds - y).mean()\n",
    "\n",
    "print(mae(rf_preds, valid_y))\n",
    "print(mae(nn_preds, nn_y).numpy())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Allows a more intuitive understanding, as the error says, it is the mean absolute error. But could we set it in relation to range of values to get a better understanding?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Mean absolute percentage error (MAPE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.34469792750873274\n",
      "0.31876716\n"
     ]
    }
   ],
   "source": [
    "# Mean absolute percentage error (MAPE)\n",
    "def mape(preds, y): return np.abs((y - preds) / y).mean()\n",
    "\n",
    "print(mape(rf_preds, valid_y))\n",
    "print(mape(nn_preds, nn_y).numpy())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ok so we are arround 28% off. Seems pretty high to me but maybe it is still useful for traffic planer?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### normalized Mean Absolute Error / nMAE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.214701915905157\n",
      "0.20063037\n"
     ]
    }
   ],
   "source": [
    "def nmae(preds, y): return np.abs((preds - y)).mean() / preds.mean()\n",
    "\n",
    "print(nmae(rf_preds, valid_y))\n",
    "print(nmae(nn_preds, nn_y).numpy())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Another relative metric. Loses some of the details (outliers) etc. because the averaging is done after summing up all errors."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.2 Visualize results\n",
    "\n",
    "Could I get more insights by visualizing the data and results?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Plot the results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1080 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot a bar chart\n",
    "plt.bar(valid_xs['datumDayofyear'], rf_preds, label='RandomForest')\n",
    "plt.bar(valid_xs['datumDayofyear'], nn_preds, label='Neuronal Net')\n",
    "plt.bar(valid_xs['datumDayofyear'], valid_y, label='Real data')\n",
    "plt.xlabel('day')\n",
    "plt.ylabel('total')\n",
    "plt.title(\"Arnulf\")\n",
    "plt.legend();"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Looks quite chaotic because it is too much data and small values are hidden behind the other bars. I played arround with a few different visualization but I think the cleanest one is following: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = getDataframe(prefixes=('rad2019'), zaehlstelle='Arnulf')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "def addMissingDays(df):\n",
    "    df['date'] = pd.to_datetime(df['datum'], format=\"%Y.%m.%d\")\n",
    "    idx = pd.date_range(df['date'].iloc[0], df['date'].iloc[-1])\n",
    "    df.index = pd.DatetimeIndex(df['date'])\n",
    "    df = df.reindex(idx)\n",
    "    return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = addMissingDays(df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [],
   "source": [
    "def getPredsForRF(df):\n",
    "    df = add_datepart(df, 'datum')\n",
    "    results = pd.DataFrame(columns=['date', 'pred'])\n",
    "    for index, x in df.iterrows():\n",
    "        pred = -1\n",
    "        if not pd.isnull(x['gesamt']):\n",
    "            pred = rf_m.predict([x[list(xs_final.columns)]])[0]     \n",
    "        dfTemp = pd.DataFrame(data={'date': [x['date']], 'pred': [pred]})\n",
    "        results = results.append(dfTemp)\n",
    "    return results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [],
   "source": [
    "rf_2019 = getPredsForRF(df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plotResults(df, preds, s=0, n=365, title='Arnulf', label_df='2019', label_pred='RandomForest'):\n",
    "    plt.bar(preds['date'][s:n], preds['pred'][s:n], label=label_pred, color='blue', width=0.8)\n",
    "    plt.bar(df['date'][s:n], df['gesamt'][s:n], label=label_df, color='red', width=0.4)\n",
    "    plt.xlabel('day')\n",
    "    plt.ylabel('total')\n",
    "    plt.title(title)\n",
    "    plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1080 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plotResults(df, rf_2019, n=50)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "By stepping through different timeframes I can see big errors. But to be true I didn't identify anything useful. Maybe there is a better visualisation out there?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.3 When do the biggest errors happen?\n",
    "\n",
    "I would like to know what the reasons are for the errors. So I think it is a good idea to look at the biggest errors."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [],
   "source": [
    "def getPredsForRF(df):\n",
    "    df = add_datepart(df, 'datum')\n",
    "    results = pd.DataFrame(columns=['date', 'pred'])\n",
    "    for index, x in df.iterrows():\n",
    "        pred = -1\n",
    "        d = x['date']\n",
    "        if not pd.isnull(x['gesamt']):\n",
    "            pred = rf_m.predict([x[list(xs_final.columns)]])[0]\n",
    "        dfTemp = pd.DataFrame(data={'date': index, 'pred': [pred]})\n",
    "        results = results.append(dfTemp)\n",
    "    return results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [],
   "source": [
    "df2019 = getDataframe(prefixes=('rad2019'), zaehlstelle='Arnulf')\n",
    "df2019 = addMissingDays(df2019)\n",
    "preds2019RF = getPredsForRF(df2019)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [],
   "source": [
    "errors = df2019['gesamt'].replace(np.NaN, -1).subtract(preds2019RF.set_index(preds2019RF.date)['pred'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2019-04-19   -1142.909949\n",
       "2019-05-30   -1099.506745\n",
       "2019-04-22   -1024.018307\n",
       "2019-05-31    -915.873021\n",
       "2019-05-01    -796.376005\n",
       "2019-08-15    -732.695634\n",
       "2019-12-05     713.842115\n",
       "2019-05-24    -692.161935\n",
       "2019-10-22     662.970737\n",
       "2019-05-27    -662.277323\n",
       "2019-04-21    -599.506519\n",
       "2019-09-20     597.690886\n",
       "2019-05-23    -589.133709\n",
       "2019-12-11     588.588973\n",
       "2019-05-26    -583.634995\n",
       "2019-05-25    -566.750939\n",
       "2019-10-03    -555.810868\n",
       "2019-11-01    -550.701965\n",
       "2019-09-06     541.161308\n",
       "2019-11-07     536.506263\n",
       "dtype: float64"
      ]
     },
     "execution_count": 83,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# top errors for random forest\n",
    "errors.sort_values(ascending=False, key=lambda x: np.abs(x))[:20]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* 8 of 20 biggest erros happen on a national holiday (feiertag)\n",
    "* 1 is durent a school holidays\n",
    "* 1 is a \"Brückentag\"\n",
    "\n",
    "So this looks like a promising way to improve my dataset by adding national holidays, school holidays to the dataset. In the future it might also be helpful to add \"Brückentage\" or days before a long weekend.\n",
    "\n",
    "\n",
    "```\n",
    "2019-04-19   -1180.442298 Karfreitag\n",
    "2019-05-30   -1179.176310 Christi Himmelfahrt\n",
    "2019-04-22   -1059.768092 Ostermontag\n",
    "2019-05-31    -954.947565 Brückentag Christi Himmelfahrt\n",
    "2019-08-15    -772.263989 Maria Himmelfahrt\n",
    "2019-05-24    -744.230733 ?\n",
    "2019-10-22     696.795180 ? \n",
    "2019-05-01    -695.081500 Schulferien Weihnachten\n",
    "2019-12-05     674.750098 ?\n",
    "2019-09-20     633.990541 ?\n",
    "2019-05-23    -630.930030 ?\n",
    "2019-05-26    -607.316037 ?\n",
    "2019-10-03    -597.941322 Tag der deutschen Einheit\n",
    "2019-11-01    -585.093491 Allerheiligen\n",
    "2019-04-21    -565.883708 ?\n",
    "2019-05-27    -565.291106 ?\n",
    "2019-05-25    -544.784515 ?\n",
    "2019-05-29    -538.861066 Mittwoch vor Feiertag\n",
    "2019-09-27     537.904750 ?\n",
    "2019-09-06     524.922845 Schulferien Sommer\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5. Improvements <a name=\"improve\"></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.1 Add more features to dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### National holidays / Feiertage"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For national holidays there exists an API https://feiertage-api.de/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [],
   "source": [
    "def getFeiertage(jahre = [2017,2018,2019], bundesland = 'BY', ignore = ['Augsburger Friedensfest']):\n",
    "    feiertage_df = pd.DataFrame(columns={'date','name'})\n",
    "    for jahr in jahre:\n",
    "        r = requests.get(f'https://feiertage-api.de/api/?jahr={jahr}&nur_land={bundesland}')\n",
    "        json = r.json()\n",
    "        for key, d in json.items():\n",
    "            if not key in ignore:\n",
    "                feiertage_df = feiertage_df.append({'date': datetime.strptime(d['datum'], \"%Y-%m-%d\"), 'name':key}, ignore_index=True)\n",
    "    return feiertage_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>date</th>\n",
       "      <th>name</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2017-01-01</td>\n",
       "      <td>Neujahrstag</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2017-01-06</td>\n",
       "      <td>Heilige Drei Könige</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2017-04-14</td>\n",
       "      <td>Karfreitag</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2017-04-17</td>\n",
       "      <td>Ostermontag</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2017-05-01</td>\n",
       "      <td>Tag der Arbeit</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>2017-05-25</td>\n",
       "      <td>Christi Himmelfahrt</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>2017-06-05</td>\n",
       "      <td>Pfingstmontag</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>2017-06-15</td>\n",
       "      <td>Fronleichnam</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>2017-08-15</td>\n",
       "      <td>Mariä Himmelfahrt</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>2017-10-03</td>\n",
       "      <td>Tag der Deutschen Einheit</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        date                       name\n",
       "0 2017-01-01                Neujahrstag\n",
       "1 2017-01-06        Heilige Drei Könige\n",
       "2 2017-04-14                 Karfreitag\n",
       "3 2017-04-17                Ostermontag\n",
       "4 2017-05-01             Tag der Arbeit\n",
       "5 2017-05-25        Christi Himmelfahrt\n",
       "6 2017-06-05              Pfingstmontag\n",
       "7 2017-06-15               Fronleichnam\n",
       "8 2017-08-15          Mariä Himmelfahrt\n",
       "9 2017-10-03  Tag der Deutschen Einheit"
      ]
     },
     "execution_count": 85,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "getFeiertage()[:10]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### School holidays / Schulferien"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Also for this an open API exists: https://ferien-api.de/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [],
   "source": [
    "def getFerientage(jahre = [2017,2018,2019], bundesland = 'BY'):\n",
    "    ferientage_df = pd.DataFrame(columns={'date', 'name'})\n",
    "    for jahr in jahre:\n",
    "        r = requests.get(f'https://ferien-api.de/api/v1/holidays/{bundesland}/{jahr}')\n",
    "        json = r.json()\n",
    "\n",
    "        for f in json:\n",
    "            start = datetime.strptime(f['start'], '%Y-%m-%dT%H:%MZ')\n",
    "            end = datetime.strptime(f['end'], '%Y-%m-%dT%H:%MZ')\n",
    "            dates = pd.date_range(start, end)\n",
    "\n",
    "            df = pd.DataFrame({'date': dates, 'name':f['name']})\n",
    "\n",
    "            ferientage_df = ferientage_df.append(df)\n",
    "    return ferientage_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "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>date</th>\n",
       "      <th>name</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2017-02-27</td>\n",
       "      <td>winterferien</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2017-02-28</td>\n",
       "      <td>winterferien</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2017-03-01</td>\n",
       "      <td>winterferien</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2017-03-02</td>\n",
       "      <td>winterferien</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2017-03-03</td>\n",
       "      <td>winterferien</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>2017-03-04</td>\n",
       "      <td>winterferien</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2017-04-10</td>\n",
       "      <td>osterferien</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2017-04-11</td>\n",
       "      <td>osterferien</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2017-04-12</td>\n",
       "      <td>osterferien</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2017-04-13</td>\n",
       "      <td>osterferien</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        date          name\n",
       "0 2017-02-27  winterferien\n",
       "1 2017-02-28  winterferien\n",
       "2 2017-03-01  winterferien\n",
       "3 2017-03-02  winterferien\n",
       "4 2017-03-03  winterferien\n",
       "5 2017-03-04  winterferien\n",
       "0 2017-04-10   osterferien\n",
       "1 2017-04-11   osterferien\n",
       "2 2017-04-12   osterferien\n",
       "3 2017-04-13   osterferien"
      ]
     },
     "execution_count": 87,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "getFerientage()[:10]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Add as additional columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [],
   "source": [
    "feiertage_df = getFeiertage()\n",
    "ferientage_df = getFerientage()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [],
   "source": [
    "def add_date_feature(df, name, dates):\n",
    "    df[name] = False\n",
    "    df.loc[df['date'].isin(dates), name] = True"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "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>uhrzeit_start</th>\n",
       "      <th>uhrzeit_ende</th>\n",
       "      <th>datumWeek</th>\n",
       "      <th>zaehlstelle</th>\n",
       "      <th>richtung_1</th>\n",
       "      <th>richtung_2</th>\n",
       "      <th>gesamt</th>\n",
       "      <th>min-temp</th>\n",
       "      <th>max-temp</th>\n",
       "      <th>niederschlag</th>\n",
       "      <th>...</th>\n",
       "      <th>datumDayofyear</th>\n",
       "      <th>datumIs_month_end</th>\n",
       "      <th>datumIs_month_start</th>\n",
       "      <th>datumIs_quarter_end</th>\n",
       "      <th>datumIs_quarter_start</th>\n",
       "      <th>datumIs_year_end</th>\n",
       "      <th>datumIs_year_start</th>\n",
       "      <th>datumElapsed</th>\n",
       "      <th>feiertag</th>\n",
       "      <th>ferientag</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>00:00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>1</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>138</td>\n",
       "      <td>14</td>\n",
       "      <td>152</td>\n",
       "      <td>2.7</td>\n",
       "      <td>5.2</td>\n",
       "      <td>1.6</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>1546300800</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1 rows × 27 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "  uhrzeit_start  uhrzeit_ende  datumWeek zaehlstelle  richtung_1  richtung_2  \\\n",
       "0      00:00:00         23.59          1      Arnulf         138          14   \n",
       "\n",
       "   gesamt  min-temp  max-temp  niederschlag  ...  datumDayofyear  \\\n",
       "0     152       2.7       5.2           1.6  ...               1   \n",
       "\n",
       "   datumIs_month_end datumIs_month_start  datumIs_quarter_end  \\\n",
       "0              False                True                False   \n",
       "\n",
       "   datumIs_quarter_start  datumIs_year_end  datumIs_year_start  datumElapsed  \\\n",
       "0                   True             False                True    1546300800   \n",
       "\n",
       "   feiertag  ferientag  \n",
       "0      True       True  \n",
       "\n",
       "[1 rows x 27 columns]"
      ]
     },
     "execution_count": 90,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = getDataframe(prefixes=('rad2019'), zaehlstelle='Arnulf')\n",
    "df['date'] = pd.to_datetime(df['datum'], format=\"%Y.%m.%d\")\n",
    "df = add_datepart(df, 'datum')\n",
    "add_date_feature(df, 'feiertag', feiertage_df['date'])\n",
    "add_date_feature(df, 'ferientag', ferientage_df['date'])\n",
    "df[:1]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.2 RandomForest with new data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [],
   "source": [
    "dep_var = 'gesamt' # dependent variable, the number we want to predict\n",
    "usedColumns = np.append(xs_final.columns.values, ['feiertag', 'ferientag'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "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>uhrzeit_start</th>\n",
       "      <th>uhrzeit_ende</th>\n",
       "      <th>datumWeek</th>\n",
       "      <th>zaehlstelle</th>\n",
       "      <th>gesamt</th>\n",
       "      <th>min-temp</th>\n",
       "      <th>max-temp</th>\n",
       "      <th>niederschlag</th>\n",
       "      <th>bewoelkung</th>\n",
       "      <th>sonnenstunden</th>\n",
       "      <th>...</th>\n",
       "      <th>datumDayofyear</th>\n",
       "      <th>datumIs_month_end</th>\n",
       "      <th>datumIs_month_start</th>\n",
       "      <th>datumIs_quarter_end</th>\n",
       "      <th>datumIs_quarter_start</th>\n",
       "      <th>datumIs_year_end</th>\n",
       "      <th>datumIs_year_start</th>\n",
       "      <th>datumElapsed</th>\n",
       "      <th>feiertag</th>\n",
       "      <th>ferientag</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>00:00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>52</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>135</td>\n",
       "      <td>-6.1</td>\n",
       "      <td>2.6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>36</td>\n",
       "      <td>5.9</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>1483228800</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>00:00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>1</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>400</td>\n",
       "      <td>-5.4</td>\n",
       "      <td>0.8</td>\n",
       "      <td>3.2</td>\n",
       "      <td>69</td>\n",
       "      <td>0.3</td>\n",
       "      <td>...</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>1483315200</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>00:00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>1</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>262</td>\n",
       "      <td>-2.1</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>85</td>\n",
       "      <td>3.6</td>\n",
       "      <td>...</td>\n",
       "      <td>3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>1483401600</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>00:00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>1</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>207</td>\n",
       "      <td>-0.9</td>\n",
       "      <td>1.3</td>\n",
       "      <td>2.2</td>\n",
       "      <td>94</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>1483488000</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>00:00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>1</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>224</td>\n",
       "      <td>-7.1</td>\n",
       "      <td>-0.2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>88</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>1483574400</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>52</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>342</td>\n",
       "      <td>1.7</td>\n",
       "      <td>4.9</td>\n",
       "      <td>4.9</td>\n",
       "      <td>98</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>361</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>1577404800</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>52</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>374</td>\n",
       "      <td>-2.1</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>69</td>\n",
       "      <td>3.2</td>\n",
       "      <td>...</td>\n",
       "      <td>362</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>1577491200</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>52</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>285</td>\n",
       "      <td>-4.8</td>\n",
       "      <td>2.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>26</td>\n",
       "      <td>7.9</td>\n",
       "      <td>...</td>\n",
       "      <td>363</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>1577577600</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>1</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>602</td>\n",
       "      <td>-4.4</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>9</td>\n",
       "      <td>8.0</td>\n",
       "      <td>...</td>\n",
       "      <td>364</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>1577664000</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>00:00</td>\n",
       "      <td>23.59</td>\n",
       "      <td>1</td>\n",
       "      <td>Arnulf</td>\n",
       "      <td>363</td>\n",
       "      <td>-0.5</td>\n",
       "      <td>8.6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>60</td>\n",
       "      <td>6.8</td>\n",
       "      <td>...</td>\n",
       "      <td>365</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>1577750400</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1034 rows × 25 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   uhrzeit_start uhrzeit_ende  datumWeek zaehlstelle  gesamt  min-temp  \\\n",
       "0       00:00:00        23.59         52      Arnulf     135      -6.1   \n",
       "1       00:00:00        23.59          1      Arnulf     400      -5.4   \n",
       "2       00:00:00        23.59          1      Arnulf     262      -2.1   \n",
       "3       00:00:00        23.59          1      Arnulf     207      -0.9   \n",
       "4       00:00:00        23.59          1      Arnulf     224      -7.1   \n",
       "..           ...          ...        ...         ...     ...       ...   \n",
       "26         00:00        23.59         52      Arnulf     342       1.7   \n",
       "27         00:00        23.59         52      Arnulf     374      -2.1   \n",
       "28         00:00        23.59         52      Arnulf     285      -4.8   \n",
       "29         00:00        23.59          1      Arnulf     602      -4.4   \n",
       "30         00:00        23.59          1      Arnulf     363      -0.5   \n",
       "\n",
       "    max-temp  niederschlag  bewoelkung  sonnenstunden  ... datumDayofyear  \\\n",
