Explainer notebook_V4.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img align=\"left\" width=\"220px\" height=\"220px\" src=\"https://astra.dk/sites/default/files/DTU%20logo.png\" alt=\"logo\"/>\n",
    "<div>\n",
    "    <h1 style=\"text-align: right\"> Social Data Analysis and Visualization</h1>\n",
    "    <h2 style=\"text-align: right\"> Final project</h2>\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# <span style=\"color:#990000\"> Final project - Social data analysis and vizualisations"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <span style=\"color:#008000\"> Libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# usual libraries\n",
    "import numpy as np \n",
    "import pandas as pd \n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "import seaborn as sns\n",
    "sns.set(style='darkgrid', palette='muted', color_codes=True)\n",
    "import json\n",
    "import matplotlib.dates as dates\n",
    "import datetime\n",
    "from math import log\n",
    "\n",
    "# import from bokeh library\n",
    "from bokeh.models import ColorBar, LogColorMapper, LogTicker, LinearColorMapper\n",
    "from bokeh.plotting import figure\n",
    "from bokeh.io import output_notebook, push_notebook, show, curdoc\n",
    "from bokeh.layouts import layout\n",
    "from bokeh.models import (Button, CategoricalColorMapper, ColumnDataSource,\n",
    "                          HoverTool, Label, SingleIntervalTicker, Slider, RadioButtonGroup, PreText, Legend)\n",
    "from bokeh.plotting import figure\n",
    "from bokeh.application import Application\n",
    "from bokeh.application.handlers import FunctionHandler\n",
    "from bokeh.layouts import row, widgetbox, layout\n",
    "from bokeh.models.renderers import GlyphRenderer\n",
    "from bokeh.transform import linear_cmap\n",
    "\n",
    "# import from sklearn library\n",
    "from sklearn.preprocessing import PolynomialFeatures\n",
    "from sklearn.linear_model import LinearRegression\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from statsmodels.graphics.tsaplots import plot_acf, plot_pacf\n",
    "from statsmodels.tsa.arima_model import ARIMA\n",
    "\n",
    "# Function to convert the date in pandas format datetime64 to format datetime\n",
    "def convert_datetime64_to_datetime( usert: np.datetime64 )->datetime.datetime:\n",
    "    t = np.datetime64( usert, 'us').astype(datetime.datetime)\n",
    "    return(t)\n",
    "\n",
    "# Filters unwanted warnings\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# <span style=\"color:#008000\">  If you want to run and try this notebook\n",
    "    \n",
    "**You need to put your notebook url in the following variable.**\n",
    "**It must have the following shape:** \"http://localhost:XXXX\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Change the value of this variable\n",
    "notebook_url = \"http://localhost:8888\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# <span style=\"color:#008000\"> Table of content:\n",
    "* [1. Motivation](#first-part)\n",
    "* [2. Basic stats](#second-part)\n",
    "* [3. Data Analysis](#third-part)\n",
    "* [4. Genre](#fourth-part)\n",
    "* [5. Visualizations](#fifth-part)\n",
    "* [6. Discussion](#sixth-part)\n",
    "* [7. Contributions](#seventh-part)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# <span style=\"color:#990000\"> I - Motivation <a class=\"anchor\" id=\"first-part\">"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### **Objective**\n",
    "\n",
    "Since the beginning of the lockdown in France, hardly a day goes by without the media spreading anything and everything about Covid-19. Thus it has become tough to distinguish between true and false information. \n",
    "At the same time, the evolution of the outbreak has been more and more compared on a regional scale (the French departments) in spite of this notion of department tended to vanish in the last years.\n",
    "By visualizing the Covid-19 data on the French departments scale, this study aims to assess correctly the situation and to the state whether this regional grid is useful and legitimate.\n",
    "\n",
    "Moreover, many business activities were disrupted due to the lockdown. Thus, this study will also assess the impacts of the outbreak on road and air traffic, air quality and energy consumption. The evolution of these various indicators will be visualized before and during the period of lockdown."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### **What are the datasets?**\n",
    "\n",
    "Many datasets were used to achieve the goal:\n",
    "* The [hospital dataset](https://www.data.gouv.fr/fr/datasets/donnees-hospitalieres-relatives-a-lepidemie-de-covid-19/## ) \n",
    "with Covid hospital data by department, provided by the French government.\n",
    "\n",
    "* The [energy dataset](https://www.data.gouv.fr/en/datasets/donnees-eco2mix-regionales-temps-reel-1/) with energy consumption data by region, also provided by french government.\n",
    "\n",
    "* This [flights database](https://opensky-network.org/?fbclid=IwAR0kv1M_c0D27yvT__BfF7T0CCOoqi8V2cxT6hoj3gr-JvIHHayCdnbVOLA) in which all flights are listed in the main airports worldwide\n",
    "\n",
    "* [Air quality data](https://aqicn.org/data-platform/covid19/?fbclid=IwAR1nn_Tk9fLpg7uYIZ9tevK9wbubCfr-B-PH5JnXEjX9G_1SJWroD96tpUA) in the biggest cities worldwide, gathered by a little group of data scientists.\n",
    "\n",
    "* Regarding locations data, a [GeoJSON file](https://france-geojson.gregoiredavid.fr/) of French departments was needed. For some data (air traffic and air pollution), a [dataset of the cities of France](https://www.data.gouv.fr/fr/datasets/regions-departements-villes-et-villages-de-france-et-doutre-mer/) including their geographical coordinates was also needed. Finally, as the energy consumption is calculated by region (a region is an entity that includes several departments in France), a dataset linking regions and departments was required.\n",
    "\n",
    "Regarding road traffic, a dataset with data for the whole country does not exist. Only a few road traffic data within some departments are available. It seems that the French government does not centralize data collected by each department. Only private operators (Michelin, etc.) have this data set, but they did not wish to share it."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### **Goal for the end user's experience**\n",
    "The objective is to provide the user with an interactive dashboard enabling him to browse the various data available and to visualize the Covid-19 impacts both at a regional scale (map) and at a country scale (line plots)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# <span style=\"color:#990000\"> II - Basic stats <a class=\"anchor\" id=\"second-part\">"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### **Choices in data cleaning and preprocessing**\n",
    "* There is a lot of specific data cleaning and preprocessing for each dataset. Everything is presented below this part (good luck, it's long).\n",
    "* Regarding preprocessing, choice was made to prefill dictionaries with all the data formatted and calculated upstream rather than proceeding with modifications/calculations on databases at the request of the user. We wanted to make interactive graphs, and if this method requires a longer loading time at the start, it then makes it possible to navigate faster through the visualization."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <span style=\"color:#990000\"> **There are lot of datasets and the preprocessing work is very long. Everything was put here in order to fluidify the navigation in the notebook. You might want to skip the next section and go straight to the discussion on basic statistics. To do so, click [here](#datasets_stats).**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <span style=\"color:#008000\">II.1 Location Data (Used to render France Map)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <span style=\"color:#008000\">a - [GeoJSON file](https://france-geojson.gregoiredavid.fr/) of departments in France"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" Load GeoJson file and store data in variables\"\"\"\n",
    "\n",
    "# Load geojson file with department data\n",
    "with open('datasets/departements.geojson.txt') as f:\n",
    "    geodata = json.load(f)\n",
    "    \n",
    "\n",
    "# Initialization\n",
    "department = {}          # To store geodata for each department\n",
    "lat_by_department = []   # List of lists of lat for each department\n",
    "lon_by_department = []   # List of list of lon for each department\n",
    "name_department = []     # List of names of departments\n",
    "\n",
    "\n",
    "# Loop on all department in the geojson loaded file\n",
    "for feature in geodata['features']:\n",
    "    \n",
    "    # Retrieves the number of the department and add it to list\n",
    "    number = feature['properties']['code'] \n",
    "    \n",
    "    \n",
    "    # Retrieves the name of the department and add it to list\n",
    "    name = feature['properties']['nom']\n",
    "    name_department.append(name)\n",
    "    \n",
    "    # Retrieves the coordinates of the shape of the department\n",
    "    coordinates = feature['geometry']['coordinates']  \n",
    "    \n",
    "    lat = []\n",
    "    lon = []\n",
    "    \n",
    "    # For some region, the coordinates are entered in a really weird way\n",
    "    # You have to play with lists nested in lists\n",
    "    for j in range(len(coordinates)):\n",
    "        lat_intermediate = []\n",
    "        lon_intermediate = []\n",
    "        for i in range(len(coordinates[j])):\n",
    "            if len(coordinates[j][i]) == 2:\n",
    "                lat_intermediate.append(coordinates[j][i][1])\n",
    "                lon_intermediate.append(coordinates[j][i][0])\n",
    "            else:\n",
    "                for k in range(len(coordinates[j][i])):\n",
    "                    lat_intermediate.append(coordinates[j][i][k][1])\n",
    "                    lon_intermediate.append(coordinates[j][i][k][0])\n",
    "        if len(lat_intermediate) > len(lat):\n",
    "            lat = lat_intermediate\n",
    "            lon = lon_intermediate\n",
    "           \n",
    "\n",
    "    # Add to list of lat and lon for each department\n",
    "    lat_by_department.append(lat)\n",
    "    lon_by_department.append(lon)    \n",
    "    \n",
    "    # Add new entry to dictionary\n",
    "    # The number of department is the key\n",
    "    # The value of the dictionary is a dictionary containign both the name of the department, and list with latitudes and longitudes\n",
    "    department[number] = {'name':name, 'lat': lat, 'lon': lon}\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <span style=\"color:#008000\">b- Dataset to match departments and regions in France"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# load dataset\n",
    "df_match_region_department = pd.read_csv(\"datasets/region_to_department.csv\", sep=\";\")\n",
    "\n",
    "# Drop NA\n",
    "df_match_region_department = df_match_region_department.dropna()\n",
    "\n",
    "\"\"\" ------------ Create matching dictionary ------------\"\"\"\n",
    "# Initialize\n",
    "region_to_department = {}\n",
    "\n",
    "# Fill dictionnary\n",
    "for id in list(df_match_region_department[\"NUMÉRO\"].unique()):\n",
    "    region = list(df_match_region_department.loc[df_match_region_department['NUMÉRO'] == id, 'REGION'])[0]\n",
    "    \n",
    "    # Adapt to dataframe way of writting Grand-Est\n",
    "    if region == \"Grand Est\":\n",
    "        region = \"Grand-Est\"\n",
    "        \n",
    "    region_to_department[id] = region\n",
    "    \n",
    "\n",
    "# Check if all id from geojson file are in the matching dictionary\n",
    "for id in department:\n",
    "    if id not in region_to_department:\n",
    "        print(\"This id has no match:\", id)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <span style=\"color:#008000\">c- [Dataset of cities](https://www.data.gouv.fr/fr/datasets/regions-departements-villes-et-villages-de-france-et-doutre-mer/) in France"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>name</th>\n",
       "      <th>lon</th>\n",
       "      <th>lat</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>Ozan</td>\n",
       "      <td>4.91667</td>\n",
       "      <td>46.3833</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>Cormoranche-sur-Saône</td>\n",
       "      <td>4.83333</td>\n",
       "      <td>46.2333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>Plagne</td>\n",
       "      <td>5.73333</td>\n",
       "      <td>46.1833</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>Tossiat</td>\n",
       "      <td>5.31667</td>\n",
       "      <td>46.1333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>Pouillat</td>\n",
       "      <td>5.43333</td>\n",
       "      <td>46.3333</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                    name      lon      lat\n",
       "1                   Ozan  4.91667  46.3833\n",
       "2  Cormoranche-sur-Saône  4.83333  46.2333\n",
       "3                 Plagne  5.73333  46.1833\n",
       "4                Tossiat  5.31667  46.1333\n",
       "5               Pouillat  5.43333  46.3333"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# load dataset\n",
    "df_cities = pd.read_csv(\"datasets/villes_france.csv\", sep=\",\", names=[i for i in range(26)])\n",
    "\n",
    "# Keep only columns we are interested in\n",
    "df_cities = df_cities[[4, 18, 19]]\n",
    "\n",
    "# Rename columns\n",
    "df_cities.columns = ['name', 'lon', 'lat']\n",
    "\n",
    "df_cities.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <span style=\"color:#008000\"> II.2 Hospital [data](https://www.data.gouv.fr/fr/datasets/donnees-hospitalieres-relatives-a-lepidemie-de-covid-19/## )\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <span style=\"color:#008000\">a- General preparation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(5758, 7)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>hospitalized</th>\n",
       "      <th>intensive_care</th>\n",
       "      <th>death</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>date</th>\n",
       "      <th>department</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td rowspan=\"5\" valign=\"top\">2020-03-18</td>\n",
       "      <td>01</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>02</td>\n",
       "      <td>41</td>\n",
       "      <td>10</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>03</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>04</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>05</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                       hospitalized  intensive_care  death\n",
       "date       department                                     \n",
       "2020-03-18 01                     2               0      0\n",
       "           02                    41              10     11\n",
       "           03                     4               0      0\n",
       "           04                     3               1      0\n",
       "           05                     8               1      0"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_hospital = pd.read_csv(\"datasets\\donnees-hospitalieres-covid19-2020.csv\", sep=\";\")\n",
    "\n",
    "# Let's change columns name to make them more meaningful\n",
    "df_hospital.rename(columns={\"dep\":\"department\",\n",
    "                           \"sexe\":\"gender\",   \n",
    "                           \"jour\":\"date\",\n",
    "                           \"hosp\":\"hospitalized\",\n",
    "                           \"rea\":\"intensive_care\",\n",
    "                           \"rad\":\"returned_home\",\n",
    "                           \"dc\":\"death\"}, inplace=True)\n",
    "\n",
    "# We only need gender 0 which refers to men+women\n",
    "df_hospital = df_hospital[df_hospital.gender==0]\n",
    "\n",
    "# Convert string date to datetime\n",