       "0        2.6           0.0          36            5.9  ...              1   \n",
       "1        0.8           3.2          69            0.3  ...              2   \n",
       "2        0.6           0.0          85            3.6  ...              3   \n",
       "3        1.3           2.2          94            0.0  ...              4   \n",
       "4       -0.2           0.0          88            0.0  ...              5   \n",
       "..       ...           ...         ...            ...  ...            ...   \n",
       "26       4.9           4.9          98            0.0  ...            361   \n",
       "27       2.0           0.0          69            3.2  ...            362   \n",
       "28       2.5           0.0          26            7.9  ...            363   \n",
       "29       6.0           0.0           9            8.0  ...            364   \n",
       "30       8.6           0.0          60            6.8  ...            365   \n",
       "\n",
       "    datumIs_month_end  datumIs_month_start  datumIs_quarter_end  \\\n",
       "0               False                 True                False   \n",
       "1               False                False                False   \n",
       "2               False                False                False   \n",
       "3               False                False                False   \n",
       "4               False                False                False   \n",
       "..                ...                  ...                  ...   \n",
       "26              False                False                False   \n",
       "27              False                False                False   \n",
       "28              False                False                False   \n",
       "29              False                False                False   \n",
       "30               True                False                 True   \n",
       "\n",
       "    datumIs_quarter_start  datumIs_year_end  datumIs_year_start  datumElapsed  \\\n",
       "0                    True             False                True    1483228800   \n",
       "1                   False             False               False    1483315200   \n",
       "2                   False             False               False    1483401600   \n",
       "3                   False             False               False    1483488000   \n",
       "4                   False             False               False    1483574400   \n",
       "..                    ...               ...                 ...           ...   \n",
       "26                  False             False               False    1577404800   \n",
       "27                  False             False               False    1577491200   \n",
       "28                  False             False               False    1577577600   \n",
       "29                  False             False               False    1577664000   \n",
       "30                  False              True               False    1577750400   \n",
       "\n",
       "    feiertag  ferientag  \n",
       "0       True      False  \n",
       "1      False      False  \n",
       "2      False      False  \n",
       "3      False      False  \n",
       "4      False      False  \n",
       "..       ...        ...  \n",
       "26     False       True  \n",
       "27     False       True  \n",
       "28     False       True  \n",
       "29     False       True  \n",
       "30     False       True  \n",
       "\n",
       "[1034 rows x 25 columns]"
      ]
     },
     "execution_count": 92,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = getDataframe(zaehlstelle='Arnulf', prefixes = ('rad2017', 'rad2018', 'rad2019'))\n",
    "df['date'] = pd.to_datetime(df['datum'], format=\"%Y.%m.%d\")\n",
    "df = add_datepart(df, 'datum')\n",
    "add_date_feature(df, 'feiertag', feiertage_df['date'])\n",
    "add_date_feature(df, 'ferientag', ferientage_df['date'])\n",
    "df = df.drop(['richtung_1', 'richtung_2'], axis=1)\n",
    "# df = df[list(usedColumns) + [dep_var]] reduce columns after train valid set is created because year is used\n",
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create Tabular Pandas\n",
    "procs = [Categorify, FillMissing] # processings or transformations\n",
    "cond = (df.datumYear<2019)\n",
    "train_idx = np.where(cond)[0]\n",
    "valid_idx = np.where(~cond)[0]\n",
    "splits = (list(train_idx),list(valid_idx))\n",
    "\n",
    "cont, cat = cont_cat_split(df, 1, dep_var=dep_var)\n",
    "to = TabularPandas(df, procs, cat, cont, y_names=dep_var, splits=splits)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [],
   "source": [
    "xs,y = to.train.xs, to.train.y\n",
    "valid_xs, valid_y = to.valid.xs, to.valid.y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [],
   "source": [
    "xs = xs[list(usedColumns)]\n",
    "valid_xs = valid_xs[list(usedColumns)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [],
   "source": [
    "m = rf(xs,y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "272.74979"
      ]
     },
     "execution_count": 97,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "m_rmse(m, valid_xs, valid_y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "😳better then version without holidays: first run 273 vs 287 old"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Other metrics?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(210.09701086777324, 0.3237236660910646, 0.20322548194268697)"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rf_preds = m.predict(valid_xs)\n",
    "mae(rf_preds, valid_y), mape(rf_preds, valid_y), nmae(rf_preds, valid_y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "How important are the new features?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "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>cols</th>\n",
       "      <th>imp</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>max-temp</td>\n",
       "      <td>0.355302</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>datumDayofweek</td>\n",
       "      <td>0.199496</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>sonnenstunden</td>\n",
       "      <td>0.149990</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>min-temp</td>\n",
       "      <td>0.138572</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>datumDayofyear</td>\n",
       "      <td>0.054694</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>datumWeek</td>\n",
       "      <td>0.031816</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>niederschlag</td>\n",
       "      <td>0.029401</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>bewoelkung</td>\n",
       "      <td>0.019608</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>ferientag</td>\n",
       "      <td>0.007093</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>feiertag</td>\n",
       "      <td>0.007064</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>datumDay</td>\n",
       "      <td>0.006964</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              cols       imp\n",
       "0         max-temp  0.355302\n",
       "1   datumDayofweek  0.199496\n",
       "2    sonnenstunden  0.149990\n",
       "3         min-temp  0.138572\n",
       "4   datumDayofyear  0.054694\n",
       "6        datumWeek  0.031816\n",
       "5     niederschlag  0.029401\n",
       "7       bewoelkung  0.019608\n",
       "10       ferientag  0.007093\n",
       "9         feiertag  0.007064\n",
       "8         datumDay  0.006964"
      ]
     },
     "execution_count": 99,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rf_feat_importance(m, xs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Take another look at the biggest errors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [],
   "source": [
    "rf_pred = m.predict(valid_xs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [],
   "source": [
    "error = np.subtract(rf_pred, valid_y).abs()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['uhrzeit_start', 'uhrzeit_ende', 'datumWeek', 'zaehlstelle', 'gesamt',\n",
       "       'min-temp', 'max-temp', 'niederschlag', 'bewoelkung', 'sonnenstunden',\n",
       "       'date', 'datumYear', 'datumMonth', 'datumDay', 'datumDayofweek',\n",
       "       'datumDayofyear', 'datumIs_month_end', 'datumIs_month_start',\n",
       "       'datumIs_quarter_end', 'datumIs_quarter_start', 'datumIs_year_end',\n",
       "       'datumIs_year_start', 'datumElapsed', 'feiertag', 'ferientag', 'pred',\n",
       "       'error'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 102,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.options.mode.chained_assignment = None # avoid SettingWithCopyError, is this a good idea?\n",
    "df_errors = df.iloc[valid_idx]\n",
    "df_errors['pred'] = rf_pred\n",
    "df_errors['error'] = error\n",
    "df_errors.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>date</th>\n",
       "      <th>datumDayofweek</th>\n",
       "      <th>min-temp</th>\n",
       "      <th>feiertag</th>\n",
       "      <th>ferientag</th>\n",
       "      <th>error</th>\n",
       "      <th>gesamt</th>\n",
       "      <th>pred</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>2019-05-30</td>\n",
       "      <td>3</td>\n",
       "      <td>8.6</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>987.986242</td>\n",
       "      <td>593</td>\n",
       "      <td>1580.986242</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>2019-04-19</td>\n",
       "      <td>4</td>\n",
       "      <td>5.5</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>987.227566</td>\n",
       "      <td>785</td>\n",
       "      <td>1772.227566</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>2019-04-22</td>\n",
       "      <td>0</td>\n",
       "      <td>6.5</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>955.340798</td>\n",
       "      <td>843</td>\n",
       "      <td>1798.340798</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>2019-05-31</td>\n",
       "      <td>4</td>\n",
       "      <td>11.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>931.509480</td>\n",
       "      <td>751</td>\n",
       "      <td>1682.509480</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>2019-05-24</td>\n",
       "      <td>4</td>\n",
       "      <td>6.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>729.712157</td>\n",
       "      <td>1194</td>\n",
       "      <td>1923.712157</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>2019-08-15</td>\n",
       "      <td>3</td>\n",
       "      <td>9.8</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>717.304553</td>\n",
       "      <td>769</td>\n",
       "      <td>1486.304553</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>2019-05-27</td>\n",
       "      <td>0</td>\n",
       "      <td>11.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>660.980998</td>\n",
       "      <td>817</td>\n",
       "      <td>1477.980998</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>2019-10-22</td>\n",
       "      <td>1</td>\n",
       "      <td>10.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>645.521575</td>\n",
       "      <td>1873</td>\n",
       "      <td>1227.478425</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>2019-09-20</td>\n",
       "      <td>4</td>\n",
       "      <td>3.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>631.003814</td>\n",
       "      <td>2011</td>\n",
       "      <td>1379.996186</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2019-05-01</td>\n",
       "      <td>2</td>\n",
       "      <td>4.7</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>607.479866</td>\n",
       "      <td>1061</td>\n",
       "      <td>1668.479866</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         date  datumDayofweek  min-temp  feiertag  ferientag       error  \\\n",
       "29 2019-05-30               3       8.6      True      False  987.986242   \n",
       "18 2019-04-19               4       5.5      True       True  987.227566   \n",
       "21 2019-04-22               0       6.5      True       True  955.340798   \n",
       "30 2019-05-31               4      11.7     False      False  931.509480   \n",
       "23 2019-05-24               4       6.3     False      False  729.712157   \n",
       "14 2019-08-15               3       9.8      True       True  717.304553   \n",
       "26 2019-05-27               0      11.2     False      False  660.980998   \n",
       "21 2019-10-22               1      10.5     False      False  645.521575   \n",
       "19 2019-09-20               4       3.4     False      False  631.003814   \n",
       "0  2019-05-01               2       4.7      True      False  607.479866   \n",
       "\n",
       "    gesamt         pred  \n",
       "29     593  1580.986242  \n",
       "18     785  1772.227566  \n",
       "21     843  1798.340798  \n",
       "30     751  1682.509480  \n",
       "23    1194  1923.712157  \n",
       "14     769  1486.304553  \n",
       "26     817  1477.980998  \n",
       "21    1873  1227.478425  \n",
       "19    2011  1379.996186  \n",
       "0     1061  1668.479866  "
      ]
     },
     "execution_count": 103,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_errors.sort_values(by = 'error', ascending=False)[['date', 'datumDayofweek', 'min-temp', 'feiertag', 'ferientag', 'error', 'gesamt', 'pred']][:10]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Plot biggest errors\n",
    "\n",
    "Can I see clusters when plotting the biggest errors?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABJgAAANsCAYAAAAX8BIxAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/Il7ecAAAACXBIWXMAAAsTAAALEwEAmpwYAAA1cElEQVR4nO3de5RldX3n/c/XbgNGwRvIsEADTtAAAo20NzTGHjQSn0GNVxxHmwkZ5zGSmOVEo8kTY0zMOHk0iY63Md7QZRSNBnAtfZKIGmcQMW1sL4AIaiOtDGJ7AxMU8Pf8UbuxaKubbr5ddaq6X6+19qpzfmfvfX6nenfRvGuffWqMEQAAAAC4rW436wkAAAAAsLIJTAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAAAAtAhMAAArRFW9rar+ZN79Z1fV1VV1XVXdfZZzAwD2bgITAMBuUFUfq6rvVNU+S/R8t0/y50l+eYxxpzHGlqV4XgCAhQhMAABNVXVYkl9MMpI8dgfrrdqNT3tQkn2TXLQb9wkAcJsITAAAfc9M8skkb0uyfuvg9Ja211fVB6vqB0nWVdWmqvqdqvpcVX2vqs6qqn2n9U+rqv89f8dVNarq57cZu0+SS6e7362qjyziawMAuFUCEwBA3zOTvHNaHl1VB8177D8keVmS/ZJsjUdPSXJyksOTHJvktF15sjHGl5IcPd29yxjj393mmQMA7AYCEwBAQ1U9LMnPJXnPGOPTSb6cuai01TljjPPHGD8eY1w/jb16jPGNMca3k3wgyZolnTQAwG4mMAEA9KxP8vdjjG9N9/86894ml+TKBbb5P/Nu/0uSOy3S3AAAlsTqWU8AAGClqqo7ZO7tbquqams02ifJXarquOn+2IVd/iDJz87b/7/ZLRMFAFhkAhMAwG33+CQ3JTkmyY/mjb8nc9dl2lWfTXJ0Va1J8sUkL+lNDwBgaXiLHADAbbc+yVvHGF8bY/yfrUuS1yR5enbxl3nTxbtfmuTDSS7LTy4KDgCwrNUYu3LWNgAAAADckjOYAAAAAGgRmAAAAABoEZgAAAAAaBGYAAAAAGjZpU82WUkOOOCAcdhhh816GgAAAAB7jE9/+tPfGmMcuO34HhuYDjvssGzYsGHW0wAAAADYY1TVFQuNe4scAAAAAC0CEwAAAAAtAhMAAAAALXvsNZgAAACA5eWGG27I5s2bc/311896KtyKfffdN4ceemhuf/vb79T6AhMAAACwJDZv3pz99tsvhx12WKpq1tNhO8YY2bJlSzZv3pzDDz98p7bxFjkAAABgSVx//fW5+93vLi4tc1WVu9/97rt0ppnABAAAACwZcWll2NU/J4EJAAAAgBaBCQAAAJiJqt277IxVq1ZlzZo1ud/97pdTTjkl3/3ud3fLa3nb296WM844Y7fs67DDDssxxxyTNWvWZM2aNfnEJz6xW/a7rY0bN+aDH/zgbtmXwAQAAADsNe5whztk48aN+cIXvpC73e1uee1rXzvrKS3oox/9aDZu3JiNGzfmxBNP3Kltbrzxxl16DoEJAAAAoOkhD3lIvv71rydJPvWpT+XEE0/M8ccfnxNPPDGXXnppkrkzk57whCfk5JNPzhFHHJEXvOAFN2//1re+Nfe5z33yS7/0Szn//PNvHr/iiity0kkn5dhjj81JJ52Ur33ta0mS0047Lc9+9rOzbt263Pve984//uM/5td+7ddy5JFH5rTTTtvhXHe0z+c973lZt25dfvd3fzdf/vKXc/LJJ+eEE07IL/7iL+aLX/xikuS9731v7ne/++W4447Lwx/+8PzoRz/Ki1/84px11llZs2ZNzjrrrNb3cnVrawAAAIAV6Kabbsp5552X008/PUnyC7/wC/n4xz+e1atX58Mf/nB+7/d+L+973/uSzJ3p85nPfCb77LNP7nvf++Y3f/M3s3r16vzhH/5hPv3pT+fOd75z1q1bl+OPPz5JcsYZZ+SZz3xm1q9fn7e85S35rd/6rZx99tlJku985zv5yEc+knPPPTennHJKzj///LzpTW/KAx7wgGzcuDFr1qxJkqxbty6rVq3KPvvskwsvvHCH+/zSl76UD3/4w1m1alVOOumkvOENb8gRRxyRCy+8ML/xG7+Rj3zkI3npS1+av/u7v8shhxyS7373u/mZn/mZvPSlL82GDRvymte8pv39FJgAAACAvca//uu/Zs2aNdm0aVNOOOGEPOpRj0qSfO9738v69etz2WWXpapyww033LzNSSedlDvf+c5JkqOOOipXXHFFvvWtb+URj3hEDjzwwCTJU5/61HzpS19KklxwwQV5//vfnyR5xjOecYuznk455ZRUVY455pgcdNBBOeaYY5IkRx99dDZt2nRzYProRz+aAw444ObtdrTPJz/5yVm1alWuu+66fOITn8iTn/zkmx/74Q9/mCR56EMfmtNOOy1PecpT8oQnPGE3fCdvyVvkAAAAgL3G1mswXXHFFfnRj3508zWY/uAP/iDr1q3LF77whXzgAx/I9ddff/M2++yzz823V61adfO1jmonryw+f72t+7rd7W53i/3e7na326VrKM3f5x3veMckyY9//OPc5S53ufnaTRs3bswll1ySJHnDG96QP/mTP8mVV16ZNWvWZMuWLTv9XDtDYAIAAAD2One+853z6le/Oq94xStyww035Hvf+14OOeSQJHPXXbo1D3rQg/Kxj30sW7ZsyQ033JD3vve9Nz924okn5t3vfneS5J3vfGce9rCHtee7M/vcf//9c/jhh988lzFGPvvZzyZJvvzlL+dBD3pQXvrSl+aAAw7IlVdemf322y/XXntte26JwAQAAADMyBi7d9lVxx9/fI477ri8+93vzgte8IK86EUvykMf+tDcdNNNt7rtwQcfnJe85CV5yEMekkc+8pG5//3vf/Njr371q/PWt741xx57bN7xjnfkVa961a5Pbhs7u893vvOdefOb35zjjjsuRx99dM4555wkyfOf//wcc8wxud/97peHP/zhOe6447Ju3bpcfPHFu+Ui3zVuy5/ACrB27dqxYcOGWU8DAAAAmFxyySU58sgjZz0NdtJCf15V9ekxxtpt13UGEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAADAbVbt32QlXXnll1q1blyOPPDJHH310XvWqVyVJvv3tb+dRj3pUjjjiiDzqUY/Kd77znSTJli1bsm7dutzpTnfKGWeccYt9nXXWWTn22GNz9NFH5wUveMHu/d6sMAITAAAAsNdYvXp1XvnKV+aSSy7JJz/5ybz2ta/NxRdfnJe//OU56aSTctlll+Wkk07Ky1/+8iTJvvvumz/+4z/OK17xilvsZ8uWLXn+85+f8847LxdddFGuvvrqnHfeebN4ScuCwAQAAADsNQ4++ODc//73T5Lst99+OfLII/P1r38955xzTtavX58kWb9+fc4+++wkyR3veMc87GEPy7777nuL/XzlK1/Jfe5znxx44IFJkkc+8pF53/vet3QvZJkRmAAAAIC90qZNm/KZz3wmD3rQg3L11Vfn4IMPTjIXob75zW/ucNuf//mfzxe/+MVs2rQpN954Y84+++xceeWVSzHtZWn1rCcAAAAAsNSuu+66PPGJT8xf/uVfZv/999/l7e9617vm9a9/fZ761Kfmdre7XU488cR85StfWYSZrgzOYAIAAAD2KjfccEOe+MQn5ulPf3qe8IQnJEkOOuigXHXVVUmSq666Kve4xz1udT+nnHJKLrzwwlxwwQW5733vmyOOOGJR572cCUwAAADAXmOMkdNPPz1HHnlknve85908/tjHPjZnnnlmkuTMM8/M4x73uFvd19a30X3nO9/J6173uvz6r//64kx6BfAWOQAAAGA2xljypzz//PPzjne8I8ccc0zWrFmTJPnTP/3TvPCFL8xTnvKUvPnNb8697nWvvPe97715m8MOOyzf//7386Mf/Shnn312/v7v/z5HHXVUnvvc5+azn/1skuTFL35x7nOf+yz561kuBCYAAABgr/Gwhz0sYzth67zzzltwfNOmTQuOv+td79pd01rxvEUOAAAAgBaBCQAAAICWRQtMVXXPqvpoVV1SVRdV1XOn8ZdU1derauO0PGbeNi+qqsur6tKqevS88ROq6vPTY6+uqlqseQMAAACLZ3tvT2N52dU/p8U8g+nGJP91jHFkkgcneU5VHTU99hdjjDXT8sEkmR47NcnRSU5O8rqqWjWt//okz0pyxLScvIjzBgAAABbBvvvumy1btohMy9wYI1u2bMm+++6709ss2kW+xxhXJblqun1tVV2S5JAdbPK4JO8eY/wwyVer6vIkD6yqTUn2H2NckCRV9fYkj0/yocWaOwAAALD7HXroodm8eXOuueaaWU+FW7Hvvvvm0EMP3en1l+RT5KrqsCTHJ7kwyUOTnFFVz0yyIXNnOX0nc/Hpk/M22zyN3TDd3nZ8oed5VubOdMq97nWv3fsiAIDlZ+u75v0WFABWhNvf/vY5/PDDZz0NFsGiX+S7qu6U5H1JfnuM8f3Mvd3t3yZZk7kznF65ddUFNh87GP/pwTHeOMZYO8ZYe+CBB3anDgAAAMBOWNTAVFW3z1xceucY4/1JMsa4eoxx0xjjx0n+KskDp9U3J7nnvM0PTfKNafzQBcYBAAAAWAYW81PkKsmbk1wyxvjzeeMHz1vtV5N8Ybp9bpJTq2qfqjo8cxfz/tR0Ladrq+rB0z6fmeScxZo3AAAAALtmMa/B9NAkz0jy+araOI39XpKnVdWazL3NbVOS/5IkY4yLquo9SS7O3CfQPWeMcdO03bOTvC3JHTJ3cW8X+AYAAABYJmpP/WjAtWvXjg0bNsx6GgDAYnKRbwCAJVVVnx5jrN12fNEv8g0AAADAnk1gAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBl0QJTVd2zqj5aVZdU1UVV9dxp/G5V9Q9Vddn09a7ztnlRVV1eVZdW1aPnjZ9QVZ+fHnt1VdVizRsAAFhA1dwCAAtYzDOYbkzyX8cYRyZ5cJLnVNVRSV6Y5LwxxhFJzpvuZ3rs1CRHJzk5yeuqatW0r9cneVaSI6bl5EWcNwAAAAC7YNEC0xjjqjHGP0+3r01ySZJDkjwuyZnTamcmefx0+3FJ3j3G+OEY46tJLk/ywKo6OMn+Y4wLxhgjydvnbQMAAADAjC3JNZiq6rAkxye5MMlBY4yrkrkIleQe02qHJLly3mabp7FDptvbjgMAAACwDCx6YKqqOyV5X5LfHmN8f0erLjA2djC+0HM9q6o2VNWGa665ZtcnCwAAAMAuW9TAVFW3z1xceucY4/3T8NXT294yff3mNL45yT3nbX5okm9M44cuMP5TxhhvHGOsHWOsPfDAA3ffCwEAAABguxbzU+QqyZuTXDLG+PN5D52bZP10e32Sc+aNn1pV+1TV4Zm7mPenprfRXVtVD572+cx52wAAAAAwY6sXcd8PTfKMJJ+vqo3T2O8leXmS91TV6Um+luTJSTLGuKiq3pPk4sx9At1zxhg3Tds9O8nbktwhyYemBQAAAIBloOY+mG3Ps3bt2rFhw4ZZTwMAWEw1XapxD/33DCwr/r4BkKSqPj3GWLvt+JJ8ihwAAAAAey6BCQAAAIAWgQkAAACAFoEJAAAAgBaBCQAAAIAWgQkAAACAFoEJAAAAgBaBCQAAAIAWgQkAWPGqZj0DAIC9m8AEAAAAQIvABAAAAECLwAQAAABAi8AEAAAAQIvABAAAAECLwAQAAABAi8AEAAAAQIvABAAAAECLwAQAAABAy+pZTwCYvapb3h9jNvMAAABgZXIGEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAALtf1dwC7BUEJgAAAABaBCYAAAAAWgQmAAAAAFoEJgAAAABaBCYAAAAAWgQmAAAAAFoEJgAAAABaBCYAAAAAWgQmAAAAAFoEJgAAAABaBCYAAAAAWgQmAAAAAFoEJgAAAABaBCYAAAAAWgQmAABgeama9QwA2EUCEwAAAAAtAhMAAAAALQITAAAAAC2rZz0BYOnNv6zBGLObBwAAAHsGZzABAAAA0CIwAQAAANAiMAEAAADQ4hpMeyjX2AEAAACWijOYAAAAAGgRmAAAAABoEZgAAAAAaBGYAAAAAGgRmAAAAABoEZgAAAAAaBGYAAAAAGhZPesJsGNVt7w/xmzmAQAAALA9zmACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgZfWsJwDs3apueX+M2cwDAACA284ZTAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC0uMg3S85FnQEAAGDP4gwmAAAAAFoEJgAAAABaBCYAAAAAWgQmAAAAAFoEJgAAAABaBCYAAAAAWgQmAAAAAFoEJgAAAABaBCYAAAAAWgQmAAAAAFoEJgAAAABaBCYAAAAAWgQmAAAAAFoEJgAAAABaBCYAAAAAWgQmAAAAAFoEJgAAAABaBCYAAAAAWgQmAAAAAFoEJgAAAABaBCYAAAAAWgQmAAAAAFoEJgAAAABaBCYAAAAAWgQmAAAAAFoEJgAAAABaVs96AsDepeont8eY3TwAAADYfZzBBAAAAECLwAQAAABAi8AEAAAAQIvABAAAAECLwAQAAABAi8AEAAAAQIvABAAAAECLwAQAAABAi8AEAAAAQIvABAAAAECLwAQAAABAi8AEAAAAQIvABAAAAECLwAQAAABAi8AEAAAAQIvABAAAAECLwAQAAABAi8AEAAAAQIvABAAAAECLwAQAAABAi8AEAAAAQMuiBaaqektVfbOqvjBv7CVV9fWq2jgtj5n32Iuq6vKqurSqHj1v/ISq+vz02KurqhZrzgAAAADsusU8g+ltSU5eYPwvxhhrpuWDSVJVRyU5NcnR0zavq6pV0/qvT/KsJEdMy0L7BAAAAGBGFi0wjTE+nuTbO7n645K8e4zxwzHGV5NcnuSBVXVwkv3HGBeMMUaStyd5/KJMGAAAAIDbZBbXYDqjqj43vYXurtPYIUmunLfO5mnskOn2tuMLqqpnVdWGqtpwzTXX7O55AwAAALCApQ5Mr0/yb5OsSXJVkldO4wtdV2nsYHxBY4w3jjHWjjHWHnjggc2pAgAAALAzljQwjTGuHmPcNMb4cZK/SvLA6aHNSe45b9VDk3xjGj90gXEAAAAAloklDUzTNZW2+tUkWz9h7twkp1bVPlV1eOYu5v2pMcZVSa6tqgdPnx73zCTnLOWcAQAAANix1Yu146p6V5JHJDmgqjYn+cMkj6iqNZl7m9umJP8lScYYF1XVe5JcnOTGJM8ZY9w07erZmftEujsk+dC0AADslWqbCwiM7V48AABg6dTYQ/9Vsnbt2rFhw4ZZT6Pttv4jcv52y+2P2D+MZ2/b42Mp/0xm+dzAHmj6IVIZe83PDz83mZmtB99SHHRVDu49wVIeM8CSqapPjzHWbjs+i0+RAwAAAGAPsmhvkQP2bH6DDgAAwFbOYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAdqhq4dsA7EDVXvVDU2ACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgZfX2HqiqzycZCz2UZIwxjl20WQEAAACwYmw3MCX590s2CwAAAABWrO0GpjHGFUs5EQAAAABWplu9BlNVPbiq/qmqrquqH1XVTVX1/aWYHAAAAADL385c5Ps1SZ6W5LIkd0jy60n+x2JOCgAAAICVY0fXYLrZGOPyqlo1xrgpyVur6hOLPC8AAAAAVoidCUz/UlU/k2RjVf1ZkquS3HFxpwUAAADASrEzb5F7xrTeGUl+kOSeSZ6wmJMCAAAAYOXYmcD0+DHG9WOM748x/miM8bwk/36xJwYAAADAyrAzgWn9AmOn7eZ5AAAAALBCbfcaTFX1tCT/IcnhVXXuvIf2T7JlsScGAAAAwMqwo4t8fyJzF/Q+IMkr541fm+RzizkpAAAAAFaO7QamMcYVSa5I8pCqOijJA6aHLhlj3LgUkwNWtqpb3h9jNvMAAABgcd3qNZiq6slJPpXkyUmekuTCqnrSYk8MAADY+2z9BdW2v6gCYHnb0Vvktvp/kjxgjPHNJKmqA5N8OMnfLObEAAAAAFgZduZT5G63NS5NtuzkdgAAAADsBXbmDKYPVdXfJXnXdP+pST64eFMCAAAAYCXZmTORRpL/meTYJMcleeOizggAAACAFWVnzmB61Bjjd5O8f+tAVf1Rkt9dtFkBAAAAsGJsNzBV1bOT/EaSe1fV5+Y9tF+S8xd7YgAAAACsDDs6g+mvk3woyX9L8sJ549eOMb69qLMCAAAAYMXYbmAaY3wvyfeSPG3ppgMAAADASrMzF/kGAAAAgO3amYt8AwAA7PGqfnJ7jNnNA2AlcgYTAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC2rZz0BYDZG6uZbAAAA0OEMJgAAAABaBCYAAAAAWgQmAAAAAFoEJgAAAHarqoVvA3sugQkAAACAFoEJAAAAgBaBCQAAAIAWgQkAAACAFoEJAAAAgBaBCQAAAICW1bOeALC4tv1Y2DFmMw8AAAD2XM5gAgAAAKBFYAIAAACgRWACkiQjlZG69RUBAABgGwITAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC2rZz0B+qpueX+M2cwDAAAA2Ds5gwkAAACAFoEJAAAAgBaBCQAAAIAWgQkAAACAFoEJAAAAgBaBCQAAAIAWgQkAAACAFoEJAAAAgBaBCQAAAIAWgQkAAACAFoEJAAAAgBaBCQAAAICW1bOeALAyVM16BgAAACxXzmACAAAAoEVgAgAAAKBFYAIAAACgxTWYgAXNv+bSGLObBwAAAMufM5gAAAAAaBGYAAAAAGgRmAAAAABoEZgAAAAAaBGYAAAAAGgRmAAAAABoEZgAAAAAaBGYAAAAAGgRmAAAAABoEZgAAAAAaBGYAAAAAGgRmAAAAABoEZiYiZHKSM16GgAAAMBuIDABAAAA0CIwAQAAANAiMAEAAADQIjABAAAA0CIwAQAAANAiMAEAAADQIjABAAAA0CIwAQAAANAiMAEAAADQIjABAAAA0CIwAQAAANAiMAEAAADQIjABAAAA0CIwAQAAANAiMAEAAADQsmiBqareUlXfrKovzBu7W1X9Q1VdNn2967zHXlRVl1fVpVX16HnjJ1TV56fHXl1VtVhzBgAAAGDXLeYZTG9LcvI2Yy9Mct4Y44gk5033U1VHJTk1ydHTNq+rqlXTNq9P8qwkR0zLtvsEAAAAYIYWLTCNMT6e5NvbDD8uyZnT7TOTPH7e+LvHGD8cY3w1yeVJHlhVByfZf4xxwRhjJHn7vG0AAAAAWAaW+hpMB40xrkqS6es9pvFDklw5b73N09gh0+1txxdUVc+qqg1VteGaa67ZrRMHAAAAYGHL5SLfC11XaexgfEFjjDeOMdaOMdYeeOCBu21yAAAAAGzfUgemq6e3vWX6+s1pfHOSe85b79Ak35jGD11gHAAAAIBlYqkD07lJ1k+31yc5Z974qVW1T1UdnrmLeX9qehvdtVX14OnT4545bxsAAAAAloHVi7XjqnpXkkckOaCqNif5wyQvT/Keqjo9ydeSPDlJxhgXVdV7klyc5MYkzxlj3DTt6tmZ+0S6OyT50LQAAAAAsEwsWmAaYzxtOw+dtJ31X5bkZQuMb0hyv904NQAAAAB2o+VykW8AAAAAViiBCQAAAIAWgQkAAACAFoEJAAAAgBaBCQAAAIAWgQkAAACAFoEJAAAAgBaBCQAAAIAWgQkAAACAFoEJAAAAgBaBCQAAAIAWgQkAAACAFoEJAAAAgBaBCQAAAIAWgQkAAACAFoEJAAAAgBaBCQAAAIAWgQkAAACAFoEJAAAAgBaBCQAAAIAWgQkAAACAFoEJAAAAgBaBCQAAAIAWgQkAAACAFoEJAAAAgBaBCQAAAIAWgQkAAACAFoEJAAAAgBaBCQAAAIAWgQkAAACAFoEJAAAAgBaBCQAAAIAWgQkAAACAFoEJAAAAgBaBCQAAAIAWgQkAAACAFoEJAAAAgBaBCQAAAIAWgQkAAACAFoEJAAAAgBaBCQAAAIAWgQkAAACAFoEJAAAAgBaBaQ81UhmpWU8DAAAA2AsITAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC0CEzAbTZSGalZTwMAAGBZqVr49p5MYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYFoBRiojNetpAAAAACxIYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAFhhfAAIALDcCEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAAAAtAhMAAAAALQITMDM+TQkAGA58G8SgNtOYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAhbVSGWkZj0NAAAAFpHABAAAAECLwAQAAABAi8AEAAAAQIvABAAAAECLwAQAAABAi8AEAAAAQIvABAAAAECLwAQAAABAi8AEAAAAQIvABAAAAECLwAQAAABAi8AEAAAAQIvABAAAAECLwAQAAABAi8AEAAAAQIvABHuBkcpIzXoaAAAA7KEEJgAAAABaBCYAAAAAWgQmAAAAAFoEJgAAAABaBCYAAAAAWgQmAAAAAFoEJgAAAABaBCYAAAAAWgQmAAAAAFoEJgAAAABaBCYAAAAAWgQmAAAAAFoEJgAAAABaBCYAAAAAWgQmAAAAAFoEJgAAAABaBCYAAAAAWgQmAAAAAFoEpj3ESGWkZj0NAAAAYC8kMAEAAADQIjABAAAA0CIwAQAAANAiMAEAAADQIjABS84F6QEAAPYsAhMAAAAALQITAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAAAAtAhMAAAAALQITsNNGKiM162kAAACwzAhMAAAAALTMJDBV1aaq+nxVbayqDdPY3arqH6rqsunrXeet/6KquryqLq2qR89izgAAAAAsbJZnMK0bY6wZY6yd7r8wyXljjCOSnDfdT1UdleTUJEcnOTnJ66pq1SwmDAAAAMBPW05vkXtckjOn22cmefy88XePMX44xvhqksuTPHDppwd7H9dbAgAAYGfMKjCNJH9fVZ+uqmdNYweNMa5KkunrPabxQ5JcOW/bzdPYT6mqZ1XVhqracM011yzS1AEAAACYb/WMnvehY4xvVNU9kvxDVX1xB+sudArFWGjFMcYbk7wxSdauXbvgOgAAAADsXjM5g2mM8Y3p6zeT/G3m3vJ2dVUdnCTT129Oq29Ocs95mx+a5BtLN1sAAAAAdmTJA1NV3bGq9tt6O8kvJ/lCknOTrJ9WW5/knOn2uUlOrap9qurwJEck+dTSzhoAAACA7ZnFW+QOSvK3VbX1+f96jPH/VdU/JXlPVZ2e5GtJnpwkY4yLquo9SS5OcmOS54wxbprBvAEAAABYwJIHpjHGV5Ict8D4liQnbWeblyV52SJPDQAAAIDbYFafIgcAAADAHkJgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgRWACAAAAoEVgAgAAAKBFYAIAAACgZcUEpqo6uaourarLq+qFs54PAAAAAHNWRGCqqlVJXpvkV5IcleRpVXXUbGcFAAAAQLJCAlOSBya5fIzxlTHGj5K8O8njZjwnAAAAAJKsnvUEdtIhSa6cd39zkgdtu1JVPSvJs6a711XVpUswt0VX825V7eo6299mlqYpHZDUt5bj/PY0Cx0fN49NA9tfZ4Ftsv1tdv65F97vvP0ckORbC++VvZDjgR3Z6/57sjM/f7kFP0N2g6U87rb9t8JutJPHgr9bu8MK/Vnl5wXztY6HFfp3YGf83EKDKyUwLfRHMX5qYIw3Jnnj4k+H3aGqNowx1s56HixPjg/mczywI44Pbo1jhK0cC9waxwjzOR52zUp5i9zmJPecd//QJN+Y0VwAAAAAmGelBKZ/SnJEVR1eVT+T5NQk5854TgAAAABkhbxFboxxY1WdkeTvkqxK8pYxxkUznhZ93s7Ijjg+mM/xwI44Prg1jhG2cixwaxwjzOd42AU1xk9dyggAAAAAdtpKeYscAAAAAMuUwAQAAABAi8DETquqe1bVR6vqkqq6qKqeO43frar+oaoum77edRq/+7T+dVX1mm329dSq+ty0nz/bwXOeUFWfr6rLq+rVVVXT+GlVdU1VbZyWX1/M186tW07Hx/TYU6rq4mkff71Yr5uFLafjoar+Yt7Pii9V1XcX8aWzE5bZ8XGvad+fmfbzmMV87dy6ZXZ8/FxVnTft42NVdehivnZ+2oyOh5dV1ZVVdd024/tU1VnTcXJhVR22CC+ZXbTMjpGHV9U/V9WNVfWkxXi9bN8yOxaeV3P/L/K56b8jP7cYr3nZGWNYLDu1JDk4yf2n2/sl+VKSo5L8WZIXTuMvTPLfp9t3TPKwJP93ktfM28/dk3wtyYHT/TOTnLSd5/xUkockqSQfSvIr0/hp8/dpmf2yzI6PI5J8Jsldp/v3mPX3Z29bltPxsM06v5m5D4qY+fdob16W0/GRuYt3Pnu6fVSSTbP+/uztyzI7Pt6bZP10+98lecesvz972zKj4+HB0/Net834byR5w3T71CRnzfr7Y1l2x8hhSY5N8vYkT5r192ZvW5bZsbAuyc9Ot5+9t/y8cAYTO22McdUY45+n29cmuSTJIUkel7m/dJm+Pn5a5wdjjP+d5PptdnXvJF8aY1wz3f9wkidu+3xVdXCS/ccYF4y5v5lv37pvlp9ldnz85ySvHWN8Z3qub+6O18jOW2bHw3xPS/Ku2/7K2B2W2fExkuw/3b5zkm90Xx89y+z4OCrJedPtj05zYAkt9fEw7eOTY4yrFnho/nP+TZKTtp7txuwsp2NkjLFpjPG5JD9uvShuk2V2LHx0jPEv091PJtkrzoAVmLhNplOCj09yYZKDtv6lmr7e41Y2vzzJL1TVYVW1OnN/we+5wHqHJNk87/7maWyrJ06nHP5NVS20PTOyDI6P+yS5T1WdX1WfrKqTb+troW8ZHA9b5/FzSQ5P8pFdfxUslmVwfLwkyX+sqs1JPpi5s9xYJpbB8fHZ/OR/Kn41yX5VdfddfyXsDkt0POzIIUmunJ7zxiTfy9yZDiwTy+AYYZlYZsfC6Zk7O3aPJzCxy6rqTknel+S3xxjf39Xtp7NKnp3krCT/K8mmJDcu9FQLbT59/UCSw8YYx2auKJ+5wLrMwDI5PlZn7m1yj8jcGStvqqq77Opc6Fsmx8NWpyb5mzHGTbs6DxbHMjk+npbkbWOMQ5M8Jsk7qsq/j5aBZXJ8/E6SX6qqzyT5pSRf384+WGRLeDzscBoL7XpX58LiWCbHCMvAcjoWquo/Jlmb5P+9LduvNP4BxS6pqttn7i/rO8cY75+Gr55OL996mvmtvh1pjPGBMcaDxhgPSXJpksuqalX95EK8L83cbxDnn0p4aKa3LowxtowxfjiN/1WSE3bH66NnuRwf02PnjDFuGGN8ddrHEbvjNbLzltHxsNWp8fa4ZWMZHR+nJ3nPtK8Lkuyb5ID+K6RjuRwfY4xvjDGeMMY4PsnvT2Pf200vk520xMfDjmzOdBbDdFbDnZN8+7a9KnanZXSMMGPL6Vioqkdm7r8dj533/657NIGJnTa9x/zNSS4ZY/z5vIfOTbJ+ur0+yTk7sa97TF/vmrkLJr5pjHHTGGPNtLx4On3x2qp68PTcz9y6760/ICaPzdz7a5mh5XR8JDk7cxfWS1UdkLm3zH2l+xrZecvseEhV3TfJXZNcsBteHk3L7Pj4WpKTpn0cmbnAdM1PPRFLZjkdH1V1wLwz2l6U5C274SWyC5b6eLiVXcx/zicl+cgYwxlMM7bMjhFmaDkdC1V1fJL/mbm4tPdcD3YsgyuNW1bGkrkr7I8kn0uycVoek7n3np+X5LLp693mbbMpc7/ZuS5zv/U5ahp/V5KLp+XUHTzn2iRfSPLlJK9JUtP4f0tyUeaujfDRJL8w6+/P3r4ss+Ojkvz5tP3nd7QPy55/PEyPvSTJy2f9fbEsv+MjcxdxPn/678nGJL886+/P3r4ss+PjSdPzfSnJm5LsM+vvz962zOh4+LNpux9PX18yje+buU8WvDxznzx471l/fyzL7hh5wHT/B0m2JLlo1t+fvWlZZsfCh5NcPW8e5876+7MUy9b/eAIAAADAbeItcgAAAAC0CEwAAAAAtAhMAAAAALQITAAAAAC0CEwAAAAAtAhMAABLqKpeUlW/M+t5AADsTgITAAAAAC0CEwDAIquq36+qS6vqw0nuO43956r6p6r6bFW9r6p+tqr2q6qvVtXtp3X2r6pNW+8DACxXAhMAwCKqqhOSnJrk+CRPSPKA6aH3jzEeMMY4LsklSU4fY1yb5GNJ/q9pnVOTvG+MccPSzhoAYNcITAAAi+sXk/ztGONfxhjfT3LuNH6/qvpfVfX5JE9PcvQ0/qYk/2m6/Z+SvHVJZwsAcBsITAAAi28sMPa2JGeMMY5J8kdJ9k2SMcb5SQ6rql9KsmqM8YUlmyUAwG0kMAEALK6PJ/nVqrpDVe2X5JRpfL8kV03XV3r6Ntu8Pcm74uwlAGCFqDEW+oUaAAC7S1X9fpJnJrkiyeYkFyf5QZIXTGOfT7LfGOO0af1/k+SrSQ4eY3x3BlMGANglAhMAwDJTVU9K8rgxxjNmPRcAgJ2xetYTAADgJ6rqfyT5lSSPmfVcAAB2ljOYAAAAAGhxkW8AAAAAWgQmAAAAAFoEJgAAAABaBCYAAAAAWgQmAAAAAFr+f9pGDGdk2T13AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 1440x1080 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "top_errors = df_errors.sort_values(by = 'error', ascending=False)[:20]\n",
    "plotResults(top_errors, top_errors)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There is a week in may in which some errors occur."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.3 NeuronalNet with new data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['max-temp', 'datumDayofweek', 'sonnenstunden', 'min-temp',\n",
       "       'datumDayofyear', 'niederschlag', 'datumWeek', 'bewoelkung',\n",
       "       'feiertag', 'ferientag'], dtype=object)"
      ]
     },
     "execution_count": 105,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "usedColumns[usedColumns!='datumDay']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "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>max-temp</th>\n",
       "      <th>datumDayofweek</th>\n",
       "      <th>sonnenstunden</th>\n",
       "      <th>min-temp</th>\n",
       "      <th>datumDayofyear</th>\n",
       "      <th>niederschlag</th>\n",
       "      <th>datumWeek</th>\n",
       "      <th>bewoelkung</th>\n",
       "      <th>datumDay</th>\n",
       "      <th>feiertag</th>\n",
       "      <th>ferientag</th>\n",
       "      <th>gesamt</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2.6</td>\n",
       "      <td>6</td>\n",
       "      <td>5.9</td>\n",
       "      <td>-6.1</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>52</td>\n",
       "      <td>36</td>\n",
       "      <td>1</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>135</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.8</td>\n",
       "      <td>0</td>\n",
       "      <td>0.3</td>\n",
       "      <td>-5.4</td>\n",
       "      <td>2</td>\n",
       "      <td>3.2</td>\n",
       "      <td>1</td>\n",
       "      <td>69</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>400</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.6</td>\n",
       "      <td>1</td>\n",
       "      <td>3.6</td>\n",
       "      <td>-2.1</td>\n",
       "      <td>3</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>85</td>\n",
       "      <td>3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>262</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1.3</td>\n",
       "      <td>2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-0.9</td>\n",
       "      <td>4</td>\n",
       "      <td>2.2</td>\n",
       "      <td>1</td>\n",
       "      <td>94</td>\n",
       "      <td>4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>207</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-0.2</td>\n",
       "      <td>3</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-7.1</td>\n",
       "      <td>5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>88</td>\n",
       "      <td>5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>224</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>4.9</td>\n",
       "      <td>4</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.7</td>\n",
       "      <td>361</td>\n",
       "      <td>4.9</td>\n",
       "      <td>52</td>\n",
       "      <td>98</td>\n",
       "      <td>27</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>342</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>2.0</td>\n",
       "      <td>5</td>\n",
       "      <td>3.2</td>\n",
       "      <td>-2.1</td>\n",
       "      <td>362</td>\n",
       "      <td>0.0</td>\n",
       "      <td>52</td>\n",
       "      <td>69</td>\n",
       "      <td>28</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>374</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>2.5</td>\n",
       "      <td>6</td>\n",
       "      <td>7.9</td>\n",
       "      <td>-4.8</td>\n",
       "      <td>363</td>\n",
       "      <td>0.0</td>\n",
       "      <td>52</td>\n",
       "      <td>26</td>\n",
       "      <td>29</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>285</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>6.0</td>\n",
       "      <td>0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>-4.4</td>\n",
       "      <td>364</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>9</td>\n",
       "      <td>30</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>602</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>8.6</td>\n",
       "      <td>1</td>\n",
       "      <td>6.8</td>\n",
       "      <td>-0.5</td>\n",
       "      <td>365</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>60</td>\n",
       "      <td>31</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>363</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1034 rows × 12 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    max-temp  datumDayofweek  sonnenstunden  min-temp  datumDayofyear  \\\n",
       "0        2.6               6            5.9      -6.1               1   \n",
       "1        0.8               0            0.3      -5.4               2   \n",
       "2        0.6               1            3.6      -2.1               3   \n",
       "3        1.3               2            0.0      -0.9               4   \n",
       "4       -0.2               3            0.0      -7.1               5   \n",
       "..       ...             ...            ...       ...             ...   \n",
       "26       4.9               4            0.0       1.7             361   \n",
       "27       2.0               5            3.2      -2.1             362   \n",
       "28       2.5               6            7.9      -4.8             363   \n",
       "29       6.0               0            8.0      -4.4             364   \n",
       "30       8.6               1            6.8      -0.5             365   \n",
       "\n",
       "    niederschlag  datumWeek  bewoelkung  datumDay  feiertag  ferientag  gesamt  \n",
       "0            0.0         52          36         1      True      False     135  \n",
       "1            3.2          1          69         2     False      False     400  \n",
       "2            0.0          1          85         3     False      False     262  \n",
       "3            2.2          1          94         4     False      False     207  \n",
       "4            0.0          1          88         5     False      False     224  \n",
       "..           ...        ...         ...       ...       ...        ...     ...  \n",
       "26           4.9         52          98        27     False       True     342  \n",
       "27           0.0         52          69        28     False       True     374  \n",
       "28           0.0         52          26        29     False       True     285  \n",
       "29           0.0          1           9        30     False       True     602  \n",
       "30           0.0          1          60        31     False       True     363  \n",
       "\n",
       "[1034 rows x 12 columns]"
      ]
     },
     "execution_count": 106,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# usedColumns = usedColumns[usedColumns!='datumDay'] #remove datumDay in future? it is the day in the month does it say sth dayInYear cannot say?\n",
    "\n",
    "df_nn = getDataframe(zaehlstelle='Arnulf', prefixes = ('rad2017', 'rad2018', 'rad2019'))\n",
    "df_nn['date'] = pd.to_datetime(df_nn['datum'], format=\"%Y.%m.%d\")\n",
    "df_nn = add_datepart(df_nn, 'datum')\n",
    "add_date_feature(df_nn, 'feiertag', feiertage_df['date'])\n",
    "add_date_feature(df_nn, 'ferientag', ferientage_df['date'])\n",
    "df_nn = df_nn.drop(['richtung_1', 'richtung_2'], axis=1)\n",
    "df_full = df_nn\n",
    "df_nn = df_nn[list(usedColumns) + [dep_var]] #reduce columns after train valid set is created because year is used\n",
    "df_nn"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I use the default value for `max_card=20` in comparison to the first run. Which leads to better continuos and categorical splits in my opinion. e.g. bewoelkung is now continous. I also tested it once with a dayOfYear as category but it let to worse performance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {},
   "outputs": [],
   "source": [
    "cont_nn, cat_nn = cont_cat_split(df_nn, dep_var=dep_var)\n",
    "#cont_nn.append('bewoelkung')\n",
    "#cat_nn.remove('bewoelkung')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(['max-temp',\n",
       "  'sonnenstunden',\n",
       "  'min-temp',\n",
       "  'datumDayofyear',\n",
       "  'niederschlag',\n",
       "  'bewoelkung',\n",
       "  'datumDay'],\n",
       " ['datumDayofweek', 'datumWeek', 'feiertag', 'ferientag'])"
      ]
     },
     "execution_count": 108,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cont_nn, cat_nn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "datumDayofweek     7\n",
       "datumWeek         52\n",
       "feiertag           2\n",
       "ferientag          2\n",
       "dtype: int64"
      ]
     },
     "execution_count": 109,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_nn[cat_nn].nunique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_nn=df_nn.astype('float')\n",
    "procs_nn = [Categorify, FillMissing, Normalize]\n",
    "to_nn = TabularPandas(df_nn, procs_nn, cat_nn, cont_nn, splits=splits, y_names=dep_var)\n",
    "dls = to_nn.dataloaders()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn = tabular_learner(dls, n_out=1, loss_func=F.mse_loss) # removed all config params and just use default, model optimization is for another time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1816845.625000</td>\n",
       "      <td>1432133.000000</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>1808576.000000</td>\n",
       "      <td>1428485.875000</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>1781624.875000</td>\n",
       "      <td>1372125.625000</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>1593979.625000</td>\n",
       "      <td>813157.750000</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>1210289.750000</td>\n",
       "      <td>181425.625000</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>907380.125000</td>\n",
       "      <td>93933.218750</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>693443.375000</td>\n",
       "      <td>78018.203125</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>541596.625000</td>\n",
       "      <td>74059.382812</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>429904.500000</td>\n",
       "      <td>69380.671875</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>346007.937500</td>\n",
       "      <td>81721.476562</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>282091.156250</td>\n",
       "      <td>56799.742188</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>233567.953125</td>\n",
       "      <td>70023.062500</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>195410.078125</td>\n",
       "      <td>73250.179688</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>162619.500000</td>\n",
       "      <td>58097.007812</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>141866.890625</td>\n",
       "      <td>58359.632812</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>122567.882812</td>\n",
       "      <td>63118.117188</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>106427.398438</td>\n",
       "      <td>62738.894531</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>91498.328125</td>\n",
       "      <td>60427.527344</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>82339.585938</td>\n",
       "      <td>59645.546875</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>73133.851562</td>\n",
       "      <td>59144.730469</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "learn.fit_one_cycle(20, 1e-2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "243.196913"
      ]
     },
     "execution_count": 113,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preds,targs = learn.get_preds()\n",
    "r_mse(preds,targs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "183.31126\n",
      "0.27675143\n",
      "0.1825398\n"
     ]
    }
   ],
   "source": [
    "print(mae(preds, targs).numpy())\n",
    "print(mape(preds, targs).numpy())\n",
    "print(nmae(preds, targs).numpy())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Also better than previous model: 253 vs 287"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Check top errors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_err_nn = df_full.iloc[valid_idx]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [],
   "source": [
    "errors = pd.DataFrame(torch.sub(preds, targs).abs().numpy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {},
   "outputs": [],
   "source": [
    "pd.options.mode.chained_assignment = None\n",
    "df_err_nn['error'] = errors.to_numpy()\n",
    "df_err_nn['pred'] = preds.numpy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['uhrzeit_start', 'uhrzeit_ende', 'datumWeek', 'zaehlstelle', 'gesamt',\n",
       "       'min-temp', 'max-temp', 'niederschlag', 'bewoelkung', 'sonnenstunden',\n",
       "       'date', 'datumYear', 'datumMonth', 'datumDay', 'datumDayofweek',\n",
       "       'datumDayofyear', 'datumIs_month_end', 'datumIs_month_start',\n",
       "       'datumIs_quarter_end', 'datumIs_quarter_start', 'datumIs_year_end',\n",
       "       'datumIs_year_start', 'datumElapsed', 'feiertag', 'ferientag', 'error',\n",
       "       'pred'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 118,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_err_nn.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>date</th>\n",
       "      <th>datumDayofweek</th>\n",
       "      <th>min-temp</th>\n",
       "      <th>feiertag</th>\n",
       "      <th>ferientag</th>\n",
       "      <th>error</th>\n",
       "      <th>gesamt</th>\n",
       "      <th>pred</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>2019-05-23</td>\n",
       "      <td>3</td>\n",
       "      <td>7.6</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>804.707520</td>\n",
       "      <td>1239</td>\n",
       "      <td>2043.707520</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>2019-05-24</td>\n",
       "      <td>4</td>\n",
       "      <td>6.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>772.359619</td>\n",
       "      <td>1194</td>\n",
       "      <td>1966.359619</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>2019-05-21</td>\n",
       "      <td>1</td>\n",
       "      <td>10.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>718.092285</td>\n",
       "      <td>317</td>\n",
       "      <td>1035.092285</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>2019-05-22</td>\n",
       "      <td>2</td>\n",
       "      <td>8.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>677.635742</td>\n",
       "      <td>669</td>\n",
       "      <td>1346.635742</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>2019-05-31</td>\n",
       "      <td>4</td>\n",
       "      <td>11.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>667.846680</td>\n",
       "      <td>751</td>\n",
       "      <td>1418.846680</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>2019-08-23</td>\n",
       "      <td>4</td>\n",
       "      <td>11.4</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>661.843994</td>\n",
       "      <td>1778</td>\n",
       "      <td>1116.156006</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>2019-05-28</td>\n",
       "      <td>1</td>\n",
       "      <td>9.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>659.985962</td>\n",
       "      <td>396</td>\n",
       "      <td>1055.985962</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>2019-05-29</td>\n",
       "      <td>2</td>\n",
       "      <td>9.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>633.642822</td>\n",
       "      <td>722</td>\n",
       "      <td>1355.642822</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>2019-08-26</td>\n",
       "      <td>0</td>\n",
       "      <td>16.1</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>602.505859</td>\n",
       "      <td>1673</td>\n",
       "      <td>1070.494141</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>2019-08-22</td>\n",
       "      <td>3</td>\n",
       "      <td>13.0</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>595.103638</td>\n",
       "      <td>1956</td>\n",
       "      <td>1360.896362</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>2019-05-25</td>\n",
       "      <td>5</td>\n",
       "      <td>10.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>576.875488</td>\n",
       "      <td>506</td>\n",
       "      <td>1082.875488</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>2019-05-27</td>\n",
       "      <td>0</td>\n",
       "      <td>11.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>558.678589</td>\n",
       "      <td>817</td>\n",
       "      <td>1375.678589</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>2019-12-16</td>\n",
       "      <td>0</td>\n",
       "      <td>2.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>523.304504</td>\n",
       "      <td>1455</td>\n",
       "      <td>931.695496</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2019-02-04</td>\n",
       "      <td>0</td>\n",
       "      <td>-4.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>521.534241</td>\n",
       "      <td>254</td>\n",
       "      <td>775.534241</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>2019-10-22</td>\n",
       "      <td>1</td>\n",