    "df_hospital[\"date\"] = pd.to_datetime(df_hospital[\"date\"], format='%Y-%m-%d')\n",
    "df_hospital=df_hospital[[\"date\",\"department\",\"gender\",\"hospitalized\",\"intensive_care\",\"death\",\"returned_home\"]]\n",
    "\n",
    "# We are not interested in returned_home and gender\n",
    "df_hospital_groupby = df_hospital.groupby(['date','department']).sum()\n",
    "del df_hospital_groupby[\"gender\"] \n",
    "del df_hospital_groupby[\"returned_home\"]\n",
    "\n",
    "print(df_hospital.shape)\n",
    "df_hospital_groupby.head()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <span style=\"color:#008000\">b- Preparation for line plots"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "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>Hospitalized</th>\n",
       "      <th>Intensive_care</th>\n",
       "      <th>Cumulated_deceased</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>date</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>2020-03-18</td>\n",
       "      <td>2972</td>\n",
       "      <td>771</td>\n",
       "      <td>218</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2020-03-19</td>\n",
       "      <td>4073</td>\n",
       "      <td>1002</td>\n",
       "      <td>327</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2020-03-20</td>\n",
       "      <td>5226</td>\n",
       "      <td>1297</td>\n",
       "      <td>450</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2020-03-21</td>\n",
       "      <td>5900</td>\n",
       "      <td>1453</td>\n",
       "      <td>525</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2020-03-22</td>\n",
       "      <td>6954</td>\n",
       "      <td>1674</td>\n",
       "      <td>632</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            Hospitalized  Intensive_care  Cumulated_deceased\n",
       "date                                                        \n",
       "2020-03-18          2972             771                 218\n",
       "2020-03-19          4073            1002                 327\n",
       "2020-03-20          5226            1297                 450\n",
       "2020-03-21          5900            1453                 525\n",
       "2020-03-22          6954            1674                 632"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#evolution over time for the whole country\n",
    "df_hospital_France = df_hospital_groupby.sum(level=0)\n",
    "\n",
    "# Rename columns\n",
    "df_hospital_France .rename(columns={\"hospitalized\":\"Hospitalized\",\n",
    "                           \"intensive_care\":\"Intensive_care\",   \n",
    "                           \"death\":\"Cumulated_deceased\"}, \n",
    "                   inplace=True)\n",
    "#df_evolution.reset_index(inplace=True)\n",
    "df_hospital_France.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Hospitalized</th>\n",
       "      <th>Intensive_care</th>\n",
       "      <th>Cumulated_deceased</th>\n",
       "      <th>integer</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>date</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>2020-03-18</td>\n",
       "      <td>2972</td>\n",
       "      <td>771</td>\n",
       "      <td>218</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2020-03-19</td>\n",
       "      <td>4073</td>\n",
       "      <td>1002</td>\n",
       "      <td>327</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            Hospitalized  Intensive_care  Cumulated_deceased  integer\n",
       "date                                                                 \n",
       "2020-03-18          2972             771                 218        0\n",
       "2020-03-19          4073            1002                 327        1"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\"\"\" Initialization\"\"\"\n",
    "\n",
    "# Get list of dates in the dataframe\n",
    "hospital_dates_str = df_hospital_France.index.strftime('%m/%d/%Y')\n",
    "\n",
    "# Add a new columns to order rows with integers\n",
    "df_hospital_France['integer']=[i for i in range(len(df_hospital_France))]\n",
    "df_hospital_France.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Hospitalized          int64\n",
       "Intensive_care        int64\n",
       "Cumulated_deceased    int64\n",
       "integer               int64\n",
       "dtype: object"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_hospital_France.dtypes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <span style=\"color:#008000\">c- Preparation for bokeh maps"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* #### c.1: Treatment for cumulated hospital data per day:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>gender</th>\n",
       "      <th>hospitalized</th>\n",
       "      <th>intensive_care</th>\n",
       "      <th>death</th>\n",
       "      <th>returned_home</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>department</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>01</td>\n",
       "      <td>0</td>\n",
       "      <td>6491</td>\n",
       "      <td>972</td>\n",
       "      <td>2633</td>\n",
       "      <td>8726</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>02</td>\n",
       "      <td>0</td>\n",
       "      <td>12672</td>\n",
       "      <td>1990</td>\n",
       "      <td>7937</td>\n",
       "      <td>21331</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>03</td>\n",
       "      <td>0</td>\n",
       "      <td>2249</td>\n",
       "      <td>684</td>\n",
       "      <td>898</td>\n",
       "      <td>4598</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>04</td>\n",
       "      <td>0</td>\n",
       "      <td>1333</td>\n",
       "      <td>139</td>\n",
       "      <td>272</td>\n",
       "      <td>3790</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>05</td>\n",
       "      <td>0</td>\n",
       "      <td>2202</td>\n",
       "      <td>474</td>\n",
       "      <td>263</td>\n",
       "      <td>3761</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            gender  hospitalized  intensive_care  death  returned_home\n",
       "department                                                            \n",
       "01               0          6491             972   2633           8726\n",
       "02               0         12672            1990   7937          21331\n",
       "03               0          2249             684    898           4598\n",
       "04               0          1333             139    272           3790\n",
       "05               0          2202             474    263           3761"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# We group by department\n",
    "df_hospital_groupby_department = df_hospital.groupby(['department']).sum()\n",
    "df_hospital_groupby_department.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Show the unmatching departments we remove from the dataframe ['971', '972', '973', '974', '976']\n",
      "The lists are identical\n"
     ]
    }
   ],
   "source": [
    "\"\"\" Filter the dataframe by departments which are both in the dataframe and the geojson \"\"\"\n",
    "\n",
    "# Initialization\n",
    "unmatching_departments = []\n",
    "region_shape_without_data = []\n",
    "\n",
    "# List of ids from Dataframe\n",
    "department_dataframe = list(df_hospital_groupby_department.index.values)\n",
    "\n",
    "# Loop on all departments in the dataframe\n",
    "for id in department_dataframe:  \n",
    "    \n",
    "    # We remove the departments which are not in the geoJson file\n",
    "    if id not in department:\n",
    "        unmatching_departments.append(id)\n",
    "        df_hospital_groupby_department = df_hospital_groupby_department.drop(id)\n",
    "\n",
    "print(\"Show the unmatching departments we remove from the dataframe\", unmatching_departments)\n",
    "\n",
    "\n",
    "\"\"\" Check if data from shapefile and data from dataframe are the same \"\"\"\n",
    "department_dataframe = sorted(list(df_hospital_groupby_department.index.values))\n",
    "department_geojson = sorted(list(department.keys()))\n",
    "  \n",
    "# using == to check if  \n",
    "# lists are equal \n",
    "if department_dataframe == department_geojson: \n",
    "    print (\"The lists are identical\") \n",
    "else : \n",
    "    print (\"The lists are not identical\") "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> The unmatching departments are the French overseas departments"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" Get data from dataframe \"\"\"\n",
    "\n",
    "# Initialization\n",
    "hospitalized_by_department = []\n",
    "intensive_care_by_department = []\n",
    "death_by_department = []\n",
    "\n",
    "\n",
    "# Fill the lists of data for each department in the same order as they appear in the department dictionary\n",
    "for id in department:\n",
    "    hospitalized_by_department.append(df_hospital_groupby_department.at[id, 'hospitalized'])\n",
    "    intensive_care_by_department.append(df_hospital_groupby_department.at[id, 'intensive_care'])\n",
    "    death_by_department.append(df_hospital_groupby_department.at[id, 'death'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* #### c.2:Treatment for hospital data per day"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "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>department</th>\n",
       "      <th>gender</th>\n",
       "      <th>hospitalized</th>\n",
       "      <th>intensive_care</th>\n",
       "      <th>death</th>\n",
       "      <th>returned_home</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>2020-03-18</td>\n",
       "      <td>01</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>2020-03-18</td>\n",
       "      <td>02</td>\n",
       "      <td>0</td>\n",
       "      <td>41</td>\n",
       "      <td>10</td>\n",
       "      <td>11</td>\n",
       "      <td>18</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>2020-03-18</td>\n",
       "      <td>03</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>2020-03-18</td>\n",
       "      <td>04</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>2020-03-18</td>\n",
       "      <td>05</td>\n",
       "      <td>0</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         date department  gender  hospitalized  intensive_care  death  \\\n",
       "0  2020-03-18         01       0             2               0      0   \n",
       "3  2020-03-18         02       0            41              10     11   \n",
       "6  2020-03-18         03       0             4               0      0   \n",
       "9  2020-03-18         04       0             3               1      0   \n",
       "12 2020-03-18         05       0             8               1      0   \n",
       "\n",
       "    returned_home  \n",
       "0               1  \n",
       "3              18  \n",
       "6               1  \n",
       "9               2  \n",
       "12              9  "
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Show again the dataset we are using\n",
    "df_hospital.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Date present for all departments!'"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check that the dataframe starts on the same date for each department.\n",
    "def check_if_minimum_date_in_all_department():\n",
    "    departments_without_date = []\n",
    "    \n",
    "    for id in department:\n",
    "        if ((df_hospital[\"date\"] == df_hospital[\"date\"].min()) & (df_hospital['department'] == id)).any() == False:\n",
    "            departments_without_date.append(id)\n",
    "    \n",
    "    if not departments_without_date:\n",
    "        return(\"Date present for all departments!\")\n",
    "    else:\n",
    "        return(\"No date for following departments\", departments_without_date)\n",
    "\n",
    "\n",
    "check_if_minimum_date_in_all_department()   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>gender</th>\n",
       "      <th>hospitalized</th>\n",
       "      <th>intensive_care</th>\n",
       "      <th>death</th>\n",
       "      <th>returned_home</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>department</th>\n",
       "      <th>date</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td rowspan=\"5\" valign=\"top\">01</td>\n",
       "      <td>2020-03-18</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2020-03-19</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2020-03-20</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2020-03-21</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2020-03-22</td>\n",
       "      <td>0</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                       gender  hospitalized  intensive_care  death  \\\n",
       "department date                                                      \n",
       "01         2020-03-18       0             2               0      0   \n",
       "           2020-03-19       0             2               0      0   \n",
       "           2020-03-20       0             2               0      0   \n",
       "           2020-03-21       0             4               0      0   \n",
       "           2020-03-22       0             8               1      0   \n",
       "\n",
       "                       returned_home  \n",
       "department date                       \n",
       "01         2020-03-18              1  \n",
       "           2020-03-19              1  \n",
       "           2020-03-20              1  \n",
       "           2020-03-21              1  \n",
       "           2020-03-22              2  "
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# There can be several lines for the same department and date. They must be grouped together.\n",
    "df_hospital.loc[(df_hospital['date'] == df_hospital[\"date\"].min()) & (df_hospital['department'] == \"01\")]\n",
    "\n",
    "# We group by department and date\n",
    "df_hospital_groupby_department_and_date = df_hospital.groupby(['department', 'date']).sum()\n",
    "df_hospital_groupby_department_and_date.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* #### c.3:Store data in dictionaries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" Initialization for all type of maps\"\"\"\n",
    "\n",
    "# Get list of dates in the dataframe\n",
    "hospital_dates = list(df_hospital['date'].unique())\n",
    "\n",
    "# Initialization of list to fill with dates from dataframe as String\n",
    "hospital_dates_str = [] \n",
    "\n",
    "# Initizialisation of a counter of dates\n",
    "# We want each date to be provided an unique int id to use if in bokeh later\n",
    "id_date = 0\n",
    "\n",
    "\"\"\" Initialization concerning maps with non cumulated data \"\"\"\n",
    "\n",
    "# Initialization of dictionaries to contain data for each date and each department\n",
    "hospitalized_by_date_as_integer = {}\n",
    "intensive_care_by_date_as_integer = {}\n",
    "death_by_date_as_integer = {}\n",
    "\n",
    "\n",
    "\"\"\" Initialization concerning maps with cumulated data \"\"\"\n",
    "\n",
    "# Initialization of dictionaries to contain cumulated data for each date and each department\n",
    "cumulated_hospitalized_by_date_as_integer = {}\n",
    "cumulated_intensive_care_by_date_as_integer = {}\n",
    "cumulated_death_by_date_as_integer = {} \n",
    "\n",
    "# Initialization of list to compute cumulated data for each day\n",
    "cumulated_hospitalized_helper = []\n",
    "cumulated_intensive_care_helper = []\n",
    "\n",
    "\n",
    "\"\"\" Fill all dictionaries \"\"\"\n",
    "# Create a dictionary with all dates as key, and list of related data for each department as the related date as value\n",
    "# The list of data for each department is field in the same order as the department appear in the department dictionary\n",
    "for date in hospital_dates:\n",
    "    \n",
    "    # Initialization\n",
    "    hospitalized_by_department_at_date = []\n",
    "    intensive_care_by_department_at_date = []\n",
    "    death_by_department_at_date = []\n",
    "    \n",
    "    # Find values in dataframe\n",
    "    for id in department:\n",