       "      <td>10.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>513.877930</td>\n",
       "      <td>1873</td>\n",
       "      <td>1359.122070</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>2019-09-06</td>\n",
       "      <td>4</td>\n",
       "      <td>10.6</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>513.791687</td>\n",
       "      <td>1416</td>\n",
       "      <td>902.208313</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2019-08-01</td>\n",
       "      <td>3</td>\n",
       "      <td>13.4</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>500.369385</td>\n",
       "      <td>2500</td>\n",
       "      <td>1999.630615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>2019-12-18</td>\n",
       "      <td>2</td>\n",
       "      <td>6.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>485.576172</td>\n",
       "      <td>1587</td>\n",
       "      <td>1101.423828</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>2019-05-26</td>\n",
       "      <td>6</td>\n",
       "      <td>12.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>481.312012</td>\n",
       "      <td>518</td>\n",
       "      <td>999.312012</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2019-12-05</td>\n",
       "      <td>3</td>\n",
       "      <td>-4.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>464.159485</td>\n",
       "      <td>1381</td>\n",
       "      <td>916.840515</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>2019-12-17</td>\n",
       "      <td>1</td>\n",
       "      <td>2.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>455.572998</td>\n",
       "      <td>1535</td>\n",
       "      <td>1079.427002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>2019-02-26</td>\n",
       "      <td>1</td>\n",
       "      <td>-0.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>451.941589</td>\n",
       "      <td>1378</td>\n",
       "      <td>926.058411</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>2019-01-23</td>\n",
       "      <td>2</td>\n",
       "      <td>-5.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>436.599884</td>\n",
       "      <td>743</td>\n",
       "      <td>306.400116</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>2019-10-09</td>\n",
       "      <td>2</td>\n",
       "      <td>9.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>434.552124</td>\n",
       "      <td>769</td>\n",
       "      <td>1203.552124</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>2019-12-20</td>\n",
       "      <td>4</td>\n",
       "      <td>3.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>433.591248</td>\n",
       "      <td>1356</td>\n",
       "      <td>922.408752</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>2019-10-24</td>\n",
       "      <td>3</td>\n",
       "      <td>7.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>426.324219</td>\n",
       "      <td>1883</td>\n",
       "      <td>1456.675781</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>2019-11-20</td>\n",
       "      <td>2</td>\n",
       "      <td>1.4</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>416.713806</td>\n",
       "      <td>1109</td>\n",
       "      <td>692.286194</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2019-08-05</td>\n",
       "      <td>0</td>\n",
       "      <td>15.5</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>416.055420</td>\n",
       "      <td>2269</td>\n",
       "      <td>1852.944580</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>2019-12-19</td>\n",
       "      <td>3</td>\n",
       "      <td>2.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>408.103760</td>\n",
       "      <td>1446</td>\n",
       "      <td>1037.896240</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>2019-01-21</td>\n",
       "      <td>0</td>\n",
       "      <td>-3.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>401.251068</td>\n",
       "      <td>695</td>\n",
       "      <td>293.748932</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>2019-08-13</td>\n",
       "      <td>1</td>\n",
       "      <td>13.2</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>399.941040</td>\n",
       "      <td>1674</td>\n",
       "      <td>1274.058960</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2019-08-02</td>\n",
       "      <td>4</td>\n",
       "      <td>15.8</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>398.625977</td>\n",
       "      <td>2166</td>\n",
       "      <td>1767.374023</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>2019-11-25</td>\n",
       "      <td>0</td>\n",
       "      <td>2.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>395.585571</td>\n",
       "      <td>1380</td>\n",
       "      <td>984.414429</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>2019-09-22</td>\n",
       "      <td>6</td>\n",
       "      <td>7.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>390.864929</td>\n",
       "      <td>1146</td>\n",
       "      <td>755.135071</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>2019-04-12</td>\n",
       "      <td>4</td>\n",
       "      <td>2.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>380.251526</td>\n",
       "      <td>1256</td>\n",
       "      <td>875.748474</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2019-02-05</td>\n",
       "      <td>1</td>\n",
       "      <td>-3.9</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>378.745483</td>\n",
       "      <td>439</td>\n",
       "      <td>817.745483</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>2019-01-10</td>\n",
       "      <td>3</td>\n",
       "      <td>-3.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>371.131653</td>\n",
       "      <td>175</td>\n",
       "      <td>546.131653</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>2019-05-17</td>\n",
       "      <td>4</td>\n",
       "      <td>5.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>369.991089</td>\n",
       "      <td>1299</td>\n",
       "      <td>1668.991089</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>2019-10-21</td>\n",
       "      <td>0</td>\n",
       "      <td>11.9</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>367.876831</td>\n",
       "      <td>1929</td>\n",
       "      <td>1561.123169</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>2019-02-19</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>367.789368</td>\n",
       "      <td>1277</td>\n",
       "      <td>909.210632</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>2019-12-09</td>\n",
       "      <td>0</td>\n",
       "      <td>4.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>365.835266</td>\n",
       "      <td>1357</td>\n",
       "      <td>991.164734</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>2019-08-14</td>\n",
       "      <td>2</td>\n",
       "      <td>12.3</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>362.771851</td>\n",
       "      <td>2081</td>\n",
       "      <td>1718.228149</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>2019-12-11</td>\n",
       "      <td>2</td>\n",
       "      <td>-3.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>362.269043</td>\n",
       "      <td>1259</td>\n",
       "      <td>896.730957</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2019-04-02</td>\n",
       "      <td>1</td>\n",
       "      <td>3.9</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>354.438477</td>\n",
       "      <td>1769</td>\n",
       "      <td>1414.561523</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>2019-04-30</td>\n",
       "      <td>1</td>\n",
       "      <td>5.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>354.277588</td>\n",
       "      <td>945</td>\n",
       "      <td>1299.277588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>2019-05-09</td>\n",
       "      <td>3</td>\n",
       "      <td>7.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>352.727295</td>\n",
       "      <td>1347</td>\n",
       "      <td>1699.727295</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>2019-02-21</td>\n",
       "      <td>3</td>\n",
       "      <td>4.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>349.998596</td>\n",
       "      <td>1215</td>\n",
       "      <td>865.001404</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>2019-09-27</td>\n",
       "      <td>4</td>\n",
       "      <td>13.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>348.713379</td>\n",
       "      <td>1950</td>\n",
       "      <td>1601.286621</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>2019-09-20</td>\n",
       "      <td>4</td>\n",
       "      <td>3.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>348.644043</td>\n",
       "      <td>2011</td>\n",
       "      <td>1662.355957</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>2019-02-20</td>\n",
       "      <td>2</td>\n",
       "      <td>5.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>348.166199</td>\n",
       "      <td>1158</td>\n",
       "      <td>809.833801</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>2019-01-18</td>\n",
       "      <td>4</td>\n",
       "      <td>-2.9</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>341.173370</td>\n",
       "      <td>640</td>\n",
       "      <td>298.826630</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>2019-08-08</td>\n",
       "      <td>3</td>\n",
       "      <td>15.0</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>338.103882</td>\n",
       "      <td>2354</td>\n",
       "      <td>2015.896118</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>2019-08-21</td>\n",
       "      <td>2</td>\n",
       "      <td>12.7</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>324.406372</td>\n",
       "      <td>1593</td>\n",
       "      <td>1268.593628</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>2019-12-23</td>\n",
       "      <td>0</td>\n",
       "      <td>4.7</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>318.223663</td>\n",
       "      <td>575</td>\n",
       "      <td>256.776337</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>2019-10-25</td>\n",
       "      <td>4</td>\n",
       "      <td>8.9</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>317.963745</td>\n",
       "      <td>1758</td>\n",
       "      <td>1440.036255</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2019-10-02</td>\n",
       "      <td>2</td>\n",
       "      <td>5.9</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>315.360291</td>\n",
       "      <td>1181</td>\n",
       "      <td>865.639709</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>2019-10-19</td>\n",
       "      <td>5</td>\n",
       "      <td>8.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>315.249329</td>\n",
       "      <td>1130</td>\n",
       "      <td>814.750671</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>2019-01-22</td>\n",
       "      <td>1</td>\n",
       "      <td>-5.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>313.962555</td>\n",
       "      <td>720</td>\n",
       "      <td>406.037445</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>2019-04-21</td>\n",
       "      <td>6</td>\n",
       "      <td>4.6</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>312.186951</td>\n",
       "      <td>609</td>\n",
       "      <td>921.186951</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>2019-05-30</td>\n",
       "      <td>3</td>\n",
       "      <td>8.6</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>310.304993</td>\n",
       "      <td>593</td>\n",
       "      <td>903.304993</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>2019-01-24</td>\n",
       "      <td>3</td>\n",
       "      <td>-4.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>305.848419</td>\n",
       "      <td>764</td>\n",
       "      <td>458.151581</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2019-09-04</td>\n",
       "      <td>2</td>\n",
       "      <td>9.5</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>305.419800</td>\n",
       "      <td>2250</td>\n",
       "      <td>1944.580200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>2019-05-18</td>\n",
       "      <td>5</td>\n",
       "      <td>7.6</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>303.474365</td>\n",
       "      <td>772</td>\n",
       "      <td>1075.474365</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>2019-11-11</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>300.608154</td>\n",
       "      <td>1315</td>\n",
       "      <td>1014.391846</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>2019-11-26</td>\n",
       "      <td>1</td>\n",
       "      <td>3.9</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>299.197510</td>\n",
       "      <td>1446</td>\n",
       "      <td>1146.802490</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>2019-09-23</td>\n",
       "      <td>0</td>\n",
       "      <td>11.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>296.485107</td>\n",
       "      <td>969</td>\n",
       "      <td>1265.485107</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>2019-10-23</td>\n",
       "      <td>2</td>\n",
       "      <td>9.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>288.545166</td>\n",
       "      <td>1842</td>\n",
       "      <td>1553.454834</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>2019-05-14</td>\n",
       "      <td>1</td>\n",
       "      <td>3.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>287.237671</td>\n",
       "      <td>1335</td>\n",
       "      <td>1622.237671</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>2019-10-06</td>\n",
       "      <td>6</td>\n",
       "      <td>4.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>284.892365</td>\n",
       "      <td>561</td>\n",
       "      <td>276.107635</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>2019-02-27</td>\n",
       "      <td>2</td>\n",
       "      <td>2.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>284.727905</td>\n",
       "      <td>1351</td>\n",
       "      <td>1066.272095</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>2019-08-25</td>\n",
       "      <td>6</td>\n",
       "      <td>17.2</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>278.676208</td>\n",
       "      <td>1242</td>\n",
       "      <td>963.323792</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>2019-02-28</td>\n",
       "      <td>3</td>\n",
       "      <td>3.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>277.897705</td>\n",
       "      <td>1348</td>\n",
       "      <td>1070.102295</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2019-04-03</td>\n",
       "      <td>2</td>\n",
       "      <td>4.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>277.606201</td>\n",
       "      <td>1710</td>\n",
       "      <td>1432.393799</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2019-05-04</td>\n",
       "      <td>5</td>\n",
       "      <td>2.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>276.855408</td>\n",
       "      <td>374</td>\n",
       "      <td>650.855408</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>2019-02-08</td>\n",
       "      <td>4</td>\n",
       "      <td>1.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>274.541687</td>\n",
       "      <td>629</td>\n",
       "      <td>903.541687</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>2019-11-07</td>\n",
       "      <td>3</td>\n",
       "      <td>2.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>264.190430</td>\n",
       "      <td>1685</td>\n",
       "      <td>1420.809570</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>2019-01-11</td>\n",
       "      <td>4</td>\n",
       "      <td>-4.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>261.943756</td>\n",
       "      <td>140</td>\n",
       "      <td>401.943756</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>2019-01-25</td>\n",
       "      <td>4</td>\n",
       "      <td>-4.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>261.701752</td>\n",
       "      <td>520</td>\n",
       "      <td>258.298248</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>2019-12-10</td>\n",
       "      <td>1</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>261.452393</td>\n",
       "      <td>1289</td>\n",
       "      <td>1027.547607</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2019-12-01</td>\n",
       "      <td>6</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>261.203308</td>\n",
       "      <td>454</td>\n",
       "      <td>192.796692</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>2019-10-31</td>\n",
       "      <td>3</td>\n",
       "      <td>2.9</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>259.146973</td>\n",
       "      <td>1191</td>\n",
       "      <td>931.853027</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>2019-02-24</td>\n",
       "      <td>6</td>\n",
       "      <td>-3.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>256.283875</td>\n",
       "      <td>419</td>\n",
       "      <td>162.716141</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>2019-03-11</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>249.736877</td>\n",
       "      <td>600</td>\n",
       "      <td>849.736877</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2019-08-04</td>\n",
       "      <td>6</td>\n",
       "      <td>13.9</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>249.472412</td>\n",
       "      <td>1417</td>\n",
       "      <td>1167.527588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2019-10-01</td>\n",
       "      <td>1</td>\n",
       "      <td>9.6</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>246.936279</td>\n",
       "      <td>2245</td>\n",
       "      <td>1998.063721</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>2019-08-17</td>\n",
       "      <td>5</td>\n",
       "      <td>12.7</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>245.251282</td>\n",
       "      <td>1119</td>\n",
       "      <td>873.748718</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>2019-05-06</td>\n",
       "      <td>0</td>\n",
       "      <td>2.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>243.893066</td>\n",
       "      <td>1344</td>\n",
       "      <td>1100.106934</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>2019-12-30</td>\n",
       "      <td>0</td>\n",
       "      <td>-4.4</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>242.541046</td>\n",
       "      <td>602</td>\n",
       "      <td>359.458954</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>2019-10-30</td>\n",
       "      <td>2</td>\n",
       "      <td>3.6</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>241.603333</td>\n",
       "      <td>1159</td>\n",
       "      <td>917.396667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>2019-09-30</td>\n",
       "      <td>0</td>\n",
       "      <td>9.9</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>239.038452</td>\n",
       "      <td>1859</td>\n",
       "      <td>1619.961548</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2019-12-02</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>235.655701</td>\n",
       "      <td>1079</td>\n",
       "      <td>843.344299</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>2019-02-25</td>\n",
       "      <td>0</td>\n",
       "      <td>-1.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>232.563660</td>\n",
       "      <td>1214</td>\n",
       "      <td>981.436340</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2019-10-04</td>\n",
       "      <td>4</td>\n",
       "      <td>4.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>225.565002</td>\n",
       "      <td>1239</td>\n",
       "      <td>1013.434998</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>2019-08-28</td>\n",
       "      <td>2</td>\n",
       "      <td>16.9</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>222.806396</td>\n",
       "      <td>2139</td>\n",
       "      <td>1916.193604</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>2019-12-31</td>\n",
       "      <td>1</td>\n",
       "      <td>-0.5</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>219.739685</td>\n",
       "      <td>363</td>\n",
       "      <td>582.739685</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>2019-02-06</td>\n",
       "      <td>2</td>\n",
       "      <td>-6.6</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>218.759094</td>\n",
       "      <td>485</td>\n",
       "      <td>703.759094</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>2019-08-24</td>\n",
       "      <td>5</td>\n",
       "      <td>15.0</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>218.622192</td>\n",
       "      <td>1383</td>\n",
       "      <td>1164.377808</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2019-05-03</td>\n",
       "      <td>4</td>\n",
       "      <td>7.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>218.193359</td>\n",
       "      <td>738</td>\n",
       "      <td>956.193359</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2019-09-03</td>\n",
       "      <td>1</td>\n",
       "      <td>9.2</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>217.476074</td>\n",
       "      <td>2106</td>\n",
       "      <td>1888.523926</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>2019-09-11</td>\n",
       "      <td>2</td>\n",
       "      <td>10.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>217.060547</td>\n",
       "      <td>2058</td>\n",
       "      <td>1840.939453</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>2019-11-27</td>\n",
       "      <td>2</td>\n",
       "      <td>3.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>214.155151</td>\n",
       "      <td>1320</td>\n",
       "      <td>1105.844849</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>2019-03-15</td>\n",
       "      <td>4</td>\n",
       "      <td>4.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>213.256653</td>\n",
       "      <td>525</td>\n",
       "      <td>738.256653</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2019-02-03</td>\n",
       "      <td>6</td>\n",
       "      <td>-1.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>210.613312</td>\n",
       "      <td>36</td>\n",
       "      <td>246.613312</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>2019-01-16</td>\n",
       "      <td>2</td>\n",
       "      <td>-0.9</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>204.612000</td>\n",
       "      <td>729</td>\n",
       "      <td>933.612000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>2019-01-14</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>203.578705</td>\n",
       "      <td>552</td>\n",
       "      <td>348.421295</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>2019-09-07</td>\n",
       "      <td>5</td>\n",