    "        hospitalized_by_department_at_date.append(df_hospital_groupby_department_and_date.at[(id, date), 'hospitalized'])\n",
    "        intensive_care_by_department_at_date.append(df_hospital_groupby_department_and_date.at[(id, date), 'intensive_care'])\n",
    "        death_by_department_at_date.append(df_hospital_groupby_department_and_date.at[(id, date), 'death'])\n",
    "    \n",
    "    # Convert the date in pandas format datetime64 to format datetime\n",
    "    date = convert_datetime64_to_datetime(date)\n",
    "    \n",
    "    # Convert the date in datetime format as a String\n",
    "    date = date.strftime(\"%m/%d/%Y\")\n",
    "    \n",
    "    # Add to the String converted date list\n",
    "    hospital_dates_str.append(date)\n",
    "    \n",
    "    # Update non cumulated data dictionaries\n",
    "    hospitalized_by_date_as_integer[id_date] = hospitalized_by_department_at_date\n",
    "    intensive_care_by_date_as_integer[id_date] = intensive_care_by_department_at_date\n",
    "    \n",
    "    # Death is already entered in the dataframe as a cumulated data\n",
    "    cumulated_death_by_date_as_integer[id_date] = death_by_department_at_date\n",
    "    \n",
    "    # Compute number of new deaths\n",
    "    if id_date == 0:\n",
    "        death_by_date_as_integer[id_date] = cumulated_death_by_date_as_integer[id_date]\n",
    "    else:\n",
    "        # Zip list with pairs of elements from ancients and new cumulated data\n",
    "        death_by_department_at_date_helper = zip(cumulated_death_by_date_as_integer[id_date], cumulated_death_by_date_as_integer[id_date-1])\n",
    "    \n",
    "        # Add difference to list\n",
    "        death_by_date_as_integer[id_date] = [a_i - b_i for a_i, b_i in death_by_department_at_date_helper]\n",
    "        \n",
    "    \n",
    "    # Update helpers to compute cumulated data\n",
    "    if not cumulated_hospitalized_helper: # In the first turn, lists are empty\n",
    "        cumulated_hospitalized_helper = hospitalized_by_department_at_date\n",
    "        cumulated_intensive_care_helper = intensive_care_by_department_at_date\n",
    "    else:\n",
    "        # Create zipped lists with pairs of items from both lists\n",
    "        zipped_hospitalized_lists = zip(cumulated_hospitalized_helper, hospitalized_by_department_at_date) \n",
    "        zipped_intensive_lists = zip(cumulated_intensive_care_helper, intensive_care_by_department_at_date) \n",
    "       \n",
    "        \n",
    "        # Sum each element of zipped list\n",
    "        cumulated_hospitalized_helper = [sum(x) for x in zipped_hospitalized_lists]\n",
    "        cumulated_intensive_care_helper = [sum(x) for x in zipped_intensive_lists]\n",
    "    \n",
    "    \n",
    "    # Add to cumulated dictionaries  \n",
    "    cumulated_hospitalized_by_date_as_integer[id_date] = cumulated_hospitalized_helper\n",
    "    cumulated_intensive_care_by_date_as_integer[id_date] = cumulated_intensive_care_helper\n",
    "\n",
    "        \n",
    "    # Increase counter\n",
    "    id_date += 1\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" Arranges dictionaries for all three features, cumulated or not \"\"\"\n",
    "# Find the key of the latest updated cumulated data\n",
    "last_key = list(cumulated_hospitalized_by_date_as_integer.keys())[-1] \n",
    "\n",
    "# Create and fill dictionary\n",
    "hospital_data_by_feature = {}\n",
    "hospital_data_by_feature['synthesis'] = {\"hospitalized\":cumulated_hospitalized_by_date_as_integer[last_key],\n",
    "                             \"intensive_care\": cumulated_intensive_care_by_date_as_integer[last_key],\n",
    "                             \"death\": cumulated_death_by_date_as_integer[last_key]}\n",
    "\n",
    "hospital_data_by_feature['by_date'] = {\"hospitalized\":hospitalized_by_date_as_integer,\n",
    "                             \"intensive_care\": intensive_care_by_date_as_integer,\n",
    "                             \"death\": death_by_date_as_integer}\n",
    "\n",
    "hospital_data_by_feature['cumulated'] = {\"hospitalized\":cumulated_hospitalized_by_date_as_integer,\n",
    "                             \"intensive_care\": cumulated_intensive_care_by_date_as_integer,\n",
    "                             \"death\": cumulated_death_by_date_as_integer}\n",
    "\n",
    "# Dictionary of min / max values for each combo of mode/feature\n",
    "hospital_data_min_max = {}\n",
    "\n",
    "hospital_data_min_max['synthesis'] = {\"hospitalized\": {\"min\": df_hospital_groupby_department[\"hospitalized\"].min(),\n",
    "                                                       \"max\": df_hospital_groupby_department[\"hospitalized\"].max()},\n",
    "                             \"intensive_care\": {\"min\": df_hospital_groupby_department[\"intensive_care\"].min(),\n",
    "                                                       \"max\": df_hospital_groupby_department[\"intensive_care\"].max()},\n",
    "                             \"death\": {\"min\": df_hospital_groupby_department[\"death\"].min(),\n",
    "                                                       \"max\": df_hospital_groupby_department[\"death\"].max()}\n",
    "                                     }\n",
    "\n",
    "hospital_data_min_max['by_date'] = {\"hospitalized\": {\"min\": df_hospital_groupby_department_and_date[\"hospitalized\"].min(),\n",
    "                                                       \"max\": df_hospital_groupby_department_and_date[\"hospitalized\"].max()},\n",
    "                             \"intensive_care\": {\"min\": df_hospital_groupby_department_and_date[\"intensive_care\"].min(),\n",
    "                                                       \"max\": df_hospital_groupby_department_and_date[\"intensive_care\"].max()},\n",
    "                             \"death\": {\"min\": df_hospital_groupby_department_and_date[\"death\"].min(),\n",
    "                                                       \"max\": df_hospital_groupby_department_and_date[\"death\"].max()}\n",
    "                                     }\n",
    "\n",
    "hospital_data_min_max['cumulated'] = {\"hospitalized\": {\"min\": df_hospital_groupby_department[\"hospitalized\"].min(),\n",
    "                                                       \"max\": df_hospital_groupby_department[\"hospitalized\"].max()},\n",
    "                             \"intensive_care\": {\"min\": df_hospital_groupby_department[\"intensive_care\"].min(),\n",
    "                                                       \"max\": df_hospital_groupby_department[\"intensive_care\"].max()},\n",
    "                             \"death\": {\"min\": df_hospital_groupby_department[\"death\"].min(),\n",
    "                                                       \"max\": df_hospital_groupby_department[\"death\"].max()}\n",
    "                                     }"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <span style=\"color:#008000\"> II.3 Energy [data](https://www.data.gouv.fr/en/datasets/donnees-eco2mix-regionales-temps-reel-1/)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <span style=\"color:#008000\">a- General preparation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(101952, 6)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>region</th>\n",
       "      <th>date</th>\n",
       "      <th>hour</th>\n",
       "      <th>general_consumption</th>\n",
       "      <th>thermal_power</th>\n",
       "      <th>nuclear_power</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>Bretagne</td>\n",
       "      <td>2020-01-01</td>\n",
       "      <td>00:00</td>\n",
       "      <td>3596.0</td>\n",
       "      <td>237.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>Auvergne-Rhône-Alpes</td>\n",
       "      <td>2020-01-01</td>\n",
       "      <td>00:00</td>\n",
       "      <td>8907.0</td>\n",
       "      <td>234.0</td>\n",
       "      <td>11556.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>Pays de la Loire</td>\n",
       "      <td>2020-01-01</td>\n",
       "      <td>00:00</td>\n",
       "      <td>4053.0</td>\n",
       "      <td>347.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 region       date   hour  general_consumption  thermal_power  \\\n",
       "0              Bretagne 2020-01-01  00:00               3596.0          237.0   \n",
       "1  Auvergne-Rhône-Alpes 2020-01-01  00:00               8907.0          234.0   \n",
       "2      Pays de la Loire 2020-01-01  00:00               4053.0          347.0   \n",
       "\n",
       "   nuclear_power  \n",
       "0            NaN  \n",
       "1        11556.0  \n",
       "2            NaN  "
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# As some data can be entered in double in the two dataframes, drop the duplicate rows\n",
    "df_energy = pd.concat([pd.read_csv(\"datasets/eco2mix-regional-2020.csv\", sep=\";\"), #energy january-march \n",
    "                        pd.read_csv(\"datasets/eco2mix-regional-2020_V2.csv\", sep=\";\")], #energy april-may \n",
    "                       ignore_index=True).drop_duplicates().reset_index(drop=True)\n",
    "\n",
    "\n",
    "# Keep only data we are interested in\n",
    "df_energy = df_energy[['Région', 'Date', 'Heure', 'Consommation (MW)', 'Thermique (MW)', 'Nucléaire (MW)']]\n",
    "\n",
    "\n",
    "# Rename columns\n",
    "df_energy.rename(columns={\"Région\":\"region\",\n",
    "                           \"Date\":\"date\",    \n",
    "                           \"Heure\":\"hour\",\n",
    "                           \"Consommation (MW)\":\"general_consumption\",\n",
    "                           \"Thermique (MW)\":\"thermal_power\",\n",
    "                           \"Nucléaire (MW)\":\"nuclear_power\"\n",
    "                           }, inplace=True)\n",
    "\n",
    "# Drop NA values\n",
    "df_energy = df_energy.dropna(subset=['date'])\n",
    "\n",
    "# Convert string date to datetime\n",
    "df_energy[\"date\"] = pd.to_datetime(df_energy[\"date\"], format='%Y-%m-%d')\n",
    "\n",
    "# Show dataframe\n",
    "print(df_energy.shape)\n",
    "df_energy.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <span style=\"color:#008000\">b- Preparation for line plots"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "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>general_consumption</th>\n",
       "      <th>thermal_power</th>\n",
       "      <th>nuclear_power</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>date</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>2020-01-01</td>\n",
       "      <td>60577.291667</td>\n",
       "      <td>4714.416667</td>\n",
       "      <td>49636.041667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2020-01-02</td>\n",
       "      <td>64768.437500</td>\n",
       "      <td>7282.958333</td>\n",
       "      <td>50279.875000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            general_consumption  thermal_power  nuclear_power\n",
       "date                                                         \n",
       "2020-01-01         60577.291667    4714.416667   49636.041667\n",
       "2020-01-02         64768.437500    7282.958333   50279.875000"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Group by date and region\n",
    "df_energy_groupby = df_energy.groupby([\"date\",\"region\"]).mean()\n",
    "\n",
    "#energy consumption over time for the whole country\n",
    "df_energy2020 = df_energy_groupby.sum(level=0)\n",
    "df_energy2020.sort_index()\n",
    "\n",
    "df_energy2020.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Code INSEE région</th>\n",
       "      <th>Consommation (MW)</th>\n",
       "      <th>Thermique (MW)</th>\n",
       "      <th>Nucléaire (MW)</th>\n",
       "      <th>Eolien (MW)</th>\n",
       "      <th>Solaire (MW)</th>\n",
       "      <th>Hydraulique (MW)</th>\n",
       "      <th>Pompage (MW)</th>\n",
       "      <th>Bioénergies (MW)</th>\n",
       "      <th>Ech. physiques (MW)</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>date</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>2020-01-01</td>\n",
       "      <td>599.0</td>\n",
       "      <td>58485.291667</td>\n",
       "      <td>3339.458333</td>\n",
       "      <td>52828.500000</td>\n",
       "      <td>2127.8125</td>\n",
       "      <td>345.270833</td>\n",
       "      <td>4127.020833</td>\n",
       "      <td>-2092.5</td>\n",
       "      <td>1206.166667</td>\n",
       "      <td>-3395.583333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2020-01-02</td>\n",
       "      <td>599.0</td>\n",
       "      <td>66144.083333</td>\n",
       "      <td>6775.562500</td>\n",
       "      <td>55361.958333</td>\n",
       "      <td>3309.3125</td>\n",
       "      <td>590.479167</td>\n",
       "      <td>5332.270833</td>\n",
       "      <td>-943.5</td>\n",
       "      <td>1190.687500</td>\n",
       "      <td>-5472.625000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            Code INSEE région  Consommation (MW)  Thermique (MW)  \\\n",
       "date                                                               \n",
       "2020-01-01              599.0       58485.291667     3339.458333   \n",
       "2020-01-02              599.0       66144.083333     6775.562500   \n",
       "\n",
       "            Nucléaire (MW)  Eolien (MW)  Solaire (MW)  Hydraulique (MW)  \\\n",
       "date                                                                      \n",
       "2020-01-01    52828.500000    2127.8125    345.270833       4127.020833   \n",
       "2020-01-02    55361.958333    3309.3125    590.479167       5332.270833   \n",
       "\n",
       "            Pompage (MW)  Bioénergies (MW)  Ech. physiques (MW)  \n",
       "date                                                             \n",
       "2020-01-01       -2092.5       1206.166667         -3395.583333  \n",
       "2020-01-02        -943.5       1190.687500         -5472.625000  "
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Same manipulation for 2019 data\n",
    "df_energy2019 = pd.read_csv(\"datasets\\eco2mix-regional-2019.csv\", sep=\";\")\n",
    "\n",
    "#convert string date to timestamp\n",
    "df_energy2019.rename(columns={'Date': 'date'}, inplace=True)\n",
    "df_energy2019[\"date\"] = pd.to_datetime(df_energy2019[\"date\"].apply(lambda x:\"2020\"+x[4:]), format='%Y-%m-%d') #cheat for merging\n",
    "df_energy2019.drop(['Heure', 'Date - Heure'], axis=1, inplace=True)\n",
    "\n",
    "#energy consumption over time for the whole country\n",
    "df_energy2019 = df_energy2019.groupby([\"date\",\"Région\"]).mean().sum(level=0)\n",
    "df_energy2019.sort_index().head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "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>2019 (MW)</th>\n",
       "      <th>2020 (MW)</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>date</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>2020-01-01</td>\n",
       "      <td>3339.458333</td>\n",
       "      <td>4714.416667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2020-01-02</td>\n",
       "      <td>6775.562500</td>\n",
       "      <td>7282.958333</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              2019 (MW)    2020 (MW)\n",
       "date                                \n",
       "2020-01-01  3339.458333  4714.416667\n",
       "2020-01-02  6775.562500  7282.958333"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Merge 2019 and 2020 datasets\n",
    "df_consumption = pd.merge(df_energy2019['Consommation (MW)'], df_energy2020['general_consumption'], \n",
    "                               how='inner', on='date')\n",
    "\n",
    "# Rename such that all datasets have same features (otherwise it doesn't work on bokeh)\n",
    "df_consumption.rename(columns={'Consommation (MW)': '2019 (MW)', \n",
    "                               'general_consumption': '2020 (MW)'},\n",
    "                      inplace=True)\n",
    "\n",
    "df_thermic = pd.merge(df_energy2019['Thermique (MW)'], df_energy2020['thermal_power'], \n",
    "                               how='inner', on='date')\n",