       "      <td>9.4</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>196.182343</td>\n",
       "      <td>664</td>\n",
       "      <td>467.817657</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>2019-01-12</td>\n",
       "      <td>5</td>\n",
       "      <td>-1.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>194.013931</td>\n",
       "      <td>52</td>\n",
       "      <td>246.013931</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>2019-11-22</td>\n",
       "      <td>4</td>\n",
       "      <td>1.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>193.986572</td>\n",
       "      <td>1267</td>\n",
       "      <td>1073.013428</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2019-01-05</td>\n",
       "      <td>5</td>\n",
       "      <td>-0.9</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>193.613312</td>\n",
       "      <td>53</td>\n",
       "      <td>246.613312</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>2019-08-30</td>\n",
       "      <td>4</td>\n",
       "      <td>15.4</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>193.514648</td>\n",
       "      <td>2020</td>\n",
       "      <td>1826.485352</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>2019-08-06</td>\n",
       "      <td>1</td>\n",
       "      <td>19.1</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>192.877930</td>\n",
       "      <td>2269</td>\n",
       "      <td>2076.122070</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2019-12-03</td>\n",
       "      <td>1</td>\n",
       "      <td>-1.9</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>190.582886</td>\n",
       "      <td>1310</td>\n",
       "      <td>1119.417114</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>2019-09-12</td>\n",
       "      <td>3</td>\n",
       "      <td>12.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>190.212402</td>\n",
       "      <td>2219</td>\n",
       "      <td>2028.787598</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>2019-05-10</td>\n",
       "      <td>4</td>\n",
       "      <td>8.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>187.910889</td>\n",
       "      <td>1208</td>\n",
       "      <td>1395.910889</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>2019-03-16</td>\n",
       "      <td>5</td>\n",
       "      <td>9.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>183.107544</td>\n",
       "      <td>464</td>\n",
       "      <td>647.107544</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>2019-02-17</td>\n",
       "      <td>6</td>\n",
       "      <td>-1.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>182.188629</td>\n",
       "      <td>575</td>\n",
       "      <td>392.811371</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2019-03-03</td>\n",
       "      <td>6</td>\n",
       "      <td>7.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>181.486786</td>\n",
       "      <td>492</td>\n",
       "      <td>310.513214</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>2019-08-29</td>\n",
       "      <td>3</td>\n",
       "      <td>18.1</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>180.938965</td>\n",
       "      <td>1819</td>\n",
       "      <td>1638.061035</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2019-05-02</td>\n",
       "      <td>3</td>\n",
       "      <td>7.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>178.118896</td>\n",
       "      <td>1680</td>\n",
       "      <td>1858.118896</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>2019-10-08</td>\n",
       "      <td>1</td>\n",
       "      <td>7.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>177.955566</td>\n",
       "      <td>1568</td>\n",
       "      <td>1390.044434</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>2019-10-20</td>\n",
       "      <td>6</td>\n",
       "      <td>7.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>177.444763</td>\n",
       "      <td>1128</td>\n",
       "      <td>950.555237</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2019-12-04</td>\n",
       "      <td>2</td>\n",
       "      <td>-2.9</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>176.891602</td>\n",
       "      <td>1292</td>\n",
       "      <td>1115.108398</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>2019-11-12</td>\n",
       "      <td>1</td>\n",
       "      <td>2.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>174.079224</td>\n",
       "      <td>1271</td>\n",
       "      <td>1096.920776</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>2019-04-15</td>\n",
       "      <td>0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>169.128418</td>\n",
       "      <td>1450</td>\n",
       "      <td>1280.871582</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>2019-11-06</td>\n",
       "      <td>2</td>\n",
       "      <td>3.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>168.071777</td>\n",
       "      <td>1467</td>\n",
       "      <td>1298.928223</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>2019-02-07</td>\n",
       "      <td>3</td>\n",
       "      <td>-5.6</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>168.007751</td>\n",
       "      <td>491</td>\n",
       "      <td>659.007751</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>2019-09-26</td>\n",
       "      <td>3</td>\n",
       "      <td>12.6</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>166.968750</td>\n",
       "      <td>1826</td>\n",
       "      <td>1659.031250</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>2019-08-10</td>\n",
       "      <td>5</td>\n",
       "      <td>18.4</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>165.719910</td>\n",
       "      <td>1015</td>\n",
       "      <td>849.280090</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>2019-03-13</td>\n",
       "      <td>2</td>\n",
       "      <td>3.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>164.639282</td>\n",
       "      <td>913</td>\n",
       "      <td>1077.639282</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>2019-08-18</td>\n",
       "      <td>6</td>\n",
       "      <td>18.2</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>163.334961</td>\n",
       "      <td>1226</td>\n",
       "      <td>1062.665039</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>2019-08-19</td>\n",
       "      <td>0</td>\n",
       "      <td>16.1</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>161.171875</td>\n",
       "      <td>1205</td>\n",
       "      <td>1043.828125</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>2019-04-11</td>\n",
       "      <td>3</td>\n",
       "      <td>4.6</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>160.302490</td>\n",
       "      <td>1344</td>\n",
       "      <td>1183.697510</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>2019-11-18</td>\n",
       "      <td>0</td>\n",
       "      <td>2.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>159.606812</td>\n",
       "      <td>1198</td>\n",
       "      <td>1357.606812</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>2019-03-21</td>\n",
       "      <td>3</td>\n",
       "      <td>-1.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>158.035034</td>\n",
       "      <td>1359</td>\n",
       "      <td>1200.964966</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>2019-12-22</td>\n",
       "      <td>6</td>\n",
       "      <td>1.6</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>156.883575</td>\n",
       "      <td>329</td>\n",
       "      <td>172.116425</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>2019-04-20</td>\n",
       "      <td>5</td>\n",
       "      <td>5.9</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>155.979614</td>\n",
       "      <td>874</td>\n",
       "      <td>1029.979614</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>2019-04-09</td>\n",
       "      <td>1</td>\n",
       "      <td>8.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>153.306763</td>\n",
       "      <td>1418</td>\n",
       "      <td>1571.306763</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>2019-08-12</td>\n",
       "      <td>0</td>\n",
       "      <td>13.9</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>152.066223</td>\n",
       "      <td>903</td>\n",
       "      <td>750.933777</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>2019-04-29</td>\n",
       "      <td>0</td>\n",
       "      <td>4.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>151.984192</td>\n",
       "      <td>1013</td>\n",
       "      <td>861.015808</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>2019-03-22</td>\n",
       "      <td>4</td>\n",
       "      <td>0.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>150.120728</td>\n",
       "      <td>1316</td>\n",
       "      <td>1165.879272</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>2019-11-13</td>\n",
       "      <td>2</td>\n",
       "      <td>1.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>149.854492</td>\n",
       "      <td>985</td>\n",
       "      <td>1134.854492</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>2019-04-18</td>\n",
       "      <td>3</td>\n",
       "      <td>4.9</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>147.262939</td>\n",
       "      <td>1606</td>\n",
       "      <td>1753.262939</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>2019-04-06</td>\n",
       "      <td>5</td>\n",
       "      <td>5.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>145.920593</td>\n",
       "      <td>758</td>\n",
       "      <td>903.920593</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>2019-01-13</td>\n",
       "      <td>6</td>\n",
       "      <td>1.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>145.613312</td>\n",
       "      <td>101</td>\n",
       "      <td>246.613312</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>2019-05-13</td>\n",
       "      <td>0</td>\n",
       "      <td>5.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>145.111816</td>\n",
       "      <td>1238</td>\n",
       "      <td>1383.111816</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>2019-05-07</td>\n",
       "      <td>1</td>\n",
       "      <td>2.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>144.268066</td>\n",
       "      <td>1605</td>\n",
       "      <td>1749.268066</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>2019-11-29</td>\n",
       "      <td>4</td>\n",
       "      <td>2.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>142.913391</td>\n",
       "      <td>1070</td>\n",
       "      <td>927.086609</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>2019-03-06</td>\n",
       "      <td>2</td>\n",
       "      <td>2.9</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>140.327148</td>\n",
       "      <td>1133</td>\n",
       "      <td>992.672852</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>2019-10-07</td>\n",
       "      <td>0</td>\n",
       "      <td>6.6</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>139.905151</td>\n",
       "      <td>938</td>\n",
       "      <td>1077.905151</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>2019-09-16</td>\n",
       "      <td>0</td>\n",
       "      <td>14.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>139.746826</td>\n",
       "      <td>2199</td>\n",
       "      <td>2059.253174</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>2019-01-06</td>\n",
       "      <td>6</td>\n",
       "      <td>0.7</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>139.613312</td>\n",
       "      <td>107</td>\n",
       "      <td>246.613312</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>2019-11-30</td>\n",
       "      <td>5</td>\n",
       "      <td>-0.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>139.198456</td>\n",
       "      <td>595</td>\n",
       "      <td>455.801544</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>2019-03-26</td>\n",
       "      <td>1</td>\n",
       "      <td>1.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>138.921631</td>\n",
       "      <td>1223</td>\n",
       "      <td>1084.078369</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>2019-10-18</td>\n",
       "      <td>4</td>\n",
       "      <td>10.9</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>136.208252</td>\n",
       "      <td>1761</td>\n",
       "      <td>1624.791748</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>2019-12-25</td>\n",
       "      <td>2</td>\n",
       "      <td>3.1</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>134.646194</td>\n",
       "      <td>113</td>\n",
       "      <td>247.646194</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>2019-10-11</td>\n",
       "      <td>4</td>\n",
       "      <td>7.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>134.053223</td>\n",
       "      <td>1609</td>\n",
       "      <td>1743.053223</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>2019-11-15</td>\n",
       "      <td>4</td>\n",
       "      <td>-0.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>133.566528</td>\n",
       "      <td>1243</td>\n",
       "      <td>1109.433472</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>2019-04-24</td>\n",
       "      <td>2</td>\n",
       "      <td>10.5</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>130.541992</td>\n",
       "      <td>1800</td>\n",
       "      <td>1930.541992</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>2019-09-19</td>\n",
       "      <td>3</td>\n",
       "      <td>5.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>130.264893</td>\n",
       "      <td>1992</td>\n",
       "      <td>1861.735107</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>2019-03-24</td>\n",
       "      <td>6</td>\n",
       "      <td>3.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>130.108887</td>\n",
       "      <td>777</td>\n",
       "      <td>646.891113</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>2019-04-23</td>\n",
       "      <td>1</td>\n",
       "      <td>9.8</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>130.026367</td>\n",
       "      <td>1533</td>\n",
       "      <td>1663.026367</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2019-03-02</td>\n",
       "      <td>5</td>\n",
       "      <td>5.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>128.778625</td>\n",
       "      <td>385</td>\n",
       "      <td>256.221375</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>2019-12-28</td>\n",
       "      <td>5</td>\n",
       "      <td>-2.1</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>127.517609</td>\n",
       "      <td>374</td>\n",
       "      <td>246.482391</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>2019-02-23</td>\n",
       "      <td>5</td>\n",
       "      <td>-1.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>126.085388</td>\n",
       "      <td>490</td>\n",
       "      <td>363.914612</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>2019-11-10</td>\n",
       "      <td>6</td>\n",
       "      <td>-1.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>126.051849</td>\n",
       "      <td>465</td>\n",
       "      <td>338.948151</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>2019-11-28</td>\n",
       "      <td>3</td>\n",
       "      <td>7.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>125.962891</td>\n",
       "      <td>1299</td>\n",
       "      <td>1173.037109</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2019-01-01</td>\n",
       "      <td>1</td>\n",
       "      <td>2.7</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>123.752411</td>\n",
       "      <td>152</td>\n",
       "      <td>275.752411</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>2019-04-22</td>\n",
       "      <td>0</td>\n",
       "      <td>6.5</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>121.895142</td>\n",
       "      <td>843</td>\n",
       "      <td>964.895142</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>2019-03-17</td>\n",
       "      <td>6</td>\n",
       "      <td>3.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>118.038879</td>\n",
       "      <td>606</td>\n",
       "      <td>487.961121</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2019-11-02</td>\n",
       "      <td>5</td>\n",
       "      <td>8.9</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>117.984558</td>\n",
       "      <td>565</td>\n",
       "      <td>682.984558</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>2019-03-20</td>\n",
       "      <td>2</td>\n",
       "      <td>-1.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>117.447266</td>\n",
       "      <td>1298</td>\n",
       "      <td>1180.552734</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>2019-05-11</td>\n",
       "      <td>5</td>\n",
       "      <td>7.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>116.265320</td>\n",
       "      <td>385</td>\n",
       "      <td>501.265320</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>2019-12-07</td>\n",
       "      <td>5</td>\n",
       "      <td>6.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>116.097443</td>\n",
       "      <td>597</td>\n",
       "      <td>480.902557</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>2019-11-21</td>\n",
       "      <td>3</td>\n",
       "      <td>2.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>115.544922</td>\n",
       "      <td>1309</td>\n",
       "      <td>1193.455078</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>2019-11-08</td>\n",
       "      <td>4</td>\n",
       "      <td>3.6</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>115.079102</td>\n",
       "      <td>1180</td>\n",
       "      <td>1064.920898</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>2019-09-09</td>\n",
       "      <td>0</td>\n",
       "      <td>7.4</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>113.844482</td>\n",
       "      <td>1142</td>\n",
       "      <td>1028.155518</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>2019-09-24</td>\n",
       "      <td>1</td>\n",
       "      <td>10.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>111.355835</td>\n",
       "      <td>2003</td>\n",
       "      <td>1891.644165</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>2019-11-14</td>\n",
       "      <td>3</td>\n",
       "      <td>-0.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>111.272583</td>\n",
       "      <td>1435</td>\n",
       "      <td>1323.727417</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>2019-12-08</td>\n",
       "      <td>6</td>\n",
       "      <td>3.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>111.094208</td>\n",
       "      <td>524</td>\n",
       "      <td>412.905792</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>2019-12-24</td>\n",
       "      <td>1</td>\n",
       "      <td>3.7</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>110.969086</td>\n",
       "      <td>217</td>\n",
       "      <td>327.969086</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2019-01-02</td>\n",
       "      <td>2</td>\n",
       "      <td>-2.1</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>108.814240</td>\n",
       "      <td>391</td>\n",
       "      <td>282.185760</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>2019-03-23</td>\n",
       "      <td>5</td>\n",
       "      <td>2.6</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>108.711670</td>\n",
       "      <td>892</td>\n",
       "      <td>783.288330</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>2019-03-14</td>\n",
       "      <td>3</td>\n",
       "      <td>3.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>107.325012</td>\n",
       "      <td>758</td>\n",
       "      <td>865.325012</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>2019-03-19</td>\n",
       "      <td>1</td>\n",
       "      <td>0.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>104.738770</td>\n",
       "      <td>1142</td>\n",
       "      <td>1037.261230</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2019-02-01</td>\n",
       "      <td>4</td>\n",
       "      <td>-0.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>103.580994</td>\n",
       "      <td>762</td>\n",
       "      <td>658.419006</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>2019-01-29</td>\n",
       "      <td>1</td>\n",
       "      <td>-2.9</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>101.913879</td>\n",
       "      <td>716</td>\n",
       "      <td>817.913879</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>2019-09-10</td>\n",
       "      <td>1</td>\n",
       "      <td>5.7</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>97.481689</td>\n",
       "      <td>1927</td>\n",
       "      <td>1829.518311</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>2019-08-07</td>\n",
       "      <td>2</td>\n",
       "      <td>15.9</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>97.327026</td>\n",
       "      <td>1661</td>\n",
       "      <td>1563.672974</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>2019-08-31</td>\n",
       "      <td>5</td>\n",
       "      <td>15.8</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>97.288330</td>\n",
       "      <td>1309</td>\n",
       "      <td>1211.711670</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>2019-12-27</td>\n",
       "      <td>4</td>\n",
       "      <td>1.7</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>96.820129</td>\n",
       "      <td>342</td>\n",
       "      <td>245.179871</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>2019-04-28</td>\n",
       "      <td>6</td>\n",
       "      <td>3.9</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>96.133209</td>\n",
       "      <td>561</td>\n",
       "      <td>464.866791</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>2019-05-08</td>\n",
       "      <td>2</td>\n",
       "      <td>5.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>95.232422</td>\n",
       "      <td>1415</td>\n",
       "      <td>1319.767578</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>2019-04-07</td>\n",
       "      <td>6</td>\n",
       "      <td>5.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>95.030212</td>\n",
       "      <td>850</td>\n",
       "      <td>754.969788</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>2019-05-12</td>\n",
       "      <td>6</td>\n",
       "      <td>4.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>94.287079</td>\n",
       "      <td>160</td>\n",
       "      <td>254.287079</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>2019-12-14</td>\n",
       "      <td>5</td>\n",
       "      <td>1.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>91.483734</td>\n",
       "      <td>510</td>\n",
       "      <td>418.516266</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>2019-12-13</td>\n",
       "      <td>4</td>\n",
       "      <td>-1.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>89.696289</td>\n",
       "      <td>966</td>\n",
       "      <td>876.303711</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2019-01-03</td>\n",
       "      <td>3</td>\n",
       "      <td>-3.1</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>87.014374</td>\n",
       "      <td>377</td>\n",
       "      <td>289.985626</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>2019-04-13</td>\n",
       "      <td>5</td>\n",
       "      <td>2.6</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>86.661682</td>\n",
       "      <td>611</td>\n",
       "      <td>697.661682</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>2019-01-17</td>\n",
       "      <td>3</td>\n",
       "      <td>1.9</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>85.613586</td>\n",
       "      <td>795</td>\n",
       "      <td>709.386414</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2019-11-04</td>\n",
       "      <td>0</td>\n",
       "      <td>8.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>85.179932</td>\n",
       "      <td>1437</td>\n",
       "      <td>1351.820068</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2019-09-05</td>\n",
       "      <td>3</td>\n",
       "      <td>11.0</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>84.595459</td>\n",
       "      <td>1218</td>\n",
       "      <td>1133.404541</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>2019-01-07</td>\n",
       "      <td>0</td>\n",
       "      <td>0.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>82.346680</td>\n",
       "      <td>514</td>\n",
       "      <td>596.346680</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>2019-02-18</td>\n",
       "      <td>0</td>\n",
       "      <td>-1.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>82.087891</td>\n",
       "      <td>1154</td>\n",
       "      <td>1071.912109</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2019-03-05</td>\n",
       "      <td>1</td>\n",
       "      <td>3.8</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>80.126282</td>\n",
       "      <td>786</td>\n",
       "      <td>866.126282</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>2019-09-25</td>\n",
       "      <td>2</td>\n",
       "      <td>12.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>79.264648</td>\n",
       "      <td>1868</td>\n",
       "      <td>1788.735352</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>2019-05-16</td>\n",
       "      <td>3</td>\n",
       "      <td>4.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>78.715088</td>\n",
       "      <td>1232</td>\n",
       "      <td>1153.284912</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>2019-08-09</td>\n",
       "      <td>4</td>\n",
       "      <td>13.9</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>78.079590</td>\n",
       "      <td>2245</td>\n",
       "      <td>2166.920410</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>2019-08-20</td>\n",
       "      <td>1</td>\n",
       "      <td>12.7</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>76.604919</td>\n",
       "      <td>1015</td>\n",
       "      <td>938.395081</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2019-03-01</td>\n",
       "      <td>4</td>\n",
       "      <td>6.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>74.754944</td>\n",
       "      <td>638</td>\n",
       "      <td>712.754944</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>2019-04-16</td>\n",
       "      <td>1</td>\n",
       "      <td>2.5</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>73.001465</td>\n",
       "      <td>1679</td>\n",
       "      <td>1605.998535</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>2019-10-12</td>\n",
       "      <td>5</td>\n",
       "      <td>6.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>72.161499</td>\n",
       "      <td>1103</td>\n",
       "      <td>1175.161499</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2019-11-01</td>\n",
       "      <td>4</td>\n",
       "      <td>2.9</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>71.367645</td>\n",
       "      <td>386</td>\n",
       "      <td>457.367645</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>2019-12-06</td>\n",
       "      <td>4</td>\n",
       "      <td>-3.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>70.608032</td>\n",
       "      <td>1124</td>\n",
       "      <td>1053.391968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>2019-01-09</td>\n",
       "      <td>2</td>\n",
       "      <td>-1.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>70.584320</td>\n",
       "      <td>417</td>\n",
       "      <td>346.415680</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2019-08-03</td>\n",
       "      <td>5</td>\n",
       "      <td>14.3</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>69.559570</td>\n",
       "      <td>1021</td>\n",
       "      <td>951.440430</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>2019-12-12</td>\n",
       "      <td>3</td>\n",
       "      <td>-0.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>68.913635</td>\n",
       "      <td>1028</td>\n",
       "      <td>959.086365</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2019-05-05</td>\n",
       "      <td>6</td>\n",
       "      <td>2.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>68.455902</td>\n",
       "      <td>302</td>\n",
       "      <td>370.455902</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>2019-08-11</td>\n",
       "      <td>6</td>\n",
       "      <td>16.3</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>68.452881</td>\n",
       "      <td>1267</td>\n",
       "      <td>1198.547119</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>2019-10-27</td>\n",
       "      <td>6</td>\n",
       "      <td>10.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>66.609314</td>\n",
       "      <td>1000</td>\n",
       "      <td>933.390686</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>2019-03-25</td>\n",
       "      <td>0</td>\n",
       "      <td>1.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>65.596924</td>\n",
       "      <td>1152</td>\n",
       "      <td>1086.403076</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>2019-03-10</td>\n",
       "      <td>6</td>\n",
       "      <td>4.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>64.220215</td>\n",
       "      <td>204</td>\n",
       "      <td>268.220215</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>2019-03-29</td>\n",
       "      <td>4</td>\n",
       "      <td>4.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>63.532471</td>\n",
       "      <td>1387</td>\n",
       "      <td>1450.532471</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>2019-01-08</td>\n",
       "      <td>1</td>\n",
       "      <td>0.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>62.151672</td>\n",
       "      <td>581</td>\n",
       "      <td>643.151672</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2019-04-01</td>\n",
       "      <td>0</td>\n",
       "      <td>5.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>61.733521</td>\n",
       "      <td>1592</td>\n",
       "      <td>1530.266479</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>2019-01-28</td>\n",
       "      <td>0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>60.091980</td>\n",
       "      <td>806</td>\n",
       "      <td>745.908020</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>2019-04-08</td>\n",
       "      <td>0</td>\n",
       "      <td>8.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>59.745605</td>\n",
       "      <td>1191</td>\n",
       "      <td>1131.254395</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>2019-01-26</td>\n",
       "      <td>5</td>\n",
       "      <td>-3.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>59.561295</td>\n",
       "      <td>187</td>\n",
       "      <td>246.561295</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>2019-02-16</td>\n",
       "      <td>5</td>\n",
       "      <td>-2.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>58.482239</td>\n",
       "      <td>580</td>\n",
       "      <td>521.517761</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>2019-08-15</td>\n",
       "      <td>3</td>\n",
       "      <td>9.8</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>56.065430</td>\n",
       "      <td>769</td>\n",
       "      <td>712.934570</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>2019-05-19</td>\n",
       "      <td>6</td>\n",
       "      <td>8.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>55.986633</td>\n",
       "      <td>710</td>\n",
       "      <td>765.986633</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>2019-10-28</td>\n",
       "      <td>0</td>\n",
       "      <td>7.2</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>53.928284</td>\n",
       "      <td>902</td>\n",
       "      <td>955.928284</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>2019-01-20</td>\n",
       "      <td>6</td>\n",
       "      <td>-8.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>53.613312</td>\n",
       "      <td>193</td>\n",
       "      <td>246.613312</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>2019-04-17</td>\n",
       "      <td>2</td>\n",
       "      <td>7.0</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>53.359863</td>\n",
       "      <td>1711</td>\n",
       "      <td>1764.359863</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>2019-04-19</td>\n",
       "      <td>4</td>\n",
       "      <td>5.5</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>52.639465</td>\n",
       "      <td>785</td>\n",
       "      <td>837.639465</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>2019-08-27</td>\n",
       "      <td>1</td>\n",
       "      <td>16.3</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>52.039795</td>\n",
       "      <td>2056</td>\n",
       "      <td>2108.039795</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>2019-01-31</td>\n",
       "      <td>3</td>\n",
       "      <td>-3.6</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>51.232483</td>\n",
       "      <td>869</td>\n",
       "      <td>920.232483</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>2019-09-08</td>\n",
       "      <td>6</td>\n",
       "      <td>8.2</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>50.740677</td>\n",
       "      <td>297</td>\n",
       "      <td>246.259323</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>2019-10-13</td>\n",
       "      <td>6</td>\n",
       "      <td>8.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>50.707275</td>\n",
       "      <td>1036</td>\n",
       "      <td>1086.707275</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>2019-02-09</td>\n",
       "      <td>5</td>\n",
       "      <td>0.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>50.562988</td>\n",
       "      <td>277</td>\n",
       "      <td>327.562988</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>2019-11-24</td>\n",
       "      <td>6</td>\n",
       "      <td>3.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>49.645752</td>\n",
       "      <td>589</td>\n",
       "      <td>539.354248</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>2019-03-08</td>\n",
       "      <td>4</td>\n",
       "      <td>5.6</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>49.469482</td>\n",
       "      <td>971</td>\n",
       "      <td>921.530518</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>2019-09-17</td>\n",
       "      <td>1</td>\n",
       "      <td>11.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>49.377808</td>\n",
       "      <td>1952</td>\n",
       "      <td>1902.622192</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>2019-08-16</td>\n",
       "      <td>4</td>\n",
       "      <td>14.1</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>49.321289</td>\n",
       "      <td>1686</td>\n",
       "      <td>1636.678711</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>2019-09-13</td>\n",
       "      <td>4</td>\n",
       "      <td>11.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>48.790771</td>\n",
       "      <td>1952</td>\n",
       "      <td>2000.790771</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>2019-03-09</td>\n",
       "      <td>5</td>\n",
       "      <td>6.1</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>48.526581</td>\n",
       "      <td>310</td>\n",
       "      <td>261.473419</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>2019-10-29</td>\n",
       "      <td>1</td>\n",
       "      <td>6.0</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>47.526001</td>\n",
       "      <td>905</td>\n",
       "      <td>952.526001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>2019-04-25</td>\n",
       "      <td>3</td>\n",
       "      <td>9.8</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>47.136597</td>\n",
       "      <td>2005</td>\n",
       "      <td>1957.863403</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>2019-02-10</td>\n",
       "      <td>6</td>\n",
       "      <td>4.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>45.981781</td>\n",
       "      <td>295</td>\n",
       "      <td>249.018219</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>2019-04-26</td>\n",
       "      <td>4</td>\n",
       "      <td>6.3</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>45.411011</td>\n",
       "      <td>1095</td>\n",
       "      <td>1049.588989</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>2019-11-09</td>\n",
       "      <td>5</td>\n",
       "      <td>-0.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>42.777100</td>\n",
       "      <td>570</td>\n",
       "      <td>612.777100</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>2019-11-23</td>\n",
       "      <td>5</td>\n",
       "      <td>-0.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>42.701843</td>\n",
       "      <td>678</td>\n",
       "      <td>720.701843</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>2019-12-26</td>\n",
       "      <td>3</td>\n",
       "      <td>1.0</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>40.504700</td>\n",
       "      <td>213</td>\n",
       "      <td>253.504700</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2019-05-01</td>\n",
       "      <td>2</td>\n",
       "      <td>4.7</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>40.158691</td>\n",
       "      <td>1061</td>\n",
       "      <td>1020.841309</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>2019-02-13</td>\n",
       "      <td>2</td>\n",
       "      <td>-0.9</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>40.042542</td>\n",
       "      <td>933</td>\n",
       "      <td>892.957458</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>2019-02-12</td>\n",
       "      <td>1</td>\n",
       "      <td>-0.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>39.520142</td>\n",
       "      <td>727</td>\n",
       "      <td>687.479858</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>2019-12-29</td>\n",
       "      <td>6</td>\n",
       "      <td>-4.8</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>38.469818</td>\n",
       "      <td>285</td>\n",
       "      <td>246.530182</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2019-01-04</td>\n",
       "      <td>4</td>\n",
       "      <td>-1.7</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>36.879257</td>\n",
       "      <td>282</td>\n",
       "      <td>245.120743</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>2019-10-15</td>\n",
       "      <td>1</td>\n",
       "      <td>8.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>35.393066</td>\n",
       "      <td>2081</td>\n",
       "      <td>2045.606934</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2019-10-03</td>\n",
       "      <td>3</td>\n",
       "      <td>5.3</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>35.135437</td>\n",
       "      <td>750</td>\n",
       "      <td>714.864563</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>2019-04-14</td>\n",
       "      <td>6</td>\n",
       "      <td>3.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>35.050415</td>\n",
       "      <td>445</td>\n",
       "      <td>480.050415</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>2019-09-14</td>\n",
       "      <td>5</td>\n",
       "      <td>13.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>33.967529</td>\n",
       "      <td>1244</td>\n",
       "      <td>1210.032471</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>2019-01-19</td>\n",
       "      <td>5</td>\n",
       "      <td>-6.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>33.049194</td>\n",
       "      <td>280</td>\n",
       "      <td>246.950806</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>2019-04-10</td>\n",
       "      <td>2</td>\n",
       "      <td>8.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>32.614502</td>\n",
       "      <td>1238</td>\n",
       "      <td>1205.385498</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2019-10-05</td>\n",
       "      <td>5</td>\n",
       "      <td>5.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>32.433746</td>\n",
       "      <td>514</td>\n",
       "      <td>481.566254</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>2019-01-27</td>\n",
       "      <td>6</td>\n",
       "      <td>2.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>32.141785</td>\n",
       "      <td>215</td>\n",
       "      <td>247.141785</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2019-09-02</td>\n",
       "      <td>0</td>\n",
       "      <td>11.5</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>31.592041</td>\n",
       "      <td>1055</td>\n",
       "      <td>1086.592041</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>2019-11-16</td>\n",
       "      <td>5</td>\n",
       "      <td>2.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>31.354309</td>\n",
       "      <td>565</td>\n",
       "      <td>533.645691</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2019-04-05</td>\n",
       "      <td>4</td>\n",
       "      <td>3.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>31.036133</td>\n",
       "      <td>1138</td>\n",
       "      <td>1169.036133</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>2019-02-22</td>\n",
       "      <td>4</td>\n",
       "      <td>2.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>31.035339</td>\n",
       "      <td>678</td>\n",
       "      <td>709.035339</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>2019-03-31</td>\n",
       "      <td>6</td>\n",
       "      <td>3.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>29.763733</td>\n",
       "      <td>844</td>\n",
       "      <td>814.236267</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>2019-03-07</td>\n",
       "      <td>3</td>\n",
       "      <td>7.2</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>28.981567</td>\n",
       "      <td>961</td>\n",
       "      <td>989.981567</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>2019-01-30</td>\n",
       "      <td>2</td>\n",
       "      <td>-4.6</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>28.535583</td>\n",
       "      <td>755</td>\n",
       "      <td>726.464417</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2019-11-05</td>\n",
       "      <td>1</td>\n",
       "      <td>3.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>28.152100</td>\n",
       "      <td>1254</td>\n",
       "      <td>1282.152100</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>2019-10-10</td>\n",
       "      <td>3</td>\n",
       "      <td>8.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>28.104248</td>\n",
       "      <td>1454</td>\n",
       "      <td>1425.895752</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>2019-10-16</td>\n",
       "      <td>2</td>\n",
       "      <td>7.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>27.221436</td>\n",
       "      <td>1872</td>\n",
       "      <td>1844.778564</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>2019-02-11</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>27.177917</td>\n",
       "      <td>527</td>\n",
       "      <td>554.177917</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2019-02-02</td>\n",
       "      <td>5</td>\n",
       "      <td>0.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>25.602127</td>\n",
       "      <td>256</td>\n",
       "      <td>230.397873</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>2019-09-15</td>\n",
       "      <td>6</td>\n",
       "      <td>11.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>25.380859</td>\n",
       "      <td>1216</td>\n",
       "      <td>1190.619141</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>2019-12-21</td>\n",
       "      <td>5</td>\n",
       "      <td>2.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>24.606598</td>\n",
       "      <td>338</td>\n",
       "      <td>313.393402</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>2019-10-26</td>\n",
       "      <td>5</td>\n",
       "      <td>6.6</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>24.167480</td>\n",
       "      <td>1064</td>\n",
       "      <td>1088.167480</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>2019-12-15</td>\n",
       "      <td>6</td>\n",
       "      <td>6.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>23.127289</td>\n",
       "      <td>454</td>\n",
       "      <td>430.872711</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2019-11-03</td>\n",
       "      <td>6</td>\n",
       "      <td>7.9</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>22.729584</td>\n",
       "      <td>431</td>\n",
       "      <td>408.270416</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>2019-03-27</td>\n",
       "      <td>2</td>\n",
       "      <td>1.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>22.685364</td>\n",
       "      <td>918</td>\n",
       "      <td>895.314636</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>2019-09-18</td>\n",
       "      <td>2</td>\n",
       "      <td>6.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>22.614868</td>\n",
       "      <td>1944</td>\n",
       "      <td>1921.385132</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>2019-09-21</td>\n",
       "      <td>5</td>\n",
       "      <td>4.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>22.402344</td>\n",
       "      <td>1165</td>\n",
       "      <td>1187.402344</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>2019-03-28</td>\n",