    "df_thermic.rename(columns={'Thermique (MW)': '2019 (MW)', \n",
    "                               'thermal_power': '2020 (MW)'},\n",
    "                      inplace=True)\n",
    "\n",
    "df_nuclear = pd.merge(df_energy2019['Nucléaire (MW)'], df_energy2020['nuclear_power'], \n",
    "                               how='inner', on='date')\n",
    "df_nuclear.rename(columns={'Nucléaire (MW)': '2019 (MW)', \n",
    "                               'nuclear_power': '2020 (MW)'},\n",
    "                      inplace=True)\n",
    "\n",
    "df_thermic.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" Initialization\"\"\"\n",
    "\n",
    "# Get list of dates in the dataframe\n",
    "energy_dates_str = df_consumption.index.strftime('%m/%d/%Y')\n",
    "\n",
    "# Add a new columns to order rows with integers\n",
    "df_consumption['integer'] = [i for i in range(len(df_consumption))]\n",
    "df_thermic['integer'] = [i for i in range(len(df_thermic))]\n",
    "df_nuclear['integer'] = [i for i in range(len(df_nuclear))]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <span style=\"color:#008000\">c- Preparation for bokeh maps"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* #### c.1: Treatment for energy data per day:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "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></th>\n",
       "      <th>general_consumption</th>\n",
       "      <th>thermal_power</th>\n",
       "      <th>nuclear_power</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>region</th>\n",
       "      <th>date</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td rowspan=\"5\" valign=\"top\">Auvergne-Rhône-Alpes</td>\n",
       "      <td>2020-01-01</td>\n",
       "      <td>403305.0</td>\n",
       "      <td>10333.0</td>\n",
       "      <td>560104.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2020-01-02</td>\n",
       "      <td>442850.0</td>\n",
       "      <td>23419.0</td>\n",
       "      <td>561500.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2020-01-03</td>\n",
       "      <td>434036.0</td>\n",
       "      <td>23059.0</td>\n",
       "      <td>573405.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2020-01-04</td>\n",
       "      <td>394635.0</td>\n",
       "      <td>10534.0</td>\n",
       "      <td>556941.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2020-01-05</td>\n",
       "      <td>399483.0</td>\n",
       "      <td>10423.0</td>\n",
       "      <td>571649.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 general_consumption  thermal_power  \\\n",
       "region               date                                             \n",
       "Auvergne-Rhône-Alpes 2020-01-01             403305.0        10333.0   \n",
       "                     2020-01-02             442850.0        23419.0   \n",
       "                     2020-01-03             434036.0        23059.0   \n",
       "                     2020-01-04             394635.0        10534.0   \n",
       "                     2020-01-05             399483.0        10423.0   \n",
       "\n",
       "                                 nuclear_power  \n",
       "region               date                       \n",
       "Auvergne-Rhône-Alpes 2020-01-01       560104.0  \n",
       "                     2020-01-02       561500.0  \n",
       "                     2020-01-03       573405.0  \n",
       "                     2020-01-04       556941.0  \n",
       "                     2020-01-05       571649.0  "
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# The consumption reading is taken every quarter of an hour in a day. \n",
    "# We're interested in daily data, so we're summing by day\n",
    "df_energy_by_department_and_date = df_energy.groupby(['region', 'date']).sum()\n",
    "\n",
    "# Show dataframe\n",
    "df_energy_by_department_and_date.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "As data are entered as 12 regions, we need to map them to departments\n"
     ]
    }
   ],
   "source": [
    "print(\"As data are entered as \" + str(len(list(df_energy['region'].unique()))) + \" regions, we need to map them to departments\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* #### c.2: Sotre data in dictionaries:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" Initialization for all type of maps\"\"\"\n",
    "\n",
    "# Get list of dates in the dataframe\n",
    "energy_dates = list(df_energy['date'].unique())\n",
    "\n",
    "# Get list of regions in dataframe\n",
    "energy_regions = list(df_energy['region'].unique())\n",
    "\n",
    "# Initialization of list to fill with dates from dataframe as String\n",
    "energy_dates_str = [] \n",
    "\n",
    "# Initizialisation of a counter of dates\n",
    "# We want each date to be provided an unique int id to use if in bokeh later\n",
    "id_date = 0\n",
    "\n",
    "\"\"\" Initialization concerning maps with non cumulated data \"\"\"\n",
    "\n",
    "# Initialization of dictionaries to contain data for each date and each department\n",
    "general_consumption_by_date_as_integer = {}\n",
    "thermal_power_by_date_as_integer = {}\n",
    "nuclear_power_by_date_as_integer = {}\n",
    "\n",
    "\n",
    "\n",
    "\"\"\" Fill all dictionaries \"\"\"\n",
    "# Create a dictionary with all dates as key, and list of related data for each department as the related date as value\n",
    "# The list of data for each department is field in the same order as the department appear in the department dictionary\n",
    "for date in energy_dates:\n",
    "    \n",
    "    # Initialization\n",
    "    general_consumption_by_department_at_date = []\n",
    "    thermal_power_by_department_at_date = []\n",
    "    nuclear_power_by_department_at_date = []\n",
    "    \n",
    "    # Find values in dataframe\n",
    "    for id in department:\n",
    "        region = region_to_department[id]\n",
    "    \n",
    "        # Only 7 regions ar ein the dataset\n",
    "        # Overseas regions are not included in the dataset. They are assigned negative consumption values\n",
    "        if region not in energy_regions:\n",
    "            general_consumption_by_department_at_date.append(0)\n",
    "            thermal_power_by_department_at_date.append(0)\n",
    "            nuclear_power_by_department_at_date.append(0)\n",
    "            \n",
    "        # For some region we miss data at some date\n",
    "        # We replace the values with -1 values for now\n",
    "        # Would be better to replace it with an average of surrounding values\n",
    "        elif (region, date) not in df_energy_by_department_and_date.index:\n",
    "            general_consumption_by_department_at_date.append(0)\n",
    "            thermal_power_by_department_at_date.append(0)\n",
    "            nuclear_power_by_department_at_date.append(0)\n",
    "\n",
    "        else:\n",
    "            general_consumption_by_department_at_date.append(df_energy_by_department_and_date.at[(region, date), 'general_consumption'])\n",
    "            thermal_power_by_department_at_date.append(df_energy_by_department_and_date.at[(region, date), 'thermal_power'])\n",
    "            nuclear_power_by_department_at_date.append(df_energy_by_department_and_date.at[(region, date), 'nuclear_power'])\n",
    "    \n",
    "    \n",
    "    # Convert the date in pandas format datetime64 to format datetime\n",
    "    date = convert_datetime64_to_datetime(date)\n",
    "    \n",
    "    # Convert the date in datetime format as a String\n",
    "    date = date.strftime(\"%m/%d/%Y\")\n",
    "    \n",
    "    # Add to the String converted date list\n",
    "    energy_dates_str.append(date)\n",
    "    \n",
    "    # Update non cumulated data dictionaries\n",
    "    general_consumption_by_date_as_integer[id_date] = general_consumption_by_department_at_date\n",
    "    thermal_power_by_date_as_integer[id_date] = thermal_power_by_department_at_date\n",
    "    nuclear_power_by_date_as_integer[id_date] = nuclear_power_by_department_at_date\n",
    "    \n",
    "        \n",
    "    # Increase counter\n",
    "    id_date += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" Arranges dictionaries for all three features, cumulated or not \"\"\"\n",
    "# Create and fill dictionary\n",
    "energy_data_by_feature = {}\n",
    "\n",
    "# Here we have only one feature\n",
    "energy_data_by_feature['by_date'] = {\"general_consumption\":general_consumption_by_date_as_integer,\n",
    "                             \"thermal_power\": thermal_power_by_date_as_integer,\n",
    "                             \"nuclear_power\": nuclear_power_by_date_as_integer}\n",
    "\n",
    "# Dictionary of min / max values for each combo of mode/feature\n",
    "energy_data_min_max = {}                                    \n",
    "\n",
    "energy_data_min_max['by_date'] = {\"general_consumption\": {\"min\": df_energy_by_department_and_date[\"general_consumption\"].min(),\n",
    "                                                       \"max\": df_energy_by_department_and_date[\"general_consumption\"].max()},\n",
    "                             \"thermal_power\": {\"min\": df_energy_by_department_and_date[\"thermal_power\"].min(),\n",
    "                                                       \"max\": df_energy_by_department_and_date[\"thermal_power\"].max()},\n",
    "                             \"nuclear_power\": {\"min\": df_energy_by_department_and_date[\"nuclear_power\"].min(),\n",
    "                                                       \"max\": df_energy_by_department_and_date[\"nuclear_power\"].max()}\n",
    "                                     }\n",
    "\n",
    "\n",
    "# Dictionary of min / max values for each combo of mode/feature which doesn't take 0 values into account\n",
    "# Usefull as some regions doesn't use any thermal/nuclear energy\n",
    "energy_data_min_max_without_0 = {}          \n",
    "\n",
    "df_energy_by_department_and_date_without_0 =  df_energy_by_department_and_date[(df_energy_by_department_and_date != 0).all(1)]\n",
    "\n",
    "energy_data_min_max_without_0['by_date'] = {\"general_consumption\": {\"min\": df_energy_by_department_and_date_without_0[\"general_consumption\"].min(),\n",
    "                                                       \"max\": df_energy_by_department_and_date_without_0[\"general_consumption\"].max()},\n",
    "                             \"thermal_power\": {\"min\": df_energy_by_department_and_date_without_0[\"thermal_power\"].min(),\n",
    "                                                       \"max\": df_energy_by_department_and_date_without_0[\"thermal_power\"].max()},\n",
    "                             \"nuclear_power\": {\"min\": df_energy_by_department_and_date_without_0[\"nuclear_power\"].min(),\n",
    "                                                       \"max\": df_energy_by_department_and_date_without_0[\"nuclear_power\"].max()}\n",
    "                                     }\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <span style=\"color:#008000\"> II.4 Airline traffic [data](https://opensky-network.org/?fbclid=IwAR0kv1M_c0D27yvT__BfF7T0CCOoqi8V2cxT6hoj3gr-JvIHHayCdnbVOLA)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <span style=\"color:#008000\">a- General preparation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Unnamed: 0</th>\n",
       "      <th>callsign</th>\n",
       "      <th>day</th>\n",
       "      <th>destination</th>\n",
       "      <th>firstseen</th>\n",
       "      <th>icao24</th>\n",
       "      <th>lastseen</th>\n",
       "      <th>number</th>\n",
       "      <th>origin</th>\n",
       "      <th>registration</th>\n",
       "      <th>typecode</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>NaN</td>\n",
       "      <td>FWI16E</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>LFPO</td>\n",
       "      <td>2019-12-31 08:48:57+00:00</td>\n",
       "      <td>39be6f</td>\n",
       "      <td>2020-01-01 05:15:56+00:00</td>\n",
       "      <td>TX516</td>\n",
       "      <td>LFPO</td>\n",
       "      <td>F-HPTP</td>\n",
       "      <td>A333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>NaN</td>\n",
       "      <td>KQA113</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>VTBS</td>\n",
       "      <td>2019-12-31 10:10:18+00:00</td>\n",
       "      <td>04c11b</td>\n",
       "      <td>2020-01-01 05:57:40+00:00</td>\n",
       "      <td>KQ113</td>\n",
       "      <td>LFPG</td>\n",
       "      <td>5Y-KZE</td>\n",
       "      <td>B788</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    Unnamed: 0 callsign                       day destination  \\\n",
       "30         NaN   FWI16E 2020-01-01 00:00:00+00:00        LFPO   \n",
       "40         NaN   KQA113 2020-01-01 00:00:00+00:00        VTBS   \n",
       "\n",
       "                    firstseen  icao24                   lastseen number  \\\n",
       "30  2019-12-31 08:48:57+00:00  39be6f  2020-01-01 05:15:56+00:00  TX516   \n",
       "40  2019-12-31 10:10:18+00:00  04c11b  2020-01-01 05:57:40+00:00  KQ113   \n",
       "\n",
       "   origin registration typecode  \n",
       "30   LFPO       F-HPTP     A333  \n",
       "40   LFPG       5Y-KZE     B788  "
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Load three datasets with lists of flights\n",
    "df_flights = pd.concat([pd.read_csv(\"datasets/flightlist_20200101_20200131.csv\"), #flights on january \n",
    "                        pd.read_csv(\"datasets/flightlist_20200201_20200229.csv\"), #flights on february\n",
    "                        pd.read_csv(\"datasets/flightlist_20200301_20200331.csv\"), #flights on march\n",
    "                        pd.read_csv(\"datasets/flightlist_20200401_20200430.csv\")],\n",
    "                       ignore_index=True)\n",
    "\n",
    "# Drop na values\n",
    "df_flights.dropna()\n",
    "\n",
    "#Get only flights which come or arrive in France\n",
    "df_flights = df_flights[df_flights.origin.str.startswith(\"LF\") | df_flights.destination.str.startswith(\"LF\")]\n",
    "\n",
    "# Convert date string to timestamp\n",
    "df_flights[\"day\"] = pd.to_datetime(df_flights[\"day\"], format='%Y-%m-%d')\n",
    "\n",
    "# Show dataframe\n",
    "df_flights.head(2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Lists of airports in France taken from the  [wiki page](https://en.wikipedia.org/wiki/List_of_airports_in_France):**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "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>day</th>\n",
       "      <th>origin_airport_name</th>\n",
       "      <th>destination_airport_name</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>Orly Airport</td>\n",
       "      <td>Orly Airport</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>Charles de Gaulle Airport</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                        day        origin_airport_name  \\\n",
       "0 2020-01-01 00:00:00+00:00               Orly Airport   \n",
       "1 2020-01-01 00:00:00+00:00  Charles de Gaulle Airport   \n",
       "\n",
       "  destination_airport_name  \n",
       "0             Orly Airport  \n",
       "1                      NaN  "
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_airports = pd.read_csv(\"datasets/List_of_airports_in_France.csv\")\n",
    "\n",
    "# Merge the two datasets because df_flights only contains airport ICAO denomination.\n",
    "# Merge on origin airports\n",
    "df_flights_airports = pd.merge(df_flights, df_airports[['City served / Location','ICAO','Airport name']], \n",
    "                               how='left', left_on='origin', right_on='ICAO')\n",
    "\n",
    "# Rename columns\n",
    "df_flights_airports.rename(columns={\"City served / Location\":\"origin_location\",\n",
    "                                   \"ICAO\":\"origin_ICAO\",\n",
    "                                   \"Airport name\":\"origin_airport_name\"}, inplace=True)\n",
    "# Delete origin column\n",
    "del df_flights_airports['origin']\n",
    "\n",
    "# Merge on destination airports\n",
    "df_flights_airports = pd.merge(df_flights_airports, df_airports[['City served / Location','ICAO','Airport name']], \n",
    "                               how='left', left_on='destination', right_on='ICAO')\n",
    "\n",
    "# Rename columns\n",
    "df_flights_airports.rename(columns={\"City served / Location\":\"destination_location\",\n",
    "                                   \"ICAO\":\"destination_ICAO\",\n",
    "                                   \"Airport name\":\"destination_airport_name\"}, inplace=True)\n",
    "\n",
    "# Delete destination column\n",
    "del df_flights_airports['destination']\n",
    "\n",
    "# Keep only interesting columns\n",
    "df_flights_airports = df_flights_airports[[\"day\", \"origin_airport_name\", \"destination_airport_name\"]]\n",
    "\n",
    "# Show\n",
    "df_flights_airports.head(2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <span style=\"color:#008000\">b- Preparation for line plots"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "There are 97568 intranational flights in France\n",
      "There are 36948 international flights in France\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>total</th>\n",
       "      <th>intranational</th>\n",
       "      <th>international</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>date</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>1754</td>\n",
       "      <td>1270</td>\n",
       "      <td>484</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2020-01-02 00:00:00+00:00</td>\n",
       "      <td>1531</td>\n",
       "      <td>1004</td>\n",
       "      <td>527</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                           total  intranational  international\n",
       "date                                                          \n",
       "2020-01-01 00:00:00+00:00   1754           1270            484\n",
       "2020-01-02 00:00:00+00:00   1531           1004            527"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Get the number of flights each day\n",
    "df_total_traffic = df_flights_airports.groupby('day').count()['origin_airport_name']\n",
    "\n",
    "#Split df_flights_airports between intranational flights and international flights\n",
    "df_intranational_traffic = df_flights_airports[df_flights_airports.origin_airport_name.isin(df_airports['Airport name'].to_list())\n",
    "                                              & df_flights_airports.destination_airport_name.isin(df_airports['Airport name'].to_list())]\n",
    "\n",
    "df_international_traffic = df_flights_airports[~df_flights_airports.index.isin(df_intranational_traffic.index)]\n",
    "\n",
    "#Get the number of intranational flights each day\n",
    "df_intranational_traffic= df_intranational_traffic.groupby('day').count()['origin_airport_name']\n",
    "#Get the number of intranational flights each day\n",
    "df_international_traffic= df_international_traffic.groupby('day').count()['origin_airport_name']\n",
    "\n",
    "\n",
    "print(\"There are\",df_intranational_traffic.sum(),\"intranational flights in France\")\n",
    "print(\"There are\",df_international_traffic.sum(),\"international flights in France\")\n",
    "\n",
    "#Merge the 3 datasets\n",
    "df_air_traffic = pd.merge(df_total_traffic, df_intranational_traffic, on='day') \n",
    "df_air_traffic = pd.merge(df_air_traffic, df_international_traffic, on='day') \n",
    "\n",
    "df_air_traffic.rename(columns={\"day\":\"date\",\n",
    "                               \"origin_airport_name_x\":\"total\",\n",
    "                               \"origin_airport_name_y\":\"intranational\",\n",
    "                               \"origin_airport_name\":\"international\"},inplace=True)\n",
    "df_air_traffic.index.names = ['date']\n",
    "\n",
    "df_air_traffic.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" Initialization\"\"\"\n",
    "\n",
    "# Get list of dates in the dataframe\n",
    "air_dates_str = df_air_traffic.index.strftime('%m/%d/%Y')\n",
    "\n",
    "# Add a new columns to order rows with integers\n",
    "df_air_traffic['integer'] = [i for i in range(len(df_air_traffic))]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <span style=\"color:#008000\">c- Preparation for bokeh maps"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* #### c.1: General treatment to get the city data for each airport:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "There are 44 french airports in the dataset\n"
     ]
    }
   ],
   "source": [
    "\"\"\" Map each airport to the city in which they are\"\"\"\n",
    "\n",
    "# First we have to find all the airports in France\n",
    "# As we filtered the dataframe on the condition of having arrivel or departure in France\n",
    "# The french airports are the one which are both in the list of origin and destination aiports\n",
    "\n",
    "origin_aiports = df_flights_airports[\"origin_airport_name\"].unique()\n",
    "destination_aiports = df_flights_airports[\"destination_airport_name\"].unique()\n",
    "french_airports = [airport for airport in origin_aiports if airport in destination_aiports]\n",
    "print(\"There are \" + str(len(french_airports)) + \" french airports in the dataset\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No match found for Charles de Gaulle Airport\n",
      "No match found for Berre - La Fare Aerodrome (UAF)\n",
      "No match found for Istres-Le Tubé Air Base (BA 125)\n",
      "No match found for Alpes–Isère Airport\n"
     ]
    }
   ],
   "source": [
    "# We now try to find a match for all airport in the cities dataset\n",
    "matches = [[], []]\n",
    "\n",
    "for airport_name in french_airports:\n",
    "    match_found = False\n",
    "    for word in airport_name.replace('–', ' ').split(' '): #split airport names on both \"-\" and \" \"\n",
    "        if word in list(df_cities[\"name\"].unique()):\n",
    "            matches[0].append(airport_name)\n",
    "            matches[1].append(word)\n",
    "            match_found = True\n",
    "    if match_found == False:\n",
    "        print(\"No match found for \" + airport_name)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "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>day</th>\n",
       "      <th>origin_airport_name</th>\n",
       "      <th>destination_airport_name</th>\n",
       "      <th>french_city_origin</th>\n",
       "      <th>french_city_destination</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>Orly Airport</td>\n",
       "      <td>Orly Airport</td>\n",
       "      <td>Orly</td>\n",
       "      <td>Orly</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>Charles de Gaulle Airport</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Paris</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>Orly Airport</td>\n",
       "      <td>Orly Airport</td>\n",
       "      <td>Orly</td>\n",
       "      <td>Orly</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>Charles de Gaulle Airport</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Paris</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>Orly Airport</td>\n",
       "      <td>Orly Airport</td>\n",
       "      <td>Orly</td>\n",
       "      <td>Orly</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                        day        origin_airport_name  \\\n",
       "0 2020-01-01 00:00:00+00:00               Orly Airport   \n",
       "1 2020-01-01 00:00:00+00:00  Charles de Gaulle Airport   \n",
       "2 2020-01-01 00:00:00+00:00               Orly Airport   \n",
       "3 2020-01-01 00:00:00+00:00  Charles de Gaulle Airport   \n",
       "4 2020-01-01 00:00:00+00:00               Orly Airport   \n",
       "\n",
       "  destination_airport_name french_city_origin french_city_destination  \n",
       "0             Orly Airport               Orly                    Orly  \n",
       "1                      NaN              Paris                     NaN  \n",
       "2             Orly Airport               Orly                    Orly  \n",
       "3                      NaN              Paris                     NaN  \n",
       "4             Orly Airport               Orly                    Orly  "
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# For the unfound airports, it is because they have not the name of a city\n",
    "# We map them by hand \n",
    "matches[0].append(\"Charles de Gaulle Airport\")\n",
    "matches[1].append(\"Paris\")\n",
    "matches[0].append(\"Berre - La Fare Aerodrome (UAF)\")\n",
    "matches[1].append(\"Berre-l'Étang\")\n",
    "matches[0].append(\"Istres-Le Tubé Air Base (BA 125)\")\n",
    "matches[1].append(\"Istres\")\n",
    "matches[0].append(\"Alpes–Isère Airport\")\n",
    "matches[1].append(\"Saint-Étienne-de-Saint-Geoirs\")\n",
    "\n",
    "# Create mapping dictionary\n",
    "matching_airport_city = dict(zip(matches[0], matches[1]))\n",
    "\n",
    "# Add two columns city in the airport dataframe using the matching dictionaries\n",
    "df_flights_airports['french_city_origin'] = df_flights_airports['origin_airport_name'].map(matching_airport_city)\n",
    "df_flights_airports['french_city_destination'] = df_flights_airports['destination_airport_name'].map(matching_airport_city)\n",
    "\n",
    "# Show dataframe\n",
    "df_flights_airports.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "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>day</th>\n",
       "      <th>origin_airport_name</th>\n",
       "      <th>destination_airport_name</th>\n",
       "      <th>french_city_origin</th>\n",
       "      <th>french_city_destination</th>\n",
       "      <th>lon_origin</th>\n",
       "      <th>lat_origin</th>\n",
       "      <th>lon_destination</th>\n",
       "      <th>lat_destination</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>Orly Airport</td>\n",
       "      <td>Orly Airport</td>\n",
       "      <td>Orly</td>\n",
       "      <td>Orly</td>\n",
       "      <td>2.40000</td>\n",
       "      <td>48.75</td>\n",
       "      <td>2.4</td>\n",
       "      <td>48.75</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>Charles de Gaulle Airport</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Paris</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.34445</td>\n",
       "      <td>48.86</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>Orly Airport</td>\n",
       "      <td>Orly Airport</td>\n",
       "      <td>Orly</td>\n",
       "      <td>Orly</td>\n",
       "      <td>2.40000</td>\n",
       "      <td>48.75</td>\n",
       "      <td>2.4</td>\n",
       "      <td>48.75</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>Charles de Gaulle Airport</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Paris</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.34445</td>\n",
       "      <td>48.86</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>Orly Airport</td>\n",
       "      <td>Orly Airport</td>\n",
       "      <td>Orly</td>\n",
       "      <td>Orly</td>\n",
       "      <td>2.40000</td>\n",
       "      <td>48.75</td>\n",
       "      <td>2.4</td>\n",
       "      <td>48.75</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                        day        origin_airport_name  \\\n",
       "0 2020-01-01 00:00:00+00:00               Orly Airport   \n",
       "1 2020-01-01 00:00:00+00:00  Charles de Gaulle Airport   \n",
       "2 2020-01-01 00:00:00+00:00               Orly Airport   \n",
       "3 2020-01-01 00:00:00+00:00  Charles de Gaulle Airport   \n",
       "4 2020-01-01 00:00:00+00:00               Orly Airport   \n",
       "\n",
       "  destination_airport_name french_city_origin french_city_destination  \\\n",
       "0             Orly Airport               Orly                    Orly   \n",
       "1                      NaN              Paris                     NaN   \n",
       "2             Orly Airport               Orly                    Orly   \n",
       "3                      NaN              Paris                     NaN   \n",
       "4             Orly Airport               Orly                    Orly   \n",
       "\n",
       "   lon_origin  lat_origin  lon_destination  lat_destination  \n",
       "0     2.40000       48.75              2.4            48.75  \n",
       "1     2.34445       48.86              NaN              NaN  \n",
       "2     2.40000       48.75              2.4            48.75  \n",
       "3     2.34445       48.86              NaN              NaN  \n",
       "4     2.40000       48.75              2.4            48.75  "
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Merge with city dataframe to have lat/lon data\n",
    "df_flight_airports_locations = pd.merge(df_flights_airports, df_cities, how='left', left_on='french_city_origin', right_on='name')\n",
    "del df_flight_airports_locations[\"name\"]\n",
    "df_flight_airports_locations.rename(columns={\"lon\":'lon_origin',\"lat\":'lat_origin'}, inplace=True)\n",
    "\n",
    "df_flight_airports_locations = pd.merge(df_flight_airports_locations, df_cities, how='left', left_on='french_city_destination', right_on='name')\n",
    "del df_flight_airports_locations[\"name\"]\n",
    "df_flight_airports_locations.rename(columns={\"lon\":'lon_destination',\"lat\":'lat_destination'}, inplace=True)\n",
    "\n",
    "# Show dataframe\n",
    "df_flight_airports_locations.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* #### c.2: Treatment for intranational flights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "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></th>\n",
       "      <th>lon</th>\n",
       "      <th>lat</th>\n",
       "      <th>number_flights</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>french_city</th>\n",
       "      <th>day</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>Bastia</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>9.44945</td>\n",
       "      <td>42.7000</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Berre-l'Étang</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>5.18333</td>\n",
       "      <td>43.4667</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Blagnac</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>1.40000</td>\n",
       "      <td>43.6333</td>\n",
       "      <td>42</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Istres</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>4.98333</td>\n",
       "      <td>43.5167</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Landivisiau</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>-4.06667</td>\n",
       "      <td>48.5167</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                             lon      lat  number_flights\n",
       "french_city   day                                                        \n",
       "Bastia        2020-01-01 00:00:00+00:00  9.44945  42.7000               1\n",
       "Berre-l'Étang 2020-01-01 00:00:00+00:00  5.18333  43.4667              11\n",
       "Blagnac       2020-01-01 00:00:00+00:00  1.40000  43.6333              42\n",
       "Istres        2020-01-01 00:00:00+00:00  4.98333  43.5167               3\n",
       "Landivisiau   2020-01-01 00:00:00+00:00 -4.06667  48.5167               5"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Dataframe for intranational travels\n",
    "# Drop NA values is efficient, as international airport have NAN lat/lon\n",
    "df_intranational_flight = df_flight_airports_locations.dropna() \n",
    "\n",
    "# Departures\n",
    "df_intranational_flight_departures = df_intranational_flight.groupby(['day', 'french_city_origin', 'lon_origin', 'lat_origin']).size().reset_index(name=\"number_flights\")\n",
    "\n",
    "# Arrivals\n",
    "df_intranational_flight_arrivals = df_intranational_flight.groupby(['day', 'french_city_destination', 'lon_destination', 'lat_destination']).size().reset_index(name=\"number_flights\")\n",
    "\n",
    "# Get same name of columns in the two datasets\n",
    "df_intranational_flight_departures.columns = [\"day\", \"french_city\", \"lon\", \"lat\", \"number_flights\"]\n",
    "df_intranational_flight_arrivals.columns = [\"day\", \"french_city\", \"lon\", \"lat\", \"number_flights\"]\n",