       "      <td>3</td>\n",
       "      <td>4.9</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>22.384521</td>\n",
       "      <td>1348</td>\n",
       "      <td>1325.615479</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>2019-10-17</td>\n",
       "      <td>3</td>\n",
       "      <td>6.1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>18.623779</td>\n",
       "      <td>1952</td>\n",
       "      <td>1933.376221</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2019-03-04</td>\n",
       "      <td>0</td>\n",
       "      <td>4.5</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>18.393860</td>\n",
       "      <td>993</td>\n",
       "      <td>974.606140</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>2019-10-14</td>\n",
       "      <td>0</td>\n",
       "      <td>9.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>14.355713</td>\n",
       "      <td>1938</td>\n",
       "      <td>1952.355713</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>2019-04-27</td>\n",
       "      <td>5</td>\n",
       "      <td>5.6</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>14.051147</td>\n",
       "      <td>545</td>\n",
       "      <td>559.051147</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>2019-03-12</td>\n",
       "      <td>1</td>\n",
       "      <td>1.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>13.635010</td>\n",
       "      <td>1067</td>\n",
       "      <td>1080.635010</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>2019-03-30</td>\n",
       "      <td>5</td>\n",
       "      <td>2.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>13.035339</td>\n",
       "      <td>920</td>\n",
       "      <td>933.035339</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>2019-11-17</td>\n",
       "      <td>6</td>\n",
       "      <td>3.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>12.923615</td>\n",
       "      <td>372</td>\n",
       "      <td>384.923615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>2019-05-20</td>\n",
       "      <td>0</td>\n",
       "      <td>11.3</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>11.411102</td>\n",
       "      <td>391</td>\n",
       "      <td>379.588898</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2019-09-01</td>\n",
       "      <td>6</td>\n",
       "      <td>17.7</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>11.360901</td>\n",
       "      <td>1001</td>\n",
       "      <td>989.639099</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>2019-09-28</td>\n",
       "      <td>5</td>\n",
       "      <td>12.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>11.143677</td>\n",
       "      <td>1032</td>\n",
       "      <td>1043.143677</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>2019-09-29</td>\n",
       "      <td>6</td>\n",
       "      <td>11.2</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>9.694946</td>\n",
       "      <td>1203</td>\n",
       "      <td>1212.694946</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>2019-01-15</td>\n",
       "      <td>1</td>\n",
       "      <td>-0.8</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>9.596497</td>\n",
       "      <td>517</td>\n",
       "      <td>526.596497</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>2019-02-15</td>\n",
       "      <td>4</td>\n",
       "      <td>-1.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>6.909363</td>\n",
       "      <td>992</td>\n",
       "      <td>985.090637</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2019-04-04</td>\n",
       "      <td>3</td>\n",
       "      <td>2.9</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>5.662598</td>\n",
       "      <td>1326</td>\n",
       "      <td>1320.337402</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>2019-05-15</td>\n",
       "      <td>2</td>\n",
       "      <td>4.4</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>4.881958</td>\n",
       "      <td>1059</td>\n",
       "      <td>1054.118042</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>2019-03-18</td>\n",
       "      <td>0</td>\n",
       "      <td>1.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>3.610535</td>\n",
       "      <td>928</td>\n",
       "      <td>924.389465</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>2019-02-14</td>\n",
       "      <td>3</td>\n",
       "      <td>-1.5</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>1.276062</td>\n",
       "      <td>945</td>\n",
       "      <td>946.276062</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>2019-11-19</td>\n",
       "      <td>1</td>\n",
       "      <td>0.7</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>0.659058</td>\n",
       "      <td>1243</td>\n",
       "      <td>1242.340942</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         date  datumDayofweek  min-temp  feiertag  ferientag       error  \\\n",
       "22 2019-05-23               3       7.6     False      False  804.707520   \n",
       "23 2019-05-24               4       6.3     False      False  772.359619   \n",
       "20 2019-05-21               1      10.5     False      False  718.092285   \n",
       "21 2019-05-22               2       8.8     False      False  677.635742   \n",
       "30 2019-05-31               4      11.7     False      False  667.846680   \n",
       "22 2019-08-23               4      11.4     False       True  661.843994   \n",
       "27 2019-05-28               1       9.5     False      False  659.985962   \n",
       "28 2019-05-29               2       9.4     False      False  633.642822   \n",
       "25 2019-08-26               0      16.1     False       True  602.505859   \n",
       "21 2019-08-22               3      13.0     False       True  595.103638   \n",
       "24 2019-05-25               5      10.5     False      False  576.875488   \n",
       "26 2019-05-27               0      11.2     False      False  558.678589   \n",
       "15 2019-12-16               0       2.3     False      False  523.304504   \n",
       "3  2019-02-04               0      -4.5     False      False  521.534241   \n",
       "21 2019-10-22               1      10.5     False      False  513.877930   \n",
       "5  2019-09-06               4      10.6     False       True  513.791687   \n",
       "0  2019-08-01               3      13.4     False       True  500.369385   \n",
       "17 2019-12-18               2       6.3     False      False  485.576172   \n",
       "25 2019-05-26               6      12.1     False      False  481.312012   \n",
       "4  2019-12-05               3      -4.8     False      False  464.159485   \n",
       "16 2019-12-17               1       2.1     False      False  455.572998   \n",
       "25 2019-02-26               1      -0.1     False      False  451.941589   \n",
       "22 2019-01-23               2      -5.0     False      False  436.599884   \n",
       "8  2019-10-09               2       9.1     False      False  434.552124   \n",
       "19 2019-12-20               4       3.2     False      False  433.591248   \n",
       "23 2019-10-24               3       7.4     False      False  426.324219   \n",
       "19 2019-11-20               2       1.4      True      False  416.713806   \n",
       "4  2019-08-05               0      15.5     False       True  416.055420   \n",
       "18 2019-12-19               3       2.8     False      False  408.103760   \n",
       "20 2019-01-21               0      -3.5     False      False  401.251068   \n",
       "12 2019-08-13               1      13.2     False       True  399.941040   \n",
       "1  2019-08-02               4      15.8     False       True  398.625977   \n",
       "24 2019-11-25               0       2.3     False      False  395.585571   \n",
       "21 2019-09-22               6       7.8     False      False  390.864929   \n",
       "11 2019-04-12               4       2.4     False      False  380.251526   \n",
       "4  2019-02-05               1      -3.9     False      False  378.745483   \n",
       "9  2019-01-10               3      -3.5     False      False  371.131653   \n",
       "16 2019-05-17               4       5.5     False      False  369.991089   \n",
       "20 2019-10-21               0      11.9     False      False  367.876831   \n",
       "18 2019-02-19               1       1.0     False      False  367.789368   \n",
       "8  2019-12-09               0       4.1     False      False  365.835266   \n",
       "13 2019-08-14               2      12.3     False       True  362.771851   \n",
       "10 2019-12-11               2      -3.8     False      False  362.269043   \n",
       "1  2019-04-02               1       3.9     False      False  354.438477   \n",
       "29 2019-04-30               1       5.8     False      False  354.277588   \n",
       "8  2019-05-09               3       7.1     False      False  352.727295   \n",
       "20 2019-02-21               3       4.1     False      False  349.998596   \n",
       "26 2019-09-27               4      13.7     False      False  348.713379   \n",
       "19 2019-09-20               4       3.4     False      False  348.644043   \n",
       "19 2019-02-20               2       5.0     False      False  348.166199   \n",
       "17 2019-01-18               4      -2.9     False      False  341.173370   \n",
       "7  2019-08-08               3      15.0     False       True  338.103882   \n",
       "20 2019-08-21               2      12.7     False       True  324.406372   \n",
       "22 2019-12-23               0       4.7     False       True  318.223663   \n",
       "24 2019-10-25               4       8.9     False      False  317.963745   \n",
       "1  2019-10-02               2       5.9     False      False  315.360291   \n",
       "18 2019-10-19               5       8.5     False      False  315.249329   \n",
       "21 2019-01-22               1      -5.1     False      False  313.962555   \n",
       "20 2019-04-21               6       4.6     False       True  312.186951   \n",
       "29 2019-05-30               3       8.6      True      False  310.304993   \n",
       "23 2019-01-24               3      -4.4     False      False  305.848419   \n",
       "3  2019-09-04               2       9.5     False       True  305.419800   \n",
       "17 2019-05-18               5       7.6     False      False  303.474365   \n",
       "10 2019-11-11               0      -0.7     False      False  300.608154   \n",
       "25 2019-11-26               1       3.9     False      False  299.197510   \n",
       "22 2019-09-23               0      11.4     False      False  296.485107   \n",
       "22 2019-10-23               2       9.0     False      False  288.545166   \n",
       "13 2019-05-14               1       3.2     False      False  287.237671   \n",
       "5  2019-10-06               6       4.2     False      False  284.892365   \n",
       "26 2019-02-27               2       2.1     False      False  284.727905   \n",
       "24 2019-08-25               6      17.2     False       True  278.676208   \n",
       "27 2019-02-28               3       3.3     False      False  277.897705   \n",
       "2  2019-04-03               2       4.8     False      False  277.606201   \n",
       "3  2019-05-04               5       2.5     False      False  276.855408   \n",
       "7  2019-02-08               4       1.7     False      False  274.541687   \n",
       "6  2019-11-07               3       2.3     False      False  264.190430   \n",
       "10 2019-01-11               4      -4.1     False      False  261.943756   \n",
       "24 2019-01-25               4      -4.7     False      False  261.701752   \n",
       "9  2019-12-10               1      -1.0     False      False  261.452393   \n",
       "0  2019-12-01               6      -1.0     False      False  261.203308   \n",
       "30 2019-10-31               3       2.9     False       True  259.146973   \n",
       "23 2019-02-24               6      -3.2     False      False  256.283875   \n",
       "10 2019-03-11               0      -0.3     False      False  249.736877   \n",
       "3  2019-08-04               6      13.9     False       True  249.472412   \n",
       "0  2019-10-01               1       9.6     False      False  246.936279   \n",
       "16 2019-08-17               5      12.7     False       True  245.251282   \n",
       "5  2019-05-06               0       2.8     False      False  243.893066   \n",
       "29 2019-12-30               0      -4.4     False       True  242.541046   \n",
       "29 2019-10-30               2       3.6     False       True  241.603333   \n",
       "29 2019-09-30               0       9.9     False      False  239.038452   \n",
       "1  2019-12-02               0       0.0     False      False  235.655701   \n",
       "24 2019-02-25               0      -1.7     False      False  232.563660   \n",
       "3  2019-10-04               4       4.5     False      False  225.565002   \n",
       "27 2019-08-28               2      16.9     False       True  222.806396   \n",
       "30 2019-12-31               1      -0.5     False       True  219.739685   \n",
       "5  2019-02-06               2      -6.6     False      False  218.759094   \n",
       "23 2019-08-24               5      15.0     False       True  218.622192   \n",
       "2  2019-05-03               4       7.3     False      False  218.193359   \n",
       "2  2019-09-03               1       9.2     False       True  217.476074   \n",
       "10 2019-09-11               2      10.0     False      False  217.060547   \n",
       "26 2019-11-27               2       3.0     False      False  214.155151   \n",
       "14 2019-03-15               4       4.0     False      False  213.256653   \n",
       "2  2019-02-03               6      -1.5     False      False  210.613312   \n",
       "15 2019-01-16               2      -0.9     False      False  204.612000   \n",
       "13 2019-01-14               0      -0.1     False      False  203.578705   \n",
       "6  2019-09-07               5       9.4     False       True  196.182343   \n",
       "11 2019-01-12               5      -1.1     False      False  194.013931   \n",
       "21 2019-11-22               4       1.4     False      False  193.986572   \n",
       "4  2019-01-05               5      -0.9     False       True  193.613312   \n",
       "29 2019-08-30               4      15.4     False       True  193.514648   \n",
       "5  2019-08-06               1      19.1     False       True  192.877930   \n",
       "2  2019-12-03               1      -1.9     False      False  190.582886   \n",
       "11 2019-09-12               3      12.7     False      False  190.212402   \n",
       "9  2019-05-10               4       8.3     False      False  187.910889   \n",
       "15 2019-03-16               5       9.4     False      False  183.107544   \n",
       "16 2019-02-17               6      -1.7     False      False  182.188629   \n",
       "2  2019-03-03               6       7.0     False      False  181.486786   \n",
       "28 2019-08-29               3      18.1     False       True  180.938965   \n",
       "1  2019-05-02               3       7.4     False      False  178.118896   \n",
       "7  2019-10-08               1       7.5     False      False  177.955566   \n",
       "19 2019-10-20               6       7.0     False      False  177.444763   \n",
       "3  2019-12-04               2      -2.9     False      False  176.891602   \n",
       "11 2019-11-12               1       2.0     False      False  174.079224   \n",
       "14 2019-04-15               0       5.0     False       True  169.128418   \n",
       "5  2019-11-06               2       3.7     False      False  168.071777   \n",
       "6  2019-02-07               3      -5.6     False      False  168.007751   \n",
       "25 2019-09-26               3      12.6     False      False  166.968750   \n",
       "9  2019-08-10               5      18.4     False       True  165.719910   \n",
       "12 2019-03-13               2       3.3     False      False  164.639282   \n",
       "17 2019-08-18               6      18.2     False       True  163.334961   \n",
       "18 2019-08-19               0      16.1     False       True  161.171875   \n",
       "10 2019-04-11               3       4.6     False      False  160.302490   \n",
       "17 2019-11-18               0       2.8     False      False  159.606812   \n",
       "20 2019-03-21               3      -1.2     False      False  158.035034   \n",
       "21 2019-12-22               6       1.6     False      False  156.883575   \n",
       "19 2019-04-20               5       5.9     False       True  155.979614   \n",
       "8  2019-04-09               1       8.5     False      False  153.306763   \n",
       "11 2019-08-12               0      13.9     False       True  152.066223   \n",
       "28 2019-04-29               0       4.8     False      False  151.984192   \n",
       "21 2019-03-22               4       0.3     False      False  150.120728   \n",
       "12 2019-11-13               2       1.3     False      False  149.854492   \n",
       "17 2019-04-18               3       4.9     False       True  147.262939   \n",
       "5  2019-04-06               5       5.2     False      False  145.920593   \n",
       "12 2019-01-13               6       1.4     False      False  145.613312   \n",
       "12 2019-05-13               0       5.5     False      False  145.111816   \n",
       "6  2019-05-07               1       2.2     False      False  144.268066   \n",
       "28 2019-11-29               4       2.3     False      False  142.913391   \n",
       "5  2019-03-06               2       2.9     False       True  140.327148   \n",
       "6  2019-10-07               0       6.6     False      False  139.905151   \n",
       "15 2019-09-16               0      14.3     False      False  139.746826   \n",
       "5  2019-01-06               6       0.7      True       True  139.613312   \n",
       "29 2019-11-30               5      -0.2     False      False  139.198456   \n",
       "25 2019-03-26               1       1.2     False      False  138.921631   \n",
       "17 2019-10-18               4      10.9     False      False  136.208252   \n",
       "24 2019-12-25               2       3.1      True       True  134.646194   \n",
       "10 2019-10-11               4       7.3     False      False  134.053223   \n",
       "14 2019-11-15               4      -0.2     False      False  133.566528   \n",
       "23 2019-04-24               2      10.5     False       True  130.541992   \n",
       "18 2019-09-19               3       5.4     False      False  130.264893   \n",
       "23 2019-03-24               6       3.3     False      False  130.108887   \n",
       "22 2019-04-23               1       9.8     False       True  130.026367   \n",
       "1  2019-03-02               5       5.2     False      False  128.778625   \n",
       "27 2019-12-28               5      -2.1     False       True  127.517609   \n",
       "22 2019-02-23               5      -1.7     False      False  126.085388   \n",
       "9  2019-11-10               6      -1.7     False      False  126.051849   \n",
       "27 2019-11-28               3       7.5     False      False  125.962891   \n",
       "0  2019-01-01               1       2.7      True       True  123.752411   \n",
       "21 2019-04-22               0       6.5      True       True  121.895142   \n",
       "16 2019-03-17               6       3.2     False      False  118.038879   \n",
       "1  2019-11-02               5       8.9     False      False  117.984558   \n",
       "19 2019-03-20               2      -1.4     False      False  117.447266   \n",
       "10 2019-05-11               5       7.4     False      False  116.265320   \n",
       "6  2019-12-07               5       6.4     False      False  116.097443   \n",
       "20 2019-11-21               3       2.7     False      False  115.544922   \n",
       "7  2019-11-08               4       3.6     False      False  115.079102   \n",
       "8  2019-09-09               0       7.4     False       True  113.844482   \n",
       "23 2019-09-24               1      10.3     False      False  111.355835   \n",
       "13 2019-11-14               3      -0.3     False      False  111.272583   \n",
       "7  2019-12-08               6       3.8     False      False  111.094208   \n",
       "23 2019-12-24               1       3.7     False       True  110.969086   \n",
       "1  2019-01-02               2      -2.1     False       True  108.814240   \n",
       "22 2019-03-23               5       2.6     False      False  108.711670   \n",
       "13 2019-03-14               3       3.0     False      False  107.325012   \n",
       "18 2019-03-19               1       0.8     False      False  104.738770   \n",
       "0  2019-02-01               4      -0.2     False      False  103.580994   \n",
       "28 2019-01-29               1      -2.9     False      False  101.913879   \n",
       "9  2019-09-10               1       5.7     False       True   97.481689   \n",
       "6  2019-08-07               2      15.9     False       True   97.327026   \n",
       "30 2019-08-31               5      15.8     False       True   97.288330   \n",
       "26 2019-12-27               4       1.7     False       True   96.820129   \n",
       "27 2019-04-28               6       3.9     False       True   96.133209   \n",
       "7  2019-05-08               2       5.8     False      False   95.232422   \n",
       "6  2019-04-07               6       5.0     False      False   95.030212   \n",
       "11 2019-05-12               6       4.5     False      False   94.287079   \n",
       "13 2019-12-14               5       1.8     False      False   91.483734   \n",
       "12 2019-12-13               4      -1.2     False      False   89.696289   \n",
       "2  2019-01-03               3      -3.1     False       True   87.014374   \n",
       "12 2019-04-13               5       2.6     False      False   86.661682   \n",
       "16 2019-01-17               3       1.9     False      False   85.613586   \n",
       "3  2019-11-04               0       8.7     False      False   85.179932   \n",
       "4  2019-09-05               3      11.0     False       True   84.595459   \n",
       "6  2019-01-07               0       0.7     False      False   82.346680   \n",
       "17 2019-02-18               0      -1.8     False      False   82.087891   \n",
       "4  2019-03-05               1       3.8     False       True   80.126282   \n",
       "24 2019-09-25               2      12.8     False      False   79.264648   \n",
       "15 2019-05-16               3       4.5     False      False   78.715088   \n",
       "8  2019-08-09               4      13.9     False       True   78.079590   \n",
       "19 2019-08-20               1      12.7     False       True   76.604919   \n",