    "\n",
    "# Sum the two dataframe to get general intranational data\n",
    "df_intranational_flight = pd.concat([df_intranational_flight_departures, df_intranational_flight_arrivals]).groupby(['day','french_city', 'lon', 'lat'])['number_flights'].sum().reset_index()\n",
    "\n",
    "# Set day and city as index\n",
    "df_intranational_flight = df_intranational_flight.set_index(['french_city', 'day'])\n",
    "\n",
    "# Remove duplicate rows\n",
    "df_intranational_flight = df_intranational_flight.loc[~df_intranational_flight.index.duplicated(keep='first')]\n",
    "\n",
    "# Show intranational flights\n",
    "df_intranational_flight.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* #### c.3: Treatment for international flights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "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></th>\n",
       "      <th>lon</th>\n",
       "      <th>lat</th>\n",
       "      <th>number_flights</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>french_city</th>\n",
       "      <th>day</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>Berre-l'Étang</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>5.183330</td>\n",
       "      <td>43.4667</td>\n",
       "      <td>28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Blagnac</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>1.400000</td>\n",
       "      <td>43.6333</td>\n",
       "      <td>30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Lille</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>3.066670</td>\n",
       "      <td>50.6333</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Marseille</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>5.376390</td>\n",
       "      <td>43.2967</td>\n",
       "      <td>45</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Mérignac</td>\n",
       "      <td>2020-01-01 00:00:00+00:00</td>\n",
       "      <td>-0.633333</td>\n",
       "      <td>44.8333</td>\n",
       "      <td>42</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                              lon      lat  number_flights\n",
       "french_city   day                                                         \n",
       "Berre-l'Étang 2020-01-01 00:00:00+00:00  5.183330  43.4667              28\n",
       "Blagnac       2020-01-01 00:00:00+00:00  1.400000  43.6333              30\n",
       "Lille         2020-01-01 00:00:00+00:00  3.066670  50.6333               6\n",
       "Marseille     2020-01-01 00:00:00+00:00  5.376390  43.2967              45\n",
       "Mérignac      2020-01-01 00:00:00+00:00 -0.633333  44.8333              42"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Dataframe for international travels\n",
    "# Get all rows with NA values\n",
    "df_international_flight = df_flight_airports_locations[df_flight_airports_locations.isna().any(axis=1)]\n",
    "\n",
    "# All na values in the dataframe are replaced by empty Strings\n",
    "df_international_flight = df_international_flight.replace(np.nan, '', regex=True)\n",
    "\n",
    "\n",
    "# Concatenate origin and destination airports to get the french airport involve in the travel\n",
    "df_international_flight['french_city'] = df_international_flight['french_city_origin'] + df_international_flight['french_city_destination']\n",
    "\n",
    "# Remove unwanted columns\n",
    "del df_international_flight['french_city_origin']\n",
    "del df_international_flight['french_city_destination']\n",
    "\n",
    "\n",
    "# Same treatment for origin and destination airport names\n",
    "df_international_flight['airport_name'] = df_international_flight['origin_airport_name'] + df_international_flight['destination_airport_name']\n",
    "del df_international_flight['origin_airport_name']\n",
    "del df_international_flight['destination_airport_name']\n",
    "\n",
    "\n",
    "# Convert lat and lon columns to String to process the same way\n",
    "df_international_flight['lon_origin'] = df_international_flight['lon_origin'].astype(str)\n",
    "df_international_flight['lon_destination'] = df_international_flight['lon_destination'].astype(str)\n",
    "df_international_flight['lat_origin'] = df_international_flight['lat_origin'].astype(str)\n",
    "df_international_flight['lat_destination'] = df_international_flight['lat_destination'].astype(str)\n",
    "\n",
    "# Concatenate lat and lon to get the coordinates of the french airport involve in the travel\n",
    "df_international_flight['lon'] = df_international_flight['lon_origin'] + df_international_flight['lon_destination']\n",
    "df_international_flight['lat'] = df_international_flight['lat_origin'] + df_international_flight['lat_destination']\n",
    "\n",
    "# Remove unwanted columns\n",
    "del df_international_flight['lon_origin']\n",
    "del df_international_flight['lon_destination']\n",
    "del df_international_flight['lat_origin']\n",
    "del df_international_flight['lat_destination']\n",
    "\n",
    "# Reconvert lat and lon columns to floats\n",
    "df_international_flight['lon'] = pd.to_numeric(df_international_flight['lon'],errors='coerce')\n",
    "df_international_flight['lat'] = pd.to_numeric(df_international_flight['lat'],errors='coerce')\n",
    "\n",
    "# Count number of flights at each date and each airport\n",
    "df_international_flight = df_international_flight.groupby(['day', 'french_city', 'lon', 'lat']).size().reset_index(name=\"number_flights\")\n",
    "\n",
    "# Set day and city as index\n",
    "df_international_flight = df_international_flight.set_index(['french_city', 'day'])\n",
    "\n",
    "# Remove duplicate rows\n",
    "df_international_flight = df_international_flight.loc[~df_international_flight.index.duplicated(keep='first')]\n",
    "\n",
    "# Show international dataframe\n",
    "df_international_flight.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* #### c.4: Preparation of dictionaries to store data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" Initialization of variables \"\"\"\n",
    "\n",
    "# Initialization of dictionaries to contain data of each type of travel\n",
    "intranational_by_date_as_integer = {}\n",
    "international_by_date_as_integer = {}\n",
    "\n",
    "\n",
    "# Get list of ordered cities in the dataset\n",
    "airport_cities = matches[1]\n",
    "\n",
    "# Get lon and lat of each city of airport in the dataset\n",
    "lon_city_airport = []\n",
    "lat_city_airport = []\n",
    "for city in airport_cities:\n",
    "    lon_city_airport.append(list(df_cities.loc[df_cities['name'] == city, 'lon'])[0])\n",
    "    lat_city_airport.append(list(df_cities.loc[df_cities['name'] == city, 'lat'])[0])\n",
    "    \n",
    "# Get list of dates in the dataset\n",
    "flights_dates = list(df_flight_airports_locations['day'].unique())\n",
    "\n",
    "# Initialization of list to fill with dates from dataframe as String\n",
    "flights_dates_str = [] \n",
    "\n",
    "# Initizialisation of a counter of dates\n",
    "# We want each date to be provided an unique int id to use if in bokeh later\n",
    "id_date = 0\n",
    "\n",
    "\n",
    "\"\"\" Fill dictionaries  \"\"\"\n",
    "# Create a dictionary with all dates as key, and list of related data for each city\n",
    "for date in flights_dates:\n",
    "    # Initialization\n",
    "    intranational_at_date = []\n",
    "    international_at_date = []\n",
    "    \n",
    "    # Find numbler of flights in dataframe for each date, city and type of flight\n",
    "    for city in airport_cities:\n",
    "        # Fill intranational dictionary\n",
    "        # If a statement is missing on a certain date, its assigned value is 0.\n",
    "        if (city, date) not in df_intranational_flight.index:\n",
    "            intranational_at_date.append(0) \n",
    "        else:\n",
    "            intranational_at_date.append(df_intranational_flight.at[(city, date), 'number_flights'])\n",
    "            \n",
    "        # Fill international dicitonary\n",
    "        # If a statement is missing on a certain date, its assigned value is 0.\n",
    "        if (city, date) not in df_international_flight.index:\n",
    "            international_at_date.append(0) \n",
    "        else:\n",
    "            international_at_date.append(df_international_flight.at[(city, date), 'number_flights'])\n",
    "\n",
    "    # Convert the date in pandas format datetime64 to format datetime\n",
    "    date = convert_datetime64_to_datetime(date)\n",
    "    \n",
    "    # Convert the date in datetime format as a String\n",
    "    date = date.strftime(\"%m/%d/%Y\")\n",
    "    \n",
    "    # Add to the String converted date list\n",
    "    flights_dates_str.append(date)\n",
    "    \n",
    "    # Update data dicitonary per date\n",
    "    intranational_by_date_as_integer[id_date] = intranational_at_date\n",
    "    international_by_date_as_integer[id_date] = international_at_date\n",
    "    \n",
    "    # Increase counter\n",
    "    id_date += 1\n",
    "    \n",
    "    \n",
    "\"\"\" All dictionaries in one \"\"\"\n",
    "flight_by_feature = {}\n",
    "flight_by_feature['intranational'] = intranational_by_date_as_integer\n",
    "flight_by_feature['international'] = international_by_date_as_integer\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" Mapping size of points values \"\"\"\n",
    "\n",
    "# Compute minimum and maximum for all dates and all species\n",
    "intranational_minimum = df_intranational_flight[\"number_flights\"].min()\n",
    "international_minimum = df_international_flight[\"number_flights\"].min()\n",
    "\n",
    "\n",
    "intranational_maximum = df_intranational_flight[\"number_flights\"].max()\n",
    "international_maximum = df_international_flight[\"number_flights\"].max()\n",
    "\n",
    "\"\"\" Linear mapping for the size of each point \"\"\"\n",
    "# We want to map each value in range [minimum, maximum] to a value in range [5, 200]\n",
    "# I chose a minimum size of 5\n",
    "size_airport_city = {}\n",
    "\n",
    "size_airport_city[\"intranational\"] = {}\n",
    "size_airport_city[\"international\"] = {}       \n",
    "                                                \n",
    "for int_date in intranational_by_date_as_integer:\n",
    "    size_airport_city[\"intranational\"][int_date] = []                                                \n",
    "    for old_value in intranational_by_date_as_integer[int_date]:\n",
    "        new_value = old_value # no need of specific mapping as there are not a lot of intrnational flights in france\n",
    "        size_airport_city[\"intranational\"][int_date].append(new_value)\n",
    "\n",
    "for int_date in international_by_date_as_integer:\n",
    "    size_airport_city[\"international\"][int_date] = []\n",
    "    for old_value in international_by_date_as_integer[int_date]:\n",
    "        if old_value == 0:\n",
    "            new_value = 0\n",
    "        else:\n",
    "            new_value = 5 + ((old_value - international_minimum) * (100 - 5)) / (international_maximum - international_minimum)\n",
    "        size_airport_city[\"international\"][int_date].append(new_value)\n",
    "        \n",
    "        "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <span style=\"color:#008000\">II.5 Air quality [data](https://aqicn.org/data-platform/covid19/?fbclid=IwAR1nn_Tk9fLpg7uYIZ9tevK9wbubCfr-B-PH5JnXEjX9G_1SJWroD96tpUA)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <span style=\"color:#008000\">a- General preparation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "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>Country</th>\n",
       "      <th>City</th>\n",
       "      <th>Specie</th>\n",
       "      <th>count</th>\n",
       "      <th>min</th>\n",
       "      <th>max</th>\n",
       "      <th>median</th>\n",
       "      <th>variance</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>557484</td>\n",
       "      <td>2020-01-18</td>\n",
       "      <td>FR</td>\n",
       "      <td>Tours</td>\n",
       "      <td>wind gust</td>\n",
       "      <td>27</td>\n",
       "      <td>1.6</td>\n",
       "      <td>3.0</td>\n",
       "      <td>2.1</td>\n",
       "      <td>2.83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>557485</td>\n",
       "      <td>2020-01-25</td>\n",
       "      <td>FR</td>\n",
       "      <td>Tours</td>\n",
       "      <td>wind gust</td>\n",
       "      <td>3</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>557486</td>\n",
       "      <td>2020-03-01</td>\n",
       "      <td>FR</td>\n",
       "      <td>Tours</td>\n",
       "      <td>wind gust</td>\n",
       "      <td>8</td>\n",
       "      <td>13.3</td>\n",
       "      <td>18.0</td>\n",
       "      <td>13.8</td>\n",
       "      <td>45.54</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>557487</td>\n",
       "      <td>2020-03-05</td>\n",
       "      <td>FR</td>\n",
       "      <td>Tours</td>\n",
       "      <td>wind gust</td>\n",
       "      <td>72</td>\n",
       "      <td>3.3</td>\n",
       "      <td>56.0</td>\n",
       "      <td>16.0</td>\n",
       "      <td>1043.06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>557488</td>\n",
       "      <td>2020-01-28</td>\n",
       "      <td>FR</td>\n",
       "      <td>Tours</td>\n",
       "      <td>wind gust</td>\n",
       "      <td>8</td>\n",
       "      <td>12.8</td>\n",
       "      <td>25.7</td>\n",
       "      <td>15.4</td>\n",
       "      <td>279.71</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             date Country   City     Specie  count   min   max  median  \\\n",
       "557484 2020-01-18      FR  Tours  wind gust     27   1.6   3.0     2.1   \n",
       "557485 2020-01-25      FR  Tours  wind gust      3   2.0   2.0     2.0   \n",
       "557486 2020-03-01      FR  Tours  wind gust      8  13.3  18.0    13.8   \n",
       "557487 2020-03-05      FR  Tours  wind gust     72   3.3  56.0    16.0   \n",
       "557488 2020-01-28      FR  Tours  wind gust      8  12.8  25.7    15.4   \n",
       "\n",
       "        variance  \n",
       "557484      2.83  \n",
       "557485      0.00  \n",
       "557486     45.54  \n",
       "557487   1043.06  \n",
       "557488    279.71  "
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_air = pd.read_csv(\"datasets/waqi-covid19-airqualitydata-2020.csv\")\n",
    "\n",
    "# Consider only measures in France\n",
    "df_air = df_air[df_air.Country == \"FR\"]\n",
    "\n",
    "# Convert string date to timestamp\n",
    "df_air['Date'] = pd.to_datetime(df_air['Date'], format='%Y-%m-%d')\n",
    "df_air.rename(columns={'Date': 'date'}, inplace=True)\n",
    "\n",
    "# Show dataframe\n",
    "df_air.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <span style=\"color:#008000\">b- Preparation for line plots"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_air2020=df_air.groupby(['date','Specie']).mean()['median']\n",
    "\n",
    "# Load dataset\n",
    "df_air2019 =  pd.concat([pd.read_csv(\"datasets/waqi-covid19-airqualitydata-2019Q1.csv\"), # January to March 2019 \n",
    "                        pd.read_csv(\"datasets/waqi-covid19-airqualitydata-2019Q2.csv\")], # April to June 2019 \n",
    "                         ignore_index=True)\n",
    "\n",
    "df_air2019 = df_air2019[df_air2019.Country == \"FR\"]\n",
    "\n",
    "df_air2019.rename(columns={'Date': 'date'}, inplace=True)\n",
    "df_air2019[\"date\"] = pd.to_datetime(df_air2019[\"date\"].apply(lambda x:\"2020\"+x[4:]), format='%Y-%m-%d') #cheat for merging\n",
    "\n",
    "df_air2019=df_air2019.groupby(['Specie','date']).mean()['median']"
   ]
  },
  {
   "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>Specie</th>\n",
       "      <th>2020</th>\n",
       "      <th>2019</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>date</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>2020-01-01</td>\n",
       "      <td>co</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.4375</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2020-01-01</td>\n",
       "      <td>dew</td>\n",
       "      <td>3.0</td>\n",
       "      <td>4.1000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           Specie  2020    2019\n",
       "date                           \n",
       "2020-01-01     co   0.1  0.4375\n",
       "2020-01-01    dew   3.0  4.1000"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_air_merged = pd.merge(df_air2020, df_air2019,\n",
    "                         how='inner', on=['Specie','date'])\n",
    "\n",
    "df_air_merged.rename(columns={'median_x': '2020', \n",
    "                               'median_y': '2019'},\n",
    "                      inplace=True)\n",
    "\n",
    "df_air_merged.reset_index(inplace=True)\n",
    "df_air_merged.set_index('date', inplace = True)\n",
    "\n",
    "df_air_merged.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "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>2020</th>\n",
       "      <th>2019</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>date</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>2020-01-01</td>\n",
       "      <td>11.942308</td>\n",
       "      <td>8.713636</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2020-01-02</td>\n",
       "      <td>12.568000</td>\n",
       "      <td>8.736364</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 2020      2019\n",
       "date                           \n",
       "2020-01-01  11.942308  8.713636\n",
       "2020-01-02  12.568000  8.736364"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_no2=df_air_merged[df_air_merged.Specie=='no2'][['2020','2019']]\n",
    "df_so2=df_air_merged[df_air_merged.Specie=='so2'][['2020','2019']]\n",
    "df_co=df_air_merged[df_air_merged.Specie=='co'][['2020','2019']]\n",
    "df_o3=df_air_merged[df_air_merged.Specie=='o3'][['2020','2019']]\n",
    "df_pm10=df_air_merged[df_air_merged.Specie=='pm10'][['2020','2019']]\n",
    "df_pm25=df_air_merged[df_air_merged.Specie=='pm25'][['2020','2019']]\n",
    "\n",
    "df_no2.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" Initialization\"\"\"\n",
    "\n",
    "# Get list of dates in the dataframe\n",
    "quality_dates_str = df_no2.index.strftime('%m/%d/%Y')\n",
    "\n",
    "# Add a new columns to order rows with integers\n",
    "df_no2['integer'] = [i for i in range(len(df_no2))]\n",
    "df_so2['integer'] = [i for i in range(len(df_so2))]\n",
    "df_o3['integer'] = [i for i in range(len(df_o3))]\n",
    "df_pm10['integer'] = [i for i in range(len(df_pm10))]\n",
    "df_pm25['integer'] = [i for i in range(len(df_pm25))]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <span style=\"color:#008000\">c- Preparation for bokeh maps"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* #### c.1: Prepare dataset data from 2020 and data from 2019 at the same period"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" General function to prepare an air dataset \"\"\"\n",
    "\n",
    "def preparation_air_dataset(df_air):\n",
    "\n",
    "    # Consider only measures in France\n",
    "    df_air = df_air[df_air.Country == \"FR\"]\n",
    "\n",
    "    # Show the differents Species in the dataset\n",
    "    print(\"Here are the species we can find in the dataset:\\n\")\n",
    "    print(df_air[\"Specie\"].unique())\n",
    "    \n",
    "    # We are only interested in the species of pollutants\n",
    "    # Co is not very relevant, it's value is always close to 0\n",
    "    df_air = df_air[df_air['Specie'].isin([\"pm10\", \"pm25\", \"no2\", 'o3', 'so2'])]\n",
    "\n",
    "    # Convert string date to timestamp\n",
    "    df_air['Date'] = pd.to_datetime(df_air['Date'], format='%Y-%m-%d')\n",
    "\n",
    "    # Keep only data we are interested in\n",
    "    df_air = df_air[[\"Date\", \"Specie\", \"City\", \"median\"]]\n",
    "\n",
    "    # Show dataframe\n",
    "    print(\"\\n\\n\\nShow dataframe after the first work of preparation:\\n\")\n",
    "    print(df_air.head())\n",
    "    \n",
    "    # Group by City/Date/Specie and get average median value for each week for each city\n",
    "    df_air_position = df_air\n",
    "    df_air_position['Date'] = pd.to_datetime(df_air_position['Date']) - pd.to_timedelta(7, unit='d')\n",
    "    df_air_position = df_air_position.groupby(['City', pd.Grouper(key='Date', freq='W-MON'), 'Specie'])['median'].mean().reset_index().sort_values('Date')\n",
    "\n",
    "    # Merge dataset with the dataset of french cities to get their position\n",
    "    df_air_position = pd.merge(df_air_position, df_cities, how='left', left_on='City', right_on='name')\n",
    "    del df_air_position[\"name\"]\n",
    "\n",
    "    # Did we found data for all the cities ?\n",
    "    if df_air.shape[0] == df_air_position.shape[0]:\n",
    "        print(\"We found the coordinates for all cities\")\n",
    "        \n",
    "    # Add a column week number\n",
    "    df_air_position['week_number'] = df_air_position['Date'].dt.week\n",
    "\n",
    "    # Round the previously calculated average of medians\n",
    "    df_air_position[\"median\"] = df_air_position[\"median\"].round()\n",
    "\n",
    "    # Set city, date, specie as index\n",
    "    df_air_by_date_city = df_air_position.set_index(['City', 'Date', \"Specie\"])\n",
    "\n",
    "    # Show\n",
    "    print(\"\\n\\n\\nShow dataframe after adding cities data:\\n\")\n",
    "    print(df_air_by_date_city.head())\n",
    "    \n",
    "    return(df_air, df_air_position, df_air_by_date_city)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Here are the species we can find in the dataset:\n",
      "\n",
      "['wind gust' 'wind-gust' 'humidity' 'wind-speed' 'pm10' 'pm25' 'no2' 'dew'\n",
      " 'temperature' 'wind speed' 'pressure' 'co' 'o3' 'so2' 'precipitation']\n",
      "\n",
      "\n",
      "\n",
      "Show dataframe after the first work of preparation:\n",
      "\n",
      "             Date Specie   City  median\n",
      "557687 2020-01-04   pm10  Tours    17.0\n",
      "557688 2020-04-25   pm10  Tours    21.0\n",
      "557689 2020-03-27   pm10  Tours    27.0\n",
      "557690 2020-04-17   pm10  Tours    15.0\n",
      "557691 2020-02-18   pm10  Tours    12.0\n",
      "\n",
      "\n",
      "\n",
      "Show dataframe after adding cities data:\n",
      "\n",
      "                                    median      lon      lat  week_number\n",
      "City             Date       Specie                                       \n",
      "Amiens           2019-12-23 no2       13.0  2.30000  49.9000           52\n",
      "Rouen            2019-12-23 o3         1.0  1.08333  49.4333           52\n",
      "                            no2       19.0  1.08333  49.4333           52\n",
      "Clermont-Ferrand 2019-12-23 no2       19.0  3.08333  45.7833           52\n",
      "                            o3         1.0  3.08333  45.7833           52\n"
     ]
    }
   ],
   "source": [
    "\"\"\" Air dataset during corona  \"\"\"\n",
    "# Load dataset\n",
    "df_air_corona = pd.read_csv(\"datasets/waqi-covid19-airqualitydata-2020.csv\")\n",
    "\n",
    "# Prepare datasets\n",
    "df_air_corona, df_air_position_corona, df_air_by_date_city_corona = preparation_air_dataset(df_air_corona)\n",
    "\n",
    "\n",
    "\"\"\" Generate list of important elements of the corona dataset  \"\"\"\n",
    "\n",
    "# Get list of ordered dates in the dataset\n",
    "air_dates_corona = list(df_air_position_corona['Date'].unique())\n",
    "\n",
    "# Get list of ordered week number in dataset\n",
    "air_week = list(df_air_position_corona['week_number'].unique())\n",
    "\n",
    "# List of pollutants\n",
    "pollutants = list(df_air_corona[\"Specie\"].unique())\n",
    "\n",
    "# Get list of ordered cities in the dataset\n",
    "air_cities = list(df_air_corona[\"City\"].unique())\n",
    "\n",
    "# Get lon and lat of each city in the dataset\n",
    "lon_city = []\n",
    "lat_city = []\n",
    "for city in air_cities:\n",
    "    lon_city.append(list(df_air_position_corona.loc[df_air_position_corona['City'] == city, 'lon'])[0])\n",
    "    lat_city.append(list(df_air_position_corona.loc[df_air_position_corona['City'] == city, 'lat'])[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Here are the species we can find in the dataset:\n",
      "\n",
      "['dew' 'temperature' 'o3' 'wind-gust' 'co' 'pm10' 'humidity' 'pm25' 'no2'\n",
      " 'pressure' 'wind-speed' 'so2' 'precipitation']\n",
      "\n",
      "\n",
      "\n",
      "Show dataframe after the first work of preparation:\n",
      "\n",
      "             Date Specie   City  median\n",
      "141038 2019-01-04     o3  Tours     8.6\n",
      "141039 2019-01-11     o3  Tours     7.1\n",
      "141040 2019-02-01     o3  Tours    25.5\n",
      "141041 2019-02-04     o3  Tours    19.2\n",
      "141042 2019-02-28     o3  Tours    29.1\n",
      "\n",
      "\n",
      "\n",
      "Show dataframe after adding cities data:\n",
      "\n",
      "                          median      lon      lat  week_number\n",
      "City   Date       Specie                                       \n",
      "Amiens 2018-12-24 no2        6.0  2.30000  49.9000           52\n",
      "Lyon   2018-12-24 no2        9.0  4.84139  45.7589           52\n",
      "                  o3        14.0  4.84139  45.7589           52\n",
      "                  pm10       9.0  4.84139  45.7589           52\n",
      "                  pm25      29.0  4.84139  45.7589           52\n",
      "\n",
      "\n",
      "\n",
      "Final 2019 dataframe with same cities and weeks as corona dataset:\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>median</th>\n",
       "      <th>lon</th>\n",
       "      <th>lat</th>\n",
       "      <th>week_number</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>City</th>\n",
       "      <th>Date</th>\n",
       "      <th>Specie</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>Amiens</td>\n",
       "      <td>2018-12-24</td>\n",
       "      <td>no2</td>\n",
       "      <td>6.0</td>\n",
       "      <td>2.30000</td>\n",
       "      <td>49.9000</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td rowspan=\"4\" valign=\"top\">Lyon</td>\n",
       "      <td rowspan=\"4\" valign=\"top\">2018-12-24</td>\n",
       "      <td>no2</td>\n",
       "      <td>9.0</td>\n",
       "      <td>4.84139</td>\n",
       "      <td>45.7589</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>o3</td>\n",
       "      <td>14.0</td>\n",
       "      <td>4.84139</td>\n",
       "      <td>45.7589</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>pm10</td>\n",
       "      <td>9.0</td>\n",
       "      <td>4.84139</td>\n",
       "      <td>45.7589</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>pm25</td>\n",
       "      <td>29.0</td>\n",
       "      <td>4.84139</td>\n",
       "      <td>45.7589</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                          median      lon      lat  week_number\n",
       "City   Date       Specie                                       \n",
       "Amiens 2018-12-24 no2        6.0  2.30000  49.9000           52\n",
       "Lyon   2018-12-24 no2        9.0  4.84139  45.7589           52\n",
       "                  o3        14.0  4.84139  45.7589           52\n",
       "                  pm10       9.0  4.84139  45.7589           52\n",
       "                  pm25      29.0  4.84139  45.7589           52"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\"\"\" Air dataset at the same period in 2019  \"\"\"\n",
    "\n",
    "# Load dataset\n",
    "df_air_2019 =  pd.concat([pd.read_csv(\"datasets/waqi-covid19-airqualitydata-2019Q1.csv\"), # January to March 2019 \n",
    "                        pd.read_csv(\"datasets/waqi-covid19-airqualitydata-2019Q2.csv\")], # April to June 2019 \n",
    "                         ignore_index=True)\n",
    "\n",
    "# Prepare datasets\n",
    "df_air_2019, df_air_position_2019, df_air_by_date_city_2019 = preparation_air_dataset(df_air_2019)\n",
    "\n",
    "# Filter on the cities and week which are in the corona dataset\n",
    "df_air_position_2019 = df_air_position_2019[df_air_position_2019[\"City\"].isin(air_cities)]\n",
    "df_air_position_2019 = df_air_position_2019[df_air_position_2019[\"week_number\"].isin(air_week)]\n",
    "\n",
    "df_air_by_date_city_2019 = df_air_by_date_city_2019[df_air_by_date_city_2019.index.isin(air_cities, level=0)]\n",
    "df_air_by_date_city_2019 = df_air_by_date_city_2019[df_air_by_date_city_2019[\"week_number\"].isin(air_week)]\n",
    "\n",
    "# Get list of ordered dates in the dataset\n",
    "air_dates_2019 = list(df_air_position_2019['Date'].unique())\n",
    "\n",
    "# Show final dataframe\n",
    "print(\"\\n\\n\\nFinal 2019 dataframe with same cities and weeks as corona dataset:\")\n",
    "df_air_by_date_city_2019.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* #### c.2: Preparation of dictionaries to store data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" ---------------------- Function to compute dictionaries used to build the bokeh Maps ---------------------- \"\"\"\n",
    "\n",
    "def dictionaries_for_bokeh_map(df_air_by_date_city, air_dates, air_cities, pollutants):\n",
    "    \n",
    "    \"\"\" Initialization of variables \"\"\"\n",
    "    # Initialization of dictionaries to contain data of each specie for each week, city \n",
    "    pm10_by_date_as_integer = {}\n",
    "    pm25_by_date_as_integer = {}\n",
    "    no2_by_date_as_integer = {}\n",
    "    o3_by_date_as_integer = {}\n",
    "    so2_by_date_as_integer = {}\n",
    "\n",
    "\n",
    "    # Initialization of list to fill with dates from dataframe as String\n",
    "    air_dates_str = [] \n",
    "\n",
    "    # Initizialisation of a counter of dates\n",
    "    # We want each date to be provided an unique int id to use it in a bokeh slider\n",
    "    id_date = 0\n",
    "\n",
    "\n",
    "    \"\"\" Fill dictionaries  \"\"\"\n",
    "    # Create a dictionary with all dates as key, and list of related data for each city\n",
    "    for date in air_dates:\n",
    "\n",
    "        # Initialization\n",
    "        pm10_by_city_at_date = []\n",
    "        pm25_by_city_at_date = []\n",
    "        no2_by_city_at_date = []\n",
    "        o3_by_city_at_date = []\n",
    "        so2_by_city_at_date = []\n",
    "\n",
    "        # Find median value in dataframe for each city and specie\n",
    "        for city in air_cities:\n",
    "            \n",
    "            for specie in pollutants:\n",
    "                # If a statement is missing on a certain date, its assigned value is 0.\n",
    "                if (city, date, specie) not in df_air_by_date_city.index:\n",
    "                    eval(specie + '_by_city_at_date').append(0) # eval allow to retrieve a variable using its name as a String\n",
    "                else:\n",
    "                    eval(specie + '_by_city_at_date').append(df_air_by_date_city.at[(city, date, specie), 'median'])\n",
    "\n",
    "        # Convert the date in pandas format datetime64 to format datetime\n",
    "        date = convert_datetime64_to_datetime(date)\n",
    "\n",
    "        # Convert the date in datetime format as a String\n",
    "        date = date.strftime(\"%m/%d/%Y\")\n",
    "\n",
    "        # Add to the String converted date list\n",
    "        air_dates_str.append(date)\n",
    "\n",
    "        # Update non cumulated data dictionaries\n",
    "        for specie in pollutants:\n",
    "            eval(specie + '_by_date_as_integer')[id_date] = eval(specie + '_by_city_at_date')\n",
    "\n",
    "        # Increase counter\n",
    "        id_date += 1\n",
    "\n",
    "    \"\"\" All dictionaries in one \"\"\"\n",
    "    air_pollution_by_pollutant = {}\n",
    "\n",
    "    for specie in pollutants:\n",
    "        air_pollution_by_pollutant[specie] = eval(specie + '_by_date_as_integer')\n",
    "        \n",
    "    return(air_pollution_by_pollutant, air_dates_str)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" Creation of dictionaries for both corona and 2019 maps \"\"\"\n",
    "air_pollution_by_pollutant_corona, air_dates_str_corona = dictionaries_for_bokeh_map(df_air_by_date_city_corona, air_dates_corona, air_cities, pollutants)\n",