       "0  2019-03-01               4       6.2     False      False   74.754944   \n",
       "15 2019-04-16               1       2.5     False       True   73.001465   \n",
       "11 2019-10-12               5       6.2     False      False   72.161499   \n",
       "0  2019-11-01               4       2.9      True       True   71.367645   \n",
       "5  2019-12-06               4      -3.8     False      False   70.608032   \n",
       "8  2019-01-09               2      -1.5     False      False   70.584320   \n",
       "2  2019-08-03               5      14.3     False       True   69.559570   \n",
       "11 2019-12-12               3      -0.2     False      False   68.913635   \n",
       "4  2019-05-05               6       2.1     False      False   68.455902   \n",
       "10 2019-08-11               6      16.3     False       True   68.452881   \n",
       "26 2019-10-27               6      10.1     False      False   66.609314   \n",
       "24 2019-03-25               0       1.8     False      False   65.596924   \n",
       "9  2019-03-10               6       4.4     False      False   64.220215   \n",
       "28 2019-03-29               4       4.0     False      False   63.532471   \n",
       "7  2019-01-08               1       0.8     False      False   62.151672   \n",
       "0  2019-04-01               0       5.3     False      False   61.733521   \n",
       "27 2019-01-28               0       0.2     False      False   60.091980   \n",
       "7  2019-04-08               0       8.2     False      False   59.745605   \n",
       "25 2019-01-26               5      -3.1     False      False   59.561295   \n",
       "15 2019-02-16               5      -2.1     False      False   58.482239   \n",
       "14 2019-08-15               3       9.8      True       True   56.065430   \n",
       "18 2019-05-19               6       8.5     False      False   55.986633   \n",
       "27 2019-10-28               0       7.2     False       True   53.928284   \n",
       "19 2019-01-20               6      -8.0     False      False   53.613312   \n",
       "16 2019-04-17               2       7.0     False       True   53.359863   \n",
       "18 2019-04-19               4       5.5      True       True   52.639465   \n",
       "26 2019-08-27               1      16.3     False       True   52.039795   \n",
       "30 2019-01-31               3      -3.6     False      False   51.232483   \n",
       "7  2019-09-08               6       8.2     False       True   50.740677   \n",
       "12 2019-10-13               6       8.8     False      False   50.707275   \n",
       "8  2019-02-09               5       0.8     False      False   50.562988   \n",
       "23 2019-11-24               6       3.5     False      False   49.645752   \n",
       "7  2019-03-08               4       5.6     False       True   49.469482   \n",
       "16 2019-09-17               1      11.1     False      False   49.377808   \n",
       "15 2019-08-16               4      14.1     False       True   49.321289   \n",
       "12 2019-09-13               4      11.5     False      False   48.790771   \n",
       "8  2019-03-09               5       6.1     False       True   48.526581   \n",
       "28 2019-10-29               1       6.0     False       True   47.526001   \n",
       "24 2019-04-25               3       9.8     False       True   47.136597   \n",
       "9  2019-02-10               6       4.1     False      False   45.981781   \n",
       "25 2019-04-26               4       6.3     False       True   45.411011   \n",
       "8  2019-11-09               5      -0.1     False      False   42.777100   \n",
       "22 2019-11-23               5      -0.3     False      False   42.701843   \n",
       "25 2019-12-26               3       1.0      True       True   40.504700   \n",
       "0  2019-05-01               2       4.7      True      False   40.158691   \n",
       "12 2019-02-13               2      -0.9     False      False   40.042542   \n",
       "11 2019-02-12               1      -0.1     False      False   39.520142   \n",
       "28 2019-12-29               6      -4.8     False       True   38.469818   \n",
       "3  2019-01-04               4      -1.7     False       True   36.879257   \n",
       "14 2019-10-15               1       8.8     False      False   35.393066   \n",
       "2  2019-10-03               3       5.3      True      False   35.135437   \n",
       "13 2019-04-14               6       3.2     False      False   35.050415   \n",
       "13 2019-09-14               5      13.5     False      False   33.967529   \n",
       "18 2019-01-19               5      -6.0     False      False   33.049194   \n",
       "9  2019-04-10               2       8.0     False      False   32.614502   \n",
       "4  2019-10-05               5       5.0     False      False   32.433746   \n",
       "26 2019-01-27               6       2.1     False      False   32.141785   \n",
       "1  2019-09-02               0      11.5     False       True   31.592041   \n",
       "15 2019-11-16               5       2.3     False      False   31.354309   \n",
       "4  2019-04-05               4       3.3     False      False   31.036133   \n",
       "21 2019-02-22               4       2.7     False      False   31.035339   \n",
       "30 2019-03-31               6       3.4     False      False   29.763733   \n",
       "6  2019-03-07               3       7.2     False       True   28.981567   \n",
       "29 2019-01-30               2      -4.6     False      False   28.535583   \n",
       "4  2019-11-05               1       3.5     False      False   28.152100   \n",
       "9  2019-10-10               3       8.5     False      False   28.104248   \n",
       "15 2019-10-16               2       7.8     False      False   27.221436   \n",
       "10 2019-02-11               0       0.0     False      False   27.177917   \n",
       "1  2019-02-02               5       0.7     False      False   25.602127   \n",
       "14 2019-09-15               6      11.4     False      False   25.380859   \n",
       "20 2019-12-21               5       2.5     False      False   24.606598   \n",
       "25 2019-10-26               5       6.6     False      False   24.167480   \n",
       "14 2019-12-15               6       6.1     False      False   23.127289   \n",
       "2  2019-11-03               6       7.9     False      False   22.729584   \n",
       "26 2019-03-27               2       1.7     False      False   22.685364   \n",
       "17 2019-09-18               2       6.7     False      False   22.614868   \n",
       "20 2019-09-21               5       4.8     False      False   22.402344   \n",
       "27 2019-03-28               3       4.9     False      False   22.384521   \n",
       "16 2019-10-17               3       6.1     False      False   18.623779   \n",
       "3  2019-03-04               0       4.5     False       True   18.393860   \n",
       "13 2019-10-14               0       9.2     False      False   14.355713   \n",
       "26 2019-04-27               5       5.6     False       True   14.051147   \n",
       "11 2019-03-12               1       1.4     False      False   13.635010   \n",
       "29 2019-03-30               5       2.7     False      False   13.035339   \n",
       "16 2019-11-17               6       3.4     False      False   12.923615   \n",
       "19 2019-05-20               0      11.3     False      False   11.411102   \n",
       "0  2019-09-01               6      17.7     False       True   11.360901   \n",
       "27 2019-09-28               5      12.4     False      False   11.143677   \n",
       "28 2019-09-29               6      11.2     False      False    9.694946   \n",
       "14 2019-01-15               1      -0.8     False      False    9.596497   \n",
       "14 2019-02-15               4      -1.4     False      False    6.909363   \n",
       "3  2019-04-04               3       2.9     False      False    5.662598   \n",
       "14 2019-05-15               2       4.4     False      False    4.881958   \n",
       "17 2019-03-18               0       1.7     False      False    3.610535   \n",
       "13 2019-02-14               3      -1.5     False      False    1.276062   \n",
       "18 2019-11-19               1       0.7     False      False    0.659058   \n",
       "\n",
       "    gesamt         pred  \n",
       "22    1239  2043.707520  \n",
       "23    1194  1966.359619  \n",
       "20     317  1035.092285  \n",
       "21     669  1346.635742  \n",
       "30     751  1418.846680  \n",
       "22    1778  1116.156006  \n",
       "27     396  1055.985962  \n",
       "28     722  1355.642822  \n",
       "25    1673  1070.494141  \n",
       "21    1956  1360.896362  \n",
       "24     506  1082.875488  \n",
       "26     817  1375.678589  \n",
       "15    1455   931.695496  \n",
       "3      254   775.534241  \n",
       "21    1873  1359.122070  \n",
       "5     1416   902.208313  \n",
       "0     2500  1999.630615  \n",
       "17    1587  1101.423828  \n",
       "25     518   999.312012  \n",
       "4     1381   916.840515  \n",
       "16    1535  1079.427002  \n",
       "25    1378   926.058411  \n",
       "22     743   306.400116  \n",
       "8      769  1203.552124  \n",
       "19    1356   922.408752  \n",
       "23    1883  1456.675781  \n",
       "19    1109   692.286194  \n",
       "4     2269  1852.944580  \n",
       "18    1446  1037.896240  \n",
       "20     695   293.748932  \n",
       "12    1674  1274.058960  \n",
       "1     2166  1767.374023  \n",
       "24    1380   984.414429  \n",
       "21    1146   755.135071  \n",
       "11    1256   875.748474  \n",
       "4      439   817.745483  \n",
       "9      175   546.131653  \n",
       "16    1299  1668.991089  \n",
       "20    1929  1561.123169  \n",
       "18    1277   909.210632  \n",
       "8     1357   991.164734  \n",
       "13    2081  1718.228149  \n",
       "10    1259   896.730957  \n",
       "1     1769  1414.561523  \n",
       "29     945  1299.277588  \n",
       "8     1347  1699.727295  \n",
       "20    1215   865.001404  \n",
       "26    1950  1601.286621  \n",
       "19    2011  1662.355957  \n",
       "19    1158   809.833801  \n",
       "17     640   298.826630  \n",
       "7     2354  2015.896118  \n",
       "20    1593  1268.593628  \n",
       "22     575   256.776337  \n",
       "24    1758  1440.036255  \n",
       "1     1181   865.639709  \n",
       "18    1130   814.750671  \n",
       "21     720   406.037445  \n",
       "20     609   921.186951  \n",
       "29     593   903.304993  \n",
       "23     764   458.151581  \n",
       "3     2250  1944.580200  \n",
       "17     772  1075.474365  \n",
       "10    1315  1014.391846  \n",
       "25    1446  1146.802490  \n",
       "22     969  1265.485107  \n",
       "22    1842  1553.454834  \n",
       "13    1335  1622.237671  \n",
       "5      561   276.107635  \n",
       "26    1351  1066.272095  \n",
       "24    1242   963.323792  \n",
       "27    1348  1070.102295  \n",
       "2     1710  1432.393799  \n",
       "3      374   650.855408  \n",
       "7      629   903.541687  \n",
       "6     1685  1420.809570  \n",
       "10     140   401.943756  \n",
       "24     520   258.298248  \n",
       "9     1289  1027.547607  \n",
       "0      454   192.796692  \n",
       "30    1191   931.853027  \n",
       "23     419   162.716141  \n",
       "10     600   849.736877  \n",
       "3     1417  1167.527588  \n",
       "0     2245  1998.063721  \n",
       "16    1119   873.748718  \n",
       "5     1344  1100.106934  \n",
       "29     602   359.458954  \n",
       "29    1159   917.396667  \n",
       "29    1859  1619.961548  \n",
       "1     1079   843.344299  \n",
       "24    1214   981.436340  \n",
       "3     1239  1013.434998  \n",
       "27    2139  1916.193604  \n",
       "30     363   582.739685  \n",
       "5      485   703.759094  \n",
       "23    1383  1164.377808  \n",
       "2      738   956.193359  \n",
       "2     2106  1888.523926  \n",
       "10    2058  1840.939453  \n",
       "26    1320  1105.844849  \n",
       "14     525   738.256653  \n",
       "2       36   246.613312  \n",
       "15     729   933.612000  \n",
       "13     552   348.421295  \n",
       "6      664   467.817657  \n",
       "11      52   246.013931  \n",
       "21    1267  1073.013428  \n",
       "4       53   246.613312  \n",
       "29    2020  1826.485352  \n",
       "5     2269  2076.122070  \n",
       "2     1310  1119.417114  \n",
       "11    2219  2028.787598  \n",
       "9     1208  1395.910889  \n",
       "15     464   647.107544  \n",
       "16     575   392.811371  \n",
       "2      492   310.513214  \n",
       "28    1819  1638.061035  \n",
       "1     1680  1858.118896  \n",
       "7     1568  1390.044434  \n",
       "19    1128   950.555237  \n",
       "3     1292  1115.108398  \n",
       "11    1271  1096.920776  \n",
       "14    1450  1280.871582  \n",
       "5     1467  1298.928223  \n",
       "6      491   659.007751  \n",
       "25    1826  1659.031250  \n",
       "9     1015   849.280090  \n",
       "12     913  1077.639282  \n",
       "17    1226  1062.665039  \n",
       "18    1205  1043.828125  \n",
       "10    1344  1183.697510  \n",
       "17    1198  1357.606812  \n",
       "20    1359  1200.964966  \n",
       "21     329   172.116425  \n",
       "19     874  1029.979614  \n",
       "8     1418  1571.306763  \n",
       "11     903   750.933777  \n",
       "28    1013   861.015808  \n",
       "21    1316  1165.879272  \n",
       "12     985  1134.854492  \n",
       "17    1606  1753.262939  \n",
       "5      758   903.920593  \n",
       "12     101   246.613312  \n",
       "12    1238  1383.111816  \n",
       "6     1605  1749.268066  \n",
       "28    1070   927.086609  \n",
       "5     1133   992.672852  \n",
       "6      938  1077.905151  \n",
       "15    2199  2059.253174  \n",
       "5      107   246.613312  \n",
       "29     595   455.801544  \n",
       "25    1223  1084.078369  \n",
       "17    1761  1624.791748  \n",
       "24     113   247.646194  \n",
       "10    1609  1743.053223  \n",
       "14    1243  1109.433472  \n",
       "23    1800  1930.541992  \n",
       "18    1992  1861.735107  \n",
       "23     777   646.891113  \n",
       "22    1533  1663.026367  \n",
       "1      385   256.221375  \n",
       "27     374   246.482391  \n",
       "22     490   363.914612  \n",
       "9      465   338.948151  \n",
       "27    1299  1173.037109  \n",
       "0      152   275.752411  \n",
       "21     843   964.895142  \n",
       "16     606   487.961121  \n",
       "1      565   682.984558  \n",
       "19    1298  1180.552734  \n",
       "10     385   501.265320  \n",
       "6      597   480.902557  \n",
       "20    1309  1193.455078  \n",
       "7     1180  1064.920898  \n",
       "8     1142  1028.155518  \n",
       "23    2003  1891.644165  \n",
       "13    1435  1323.727417  \n",
       "7      524   412.905792  \n",
       "23     217   327.969086  \n",
       "1      391   282.185760  \n",
       "22     892   783.288330  \n",
       "13     758   865.325012  \n",
       "18    1142  1037.261230  \n",
       "0      762   658.419006  \n",
       "28     716   817.913879  \n",
       "9     1927  1829.518311  \n",
       "6     1661  1563.672974  \n",
       "30    1309  1211.711670  \n",
       "26     342   245.179871  \n",
       "27     561   464.866791  \n",
       "7     1415  1319.767578  \n",
       "6      850   754.969788  \n",
       "11     160   254.287079  \n",
       "13     510   418.516266  \n",
       "12     966   876.303711  \n",
       "2      377   289.985626  \n",
       "12     611   697.661682  \n",
       "16     795   709.386414  \n",
       "3     1437  1351.820068  \n",
       "4     1218  1133.404541  \n",
       "6      514   596.346680  \n",
       "17    1154  1071.912109  \n",
       "4      786   866.126282  \n",
       "24    1868  1788.735352  \n",
       "15    1232  1153.284912  \n",
       "8     2245  2166.920410  \n",
       "19    1015   938.395081  \n",
       "0      638   712.754944  \n",
       "15    1679  1605.998535  \n",
       "11    1103  1175.161499  \n",
       "0      386   457.367645  \n",
       "5     1124  1053.391968  \n",
       "8      417   346.415680  \n",
       "2     1021   951.440430  \n",
       "11    1028   959.086365  \n",
       "4      302   370.455902  \n",
       "10    1267  1198.547119  \n",
       "26    1000   933.390686  \n",
       "24    1152  1086.403076  \n",
       "9      204   268.220215  \n",
       "28    1387  1450.532471  \n",
       "7      581   643.151672  \n",
       "0     1592  1530.266479  \n",
       "27     806   745.908020  \n",
       "7     1191  1131.254395  \n",
       "25     187   246.561295  \n",
       "15     580   521.517761  \n",
       "14     769   712.934570  \n",
       "18     710   765.986633  \n",
       "27     902   955.928284  \n",
       "19     193   246.613312  \n",
       "16    1711  1764.359863  \n",
       "18     785   837.639465  \n",
       "26    2056  2108.039795  \n",
       "30     869   920.232483  \n",
       "7      297   246.259323  \n",
       "12    1036  1086.707275  \n",
       "8      277   327.562988  \n",
       "23     589   539.354248  \n",
       "7      971   921.530518  \n",
       "16    1952  1902.622192  \n",
       "15    1686  1636.678711  \n",
       "12    1952  2000.790771  \n",
       "8      310   261.473419  \n",
       "28     905   952.526001  \n",
       "24    2005  1957.863403  \n",
       "9      295   249.018219  \n",
       "25    1095  1049.588989  \n",
       "8      570   612.777100  \n",
       "22     678   720.701843  \n",
       "25     213   253.504700  \n",
       "0     1061  1020.841309  \n",
       "12     933   892.957458  \n",
       "11     727   687.479858  \n",
       "28     285   246.530182  \n",
       "3      282   245.120743  \n",
       "14    2081  2045.606934  \n",
       "2      750   714.864563  \n",
       "13     445   480.050415  \n",
       "13    1244  1210.032471  \n",
       "18     280   246.950806  \n",
       "9     1238  1205.385498  \n",
       "4      514   481.566254  \n",
       "26     215   247.141785  \n",
       "1     1055  1086.592041  \n",
       "15     565   533.645691  \n",
       "4     1138  1169.036133  \n",
       "21     678   709.035339  \n",
       "30     844   814.236267  \n",
       "6      961   989.981567  \n",
       "29     755   726.464417  \n",
       "4     1254  1282.152100  \n",
       "9     1454  1425.895752  \n",
       "15    1872  1844.778564  \n",
       "10     527   554.177917  \n",
       "1      256   230.397873  \n",
       "14    1216  1190.619141  \n",
       "20     338   313.393402  \n",
       "25    1064  1088.167480  \n",
       "14     454   430.872711  \n",
       "2      431   408.270416  \n",
       "26     918   895.314636  \n",
       "17    1944  1921.385132  \n",
       "20    1165  1187.402344  \n",
       "27    1348  1325.615479  \n",
       "16    1952  1933.376221  \n",
       "3      993   974.606140  \n",
       "13    1938  1952.355713  \n",
       "26     545   559.051147  \n",
       "11    1067  1080.635010  \n",
       "29     920   933.035339  \n",
       "16     372   384.923615  \n",
       "19     391   379.588898  \n",
       "0     1001   989.639099  \n",
       "27    1032  1043.143677  \n",
       "28    1203  1212.694946  \n",
       "14     517   526.596497  \n",
       "14     992   985.090637  \n",
       "3     1326  1320.337402  \n",
       "14    1059  1054.118042  \n",
       "17     928   924.389465  \n",
       "13     945   946.276062  \n",
       "18    1243  1242.340942  "
      ]
     },
     "execution_count": 119,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.set_option('display.max_rows', 500)\n",
    "df_err_nn.sort_values(by = 'error', ascending=False)[['date', 'datumDayofweek', 'min-temp', 'feiertag', 'ferientag', 'error', 'gesamt', 'pred']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "metadata": {},
   "outputs": [],
   "source": [
    "pd.reset_option('display.max_rows')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1080 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "top_errors = df_err_nn.sort_values(by = 'error', ascending=False)[:20]\n",
    "plotResults(top_errors, top_errors)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Also the neuronal net has quite high errors at the end of may, similiear to the RandomForest."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5. Conclusion <a name=\"conclusion\"></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So what did I learn?\n",
    "\n",
    "* RandomForest\n",
    "    * good to use as a base model\n",
    "    * nice visualisations for understanding important features (which can also be helpful for a neuronal net approach)\n",
    "    * nice way to learn about the features and dataset\n",
    "    * performs arround MAPE 33% and performs not as good as the neuronal net\n",
    "    * Open Questions: Results change on every rerun. Therfore: How comparable are the results with the different changes then? Should I fix a seed for the RandomForest, is this possibl\n",
    "* NeuronalNet\n",
    "    * just used the default model of fastai (tabular learner)\n",
    "    * performs arround MAPE 28%\n",
    "    * in comparison to the RandomForest the reasons for a prediction are not traceable\n",
    "* General\n",
    "    * Great way to learn about biking habits in munich\n",
    "    * Base models are good to improve dataset (e.g. add holidays)\n",
    "    * I learned that biking is used for commuting (workday vs weekend) which is the most important feature. Temperature is also important but not as much as expected.\n",
    "    * How to I evaluate models? How useful are they? What metrics should I use? I would like to compare them with real world usecases to check if they are helpful.\n",
    "    * Predictions are possible with the shown ML approaches. I'm not sure if the would be helpful for traffic planers with the shown erros, but for a first try it looks promising.\n",
    "\n",
    "#### Possible next steps:\n",
    "\n",
    "* improve the neuronal net model by not using the default tabular learner and optimizing & thinking about an own architecture\n",
    "* add more features to the dataset (e.g. Brückentage, days before long weekends, other weather information, ...?)\n",
    "* train on a different bike counting station or sumup all counting stations into a total number for munich\n",
    "* Use datasets where weekdays and weekends are seperated. Could I train a \"Should I bike or take the bus?\"-Model on this?\n",
    "\n",
    "\n",
    "Do you have any notes, questions or found an error? Contact me: sven.adolph@pixel.de"
   ]
  }
 ],
 "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": "3.8.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}