    "air_pollution_by_pollutant_2019, air_dates_str_2019 = dictionaries_for_bokeh_map(df_air_by_date_city_2019, air_dates_2019, air_cities, pollutants)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" Creation of dictionary containing min/max values for the color mapping \"\"\"\n",
    "# Compute min / max\n",
    "pm10_minimum = min(df_air_position_corona[df_air_position_corona[\"Specie\"] == \"pm10\"][\"median\"].min(), df_air_position_2019[df_air_position_2019[\"Specie\"] == \"pm10\"][\"median\"].min())\n",
    "pm25_minimum = min(df_air_position_corona[df_air_position_corona[\"Specie\"] == \"pm25\"][\"median\"].min(), df_air_position_2019[df_air_position_2019[\"Specie\"] == \"pm25\"][\"median\"].min())\n",
    "no2_minimum = min(df_air_position_corona[df_air_position_corona[\"Specie\"] == \"no2\"][\"median\"].min(), df_air_position_2019[df_air_position_2019[\"Specie\"] == \"no2\"][\"median\"].min())\n",
    "o3_minimum = min(df_air_position_corona[df_air_position_corona[\"Specie\"] == \"o3\"][\"median\"].min(), df_air_position_2019[df_air_position_2019[\"Specie\"] == \"o3\"][\"median\"].min())\n",
    "so2_minimum = min(df_air_position_corona[df_air_position_corona[\"Specie\"] == \"so2\"][\"median\"].min(), df_air_position_2019[df_air_position_2019[\"Specie\"] == \"so2\"][\"median\"].min())\n",
    "\n",
    "pm10_maximum = min(df_air_position_corona[df_air_position_corona[\"Specie\"] == \"pm10\"][\"median\"].max(), df_air_position_2019[df_air_position_2019[\"Specie\"] == \"pm10\"][\"median\"].max())\n",
    "pm25_maximum = min(df_air_position_corona[df_air_position_corona[\"Specie\"] == \"pm25\"][\"median\"].max(), df_air_position_2019[df_air_position_2019[\"Specie\"] == \"pm25\"][\"median\"].max())\n",
    "no2_maximum = min(df_air_position_corona[df_air_position_corona[\"Specie\"] == \"no2\"][\"median\"].max(), df_air_position_2019[df_air_position_2019[\"Specie\"] == \"no2\"][\"median\"].max())\n",
    "o3_maximum = min(df_air_position_corona[df_air_position_corona[\"Specie\"] == \"o3\"][\"median\"].max(), df_air_position_2019[df_air_position_2019[\"Specie\"] == \"o3\"][\"median\"].max())\n",
    "so2_maximum = min(df_air_position_corona[df_air_position_corona[\"Specie\"] == \"so2\"][\"median\"].max(), df_air_position_2019[df_air_position_2019[\"Specie\"] == \"so2\"][\"median\"].max())\n",
    "\n",
    "# Fill min / max dictionary\n",
    "air_data_min_max = {}\n",
    "for specie in pollutants:\n",
    "    air_data_min_max[specie] = {\"min\": eval(specie + \"_minimum\"), \"max\": eval(specie + \"_maximum\")}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <span style=\"color:#008000\">II.6 Covid [data](https://github.com/CSSEGISandData/COVID-19/tree/master/csse_covid_19_data/csse_covid_19_time_series) by country"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['US', 'United Kingdom', 'Italy', 'Spain', 'France', 'Brazil', 'Belgium',\n",
       "       'Germany', 'Iran', 'Netherlands'],\n",
       "      dtype='object', name='Country/Region')"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#John Hopkins data \n",
    "df_deaths_global = pd.read_csv(\"datasets/time_series_covid19_deaths_global.txt\")\n",
    "\n",
    "# Group by country and region\n",
    "df_deaths_global_groupby = df_deaths_global.groupby(\"Country/Region\").sum()\n",
    "df_deaths_global_groupby.drop(['Lat', 'Long'], axis=1, inplace=True)\n",
    "\n",
    "#10 most affected countries\n",
    "sorted_countries = df_deaths_global_groupby.iloc[:,-1].sort_values(ascending=False)[:10].index\n",
    "sorted_countries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create dictionnary to store death toll for these 10 countries indexed by days passed since 10th deaths\n",
    "dict_death = {}\n",
    "\n",
    "for i in range(len(sorted_countries)):\n",
    "    df_temp = df_deaths_global_groupby.loc[sorted_countries[i]]\n",
    "    dict_death[sorted_countries[i]]=df_temp[df_temp.values>=400].values.tolist() #day 0 is 400 daths"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='datasets_stats'></a>\n",
    "### **Some dataset stats and key points/plots from the exploratory data analysis**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here are presented some basic statistics of the dataframes that were already pre-processed to keep only the data we are interested in."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* First, a little overview of Covid-related data in France:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# last record date\n",
    "max_date = df_hospital.date.max()\n",
    "\n",
    "# We group by date and department\n",
    "df_hospital_groupby = df_hospital.groupby(['date','department']).sum()\n",
    "\n",
    "# Fundamental distributions of data \n",
    "df_deaths = df_hospital_groupby.loc[max_date].death\n",
    "\n",
    "plt.title(\"Death toll in the 10 most affected departments\", fontsize=16)\n",
    "df_deaths.sort_values(ascending=True)[-10:].plot(kind='barh', rot=0, figsize=(10, 4));"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> Ile de France (the department 75 including Paris) is the most affected territory. It was expected since it is the most populated one."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot of death against hospitalized for each departments\n",
    "plt.figure(figsize=(10, 5))\n",
    "plt.scatter(df_hospital_groupby_department['hospitalized'],\n",
    "            df_hospital_groupby.loc[max(df_hospital_groupby.index.get_level_values(0)) ,:]['death'][:-5],\n",
    "           color='R')\n",
    "plt.title(\"Death versus hospitalized for each department\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> The plot above emphasizes great disparities between French regions. They are not all affected in the same way by the Covid outbreak. <br>\n",
    "Nonetheless a linear correlation can be spotted, which proves the medical response of the regions seemed to be equivalent."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* Let's take a closer look at the energy dataframe:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "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></th>\n",
       "      <th>general_consumption</th>\n",
       "      <th>thermal_power</th>\n",
       "      <th>nuclear_power</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>date</th>\n",
       "      <th>region</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td rowspan=\"5\" valign=\"top\">2020-01-01</td>\n",
       "      <td>Auvergne-Rhône-Alpes</td>\n",
       "      <td>8402.187500</td>\n",
       "      <td>215.270833</td>\n",
       "      <td>11668.833333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Bourgogne-Franche-Comté</td>\n",
       "      <td>2643.291667</td>\n",
       "      <td>173.937500</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Bretagne</td>\n",
       "      <td>2844.937500</td>\n",
       "      <td>235.916667</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Centre-Val de Loire</td>\n",
       "      <td>2502.229167</td>\n",
       "      <td>103.520833</td>\n",
       "      <td>8151.729167</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Grand-Est</td>\n",
       "      <td>5360.645833</td>\n",
       "      <td>941.708333</td>\n",
       "      <td>12030.937500</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                    general_consumption  thermal_power  \\\n",
       "date       region                                                        \n",
       "2020-01-01 Auvergne-Rhône-Alpes             8402.187500     215.270833   \n",
       "           Bourgogne-Franche-Comté          2643.291667     173.937500   \n",
       "           Bretagne                         2844.937500     235.916667   \n",
       "           Centre-Val de Loire              2502.229167     103.520833   \n",
       "           Grand-Est                        5360.645833     941.708333   \n",
       "\n",
       "                                    nuclear_power  \n",
       "date       region                                  \n",
       "2020-01-01 Auvergne-Rhône-Alpes      11668.833333  \n",
       "           Bourgogne-Franche-Comté            NaN  \n",
       "           Bretagne                           NaN  \n",
       "           Centre-Val de Loire        8151.729167  \n",
       "           Grand-Est                 12030.937500  "
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_energy_groupby.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "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": [
    "# Energy consumption over time for the whole country during lockdown\n",
    "df_evolution = df_energy_groupby.sum(level=0)\n",
    "df_evolution.plot(y=[\"general_consumption\", \"thermal_power\", \"nuclear_power\"],\n",
    "                  marker='o', markersize=9, linestyle='--',\n",
    "                  figsize=(12,7))\n",
    "plt.title(\"Energy consumption in France during the lockdown\",fontsize=16);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> We only kept thermal and nuclear energy, which represent the most important source of energy used in France. Energy consumption is cyclical, and often lower at weekends, even during the lockdown. Maybe because some factories are still open during the week, and working from home with connected devices consumes energy."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* Now let's move to flights data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Charles de Gaulle Airport             67461\n",
       "Orly Airport                          42697\n",
       "Lyon–Saint-Exupéry Airport             7150\n",
       "Marseille Provence Airport             5406\n",
       "Toulouse–Blagnac Airport               4156\n",
       "Bordeaux–Mérignac Airport (BA 106)     3077\n",
       "Nantes Atlantique Airport              1530\n",
       "Berre - La Fare Aerodrome (UAF)         840\n",
       "Lille Airport                           579\n",
       "Strasbourg Airport (BA 124)             517\n",
       "Name: origin_airport_name, dtype: int64"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Which are the most used airports?\n",
    "df_flights_airports.origin_airport_name.value_counts(sort=True, ascending=False)[:10]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> Most flights are concentrated in the two airports near Paris: Orly and Charles de Gaulle."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAscAAAFgCAYAAABXB9TlAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nOzdd3gc93Xo/e/M9gIseiHYSXFYJFKFVLGKLVlykRSXJO4ldhI7juzcvL6Jb8qNYzu5N8XxGztxjePe4iLHVZIlW1azZImiKIt9WAE2EL1twda5f8zMYgFsxS4aeT7Po4fkYHZ2dhcQzpw5v3MUwzAQQgghhBBCgLrYJyCEEEIIIcRSIcGxEEIIIYQQFgmOhRBCCCGEsEhwLIQQQgghhEWCYyGEEEIIISwSHAshli1N05TFPodLkbzvQoiLmXOxT0AIsTxomvZq4B7gKsAHHAe+AHxe1/XkIpzPNuCTwG1VHqcB+BZwKzAGrNZ1PVHG474C7NR1/fIKnssAPqDr+sc0Tfsw8Oe6rgfndOLm8d4K/BPQDHwI8/P5qa7r79M07SXAI8AuXdf3lHm8R4Gwrut3F9nnJuD9wO9Ucd7d9nlWcYxHgRcX+HKfrusdcz22EOLSJsGxEKIkTdM+DbwH+BrwWSCMGZh8DLhN07TX67qeXuDTeh1wbQ2O8zbglcA7gGPlBMZVuAHoqeHx/g14AfgwcBIzOLbttZ7vcA2fD+APAa3KY7wWGKnBuTwJ/Hme7fP5GQohLnISHAshitI07e2YQdcf6br++Zwv/ULTtAPAt4E3A19fjPOrgSYgquv6V+f7iXRdf7rGh2wCHtB1/XEATZuKWXVdHwdq/Xw1oev68zU61Og8vKdCiEucBMdCiFI+AOybERgDoOv6dzRN2wUM2ds0TVsLfBQzs+wHfolZPnDM+vqHmVFOoGnalcDzwK26rj9qlSwEgSeA/wm0YQZ69+i6ftg6xoesxxrAO3Vd/0q+k9c07bXAXwNbgWHgK8Df6bqezL01bx3nI7qufzjPMZzA/8HMLvsxy0kcM/apB/4eeA3QiVmicT/wp7quj+Y8xwd0Xf/YjMf+CfBxoEvX9b6c7Z8HrtZ1feeM/V+CWTIB8FFN0z6q67pSYJ9sWYWmafdgvp9dmFnXr1vvxzpd17uth6qapv098C6gDngY88Ko1/pcfi/ntdif1weAPwJWAuesY/5fXdczM99L67HdTJV/vAPzDsQbgP8f2AKcAP5S1/Uf53t8uXKO/c/AX2B+n16F+buvnM/qncArgLuAOPANzO/dlLWPD/g74E1AA7AP+Atd15+wvu4E/hbz+6YNOGB9/eFqXpcQYn7JgjwhREGapnUCl2MGDnnpuv7nuq7fb+2/EtgNXIaZbX4nsA74laZpKyp8+tsxA7E/Bd5qHfMr1te+AHwRiGGWDtxX4PzfDfw38CzmrfxPYt6G/7K1yz0zjvOFAufyCeB/YNb3vgnYgRnM5foW8GrgL4GXYQZlbwY+WMZr/S8gA7w+59zdwO+SPyNvl0xgvaYb8uwzjfVefAr4AWZQeAL4jzy7vtw63jswX/Ot1uPADCjvxyzhuAHYq2naG6zt/2o99gvARzCD63LVAV8CPg3cDQwC39E0ranE4xRN05wz/5uxTwPm9+FbgP+t63qU8j+rTwADmO/XpzG/F3Nf17eBd2NeDL4G6AMe0DRto/X1/wT+DLP85TXAEevrLyrxuoQQi0gyx0KIYlZaf5ZbJ/t+zMV6d+i6PgjZhVMnMYOEP6vgueuAu3Rd77WO0wX8m6Zpzbqun9U07SyQKXRbXdM0FTPb+21d1+1a3Ic0TRsDPmdlW/eVcZwmzHrr/63r+iesbQ+T855omuYF3MB7dF3/mbX5USsIKrRoLEvX9UFN0+7HDNA+aW2+03oPvp1n/3HgaauM4nSZpQV/C3xN1/UPWP9+0Lpgmbn4bhR4lRVEomnaDsyLE3RdP6Fp2gCwxn5OTdNuAbqBz+q6bgCPaZqWBM6XcU42N2ZG/bvWMfswa6lvBb5f5HF3ArMWg2qa1mp//2Fm+P9W1/UHra9V8lk9pev6n1h/f1jTtN+ynvOz1vvyKuDtuq5/3Tr245h3QG60gvR3AO/Sdd2+6PqZdcH5f6hyIakQYv5IcCyEKMZeZFfuXaZbgEdyAhM78HuYMoLEGXrswNhy1vozQE4ZRxFbgFbgezO2/xfwOetc95VxnOswA6wH7A26rk9awewu+9+YGUi7rGQTZsZ9KzBZxnOAudjx+5qmrdN1/RRmQPpQbpnFXGmadhlmKcUPZ3zpe8wOjl+wA2NLN2b2tZCnMDPwz2qadi9mucTHiuxfSG6An/tZF/MrzAuymUZn/PuQ/ZcKP6uZFx1nMct9AOzs709yjp0AtlnHfo+1+f4Z2ez7gX/UNM09z4s/hRBzJMGxEKKY09afqwvtYGXC+qz60kbgN3l268MKGioQnfFvu3613EC9Mee5s3RdH9c0bRKor/A4gzO2X8j9h6Zpr8KsG15v7bsH8zU4KM9PMWui32h1B7kLsxygFlqsPwdmbM8XeOd73wv2NdZ1/ZtW8Pde4B8wA78XgDfpul5Jp4zc5y33sx4rs01df+4/Kvis8r0X9jk1AUm7RjmPZuvPcwW+3kJl2XUhxAKRmmMhREFWBvh5zEVJhfwC+Ln192GgPc8+HUxlew1m/79nzr1+ixi2/px2PlZfYy/lZZ/J2a9txnY7+LEzs9/DXLy2Stf1Vl3XXwno5Z6slUX8Nmad8d2Y5QI/KvfxJdhBWOuM7TP/PSe6rn9V1/VrMT/nP8QMHP+7FseutVp8VpYxwKVpWmjG8W/QNG2z9XUDM8O8K89/My+2hBBLhATHQohSPgHs0DTtD2Z+wRpCsRX4prXpV8Ctmqa15OzTArwUszsCwDjgs4JU281zOK9SfZV1zADkdTO22wvpnqQ8v8bsVPDb9gYrU3pHzj5XY9ax/pOu62etfQLATRTJuubxNetY7wHu1XU9VsFjizmDWR7xqhnbXz2HY0173zVN+4JVToGu6/26rn8Rc5FjwbsNi6xWn9VT1p/ZshRrEeV3gbdj/iwoQJ2u63vs/zAXmr4fSFX7QoQQ80PKKoQQpXwd8xb/5zVNuw4zm5nB7ExwD2YwYHd/+DjmIqSfW+3AFOBvMIcyfMLa5wHMzgZf1DTtU8CVTB9eUa5RwK+Zk/t2z6hPRtf1tKZpHwE+qWnasHXe2zE7KXxP1/UD5TyJVYbxL8BfWuUYe4E/xsySnrB2ex4zaPxnTdM+i3nL/M+tfeLlviBd15/RNO0I5sXCh8p9XBnHzVifx39qmtaPmTW9E7ODB0yVMZRjFFipadodmOUIjwFf0zTtHzDvIKzCfH+WZOaY2n1WezVN+ynm91c95sTI92DWSf+Hrus9mqZ9H/iG1XrwMPASzJ+HjxZqcyeEWHySORZCFGV1IHgTZgC7HTO7+V3MAO5PgLdY+6Dr+hlr+3lrvy9idnW4wc7S6bp+BPPW+9WYgfJrMEsJKvVt4DnMW+RvK3DunwL+ALPrwU+A92H20n1Lhc/1t5hT6O7B7J4wBmT7Puu6fhQzW7gdc8HVRzEDx3uA1RW2sfsZ5sKvxyo8x6J0Xf8SZhD4Jsz65m2YXRPAnHhYrv/ArFX+KfAyq1PDn2AG2vZrvxczQF5yavxZvQHz4vFDmC3ymoCX6rpudzJ5C+aF419hfq5vwmwf99fVvxIhxHxRDMNY7HMQQghh0TTtIPADXdf/psbHfTPwa6sThr3tHzAHfDQXfqQQQlxapKxCCCEWmaZpCmZ2+mrMDgr5hnNU6/eBv7Ju8Q9iLgr7/4B/mYfnEkKIZUvKKoQQYpFZZSmvw+xs8IdWeUqtvRWz3vbTwIOYpS0fxCwXEUIIYZGyCiGEEEIIISzLoazCg3n7r5fSrZuEEEIIIYQoxAF0As9SoEPNcgiOdwFPLPZJCCGEEEKIi8bNmP3IZ1kOwXEvwMhIhEymNiUgzc1BhoZKdy4qd7+FOJeFsJTOBZbW+Syl74VL7X0pl5xL/ueV9yW/pXQusLTOZyn9v2g5vi8L4WI+l2qOV85jVVWhsTEAVnyZz3IIjtMAmYxRs+DYPl4t96vGQjxHuZbSucDSOp+l9L1wqb0v5ZJzyf+88r7kt5TOBZbW+Syl/xctx/dlIVzM51LN8Sp4bMFSXelWIYQQQgghhEWCYyGEEEIIISwSHAshhBBCCGGR4FgIIYQQQgiLBMdCCCGEEEJYJDgWQgghhBDCIsGxEEIIIYQQFgmOhRBCCCGEsEhwvED6R5N889FBYonMYp+KEEIIIYQoQILjBXLfnlEe3DvOl38+gGEsnak2QgghhBBiigTHCyCdMdhzLELQq/K0HuGX+ybm5XkMw6CnPz4vxxZCCCGEuBRIcLwADp2OMRHL8M47Wtmxzsc3Hx3k5IXaB7HPHY/ywW+c4+xgoubHFkIIIYS4FEhwvACe0SN43Qo71vl49yvaCPmdfOqnfUQm0zV9nmPnJwE4PyzBsRBCCCHEXEhwPM9SaYM9xyNcvSGA26lS53PwvrvbGAmn+PyDta0/PtVnZqMHxlI1O6YQQgghxKVEguN5dqAnRjSe4TotkN22odPLm25p5vkTUb7/xEBNnidjGHRbwfHguATHQgghhBBzIcHxPHtGD+P3qFyxxj9t+x1X1bPrsgBffrCX8Wj15RUXRpJMJs0s9MBYsurjCSGEEEJciiQ4nkeJVIbnTkS4ZqMfp0OZ9jVFUbhhc5BMBobD1Wd6T1kL/DoaXQxI5lgIIYQQYk4kOJ5H+7tjTCYMrteCeb8e9Jlvf2Sy+sEgp/riuJ0K29f6GBxLSS9lIYQQQog5kOB4Hj2jh6nzqWxd7cv79aDXAUA4Vn1Zxam+OGvbPbQ1uEimDcZqUKohhBBCCHGpkeB4nsSTGZ4/GWXnZQEcqpJ3n6C3NpnjdMagpz/BunY3rSEnAIPSsUIIIYQQomISHM+TF05FiScNrtuUv6QCIGBnjqvsd3x+KEkiZbCu3UNrvQtA6o6FEEIIIebAudgncLF6+kiYUMDB5pXegvu4nApet0q4yszxyT5z+Me6Dg+NAStzPC4dK4QQQgghKiWZ43kQS2R44VSMXZcFUAuUVNjqfI6qa45P9SXwuRXaG1x43Sp1PlUGgQghhBBCzIEEx/Ngz7EIybTBdQW6VOSq8zuqrjnuvmAuxlMVMxBvrXdJcHwR+M7jQ+zvji72aQghhBCXFAmOayxjGNy/Z5SuZheXrfCU3L/O76yq5jiVNjg9GGdd+9RztYScUlaxzEUm09y3Z4z//vXIYp+KEEIIcUmR4LjGnj8R5dxQkt+6tjGbyS2m3u+oqub4zGCCVBrWd0wFx631TgbHU2Qy0uu4WpHJNH/91TP8+tDYgj7vmYEEACd64/SPyoWOEEIIsVAkOK4hwzD48TOjtIWcXKcFynpMnc9ZVc3xqT5zMl5u5rg15CKdgdGI9Dqu1v17xjg7lOSFE+EFfd5uKzgGeFpf2OcWQgghLmUSHNfQgZ4Yp/ri3H1tQ8HexjPZNcdznWh36kKcgFelpX6q8UiL1et4YEwyjtUYjaR4cK+ZMe4ditfsuN94ZJBDp2NF9+npjxMKONC6vDx1OCwTD4UQQogFIsFxDf34mVEagw5u3FJX9mPq/U4yBsQScwyO+8x6YyWnhKO13m7nJovyqvHjp0dJZwzWtrk5P5wo/YAyJFMGDz0/zi/3jRfd73R/gjWtbm7YEuT8cJLTA7V5fiGEEEIUJ8FxjejnJtHPTXLnzgZczvKyxmBmjsGsba1UIpnh7GBiWr0xQHO9nTmW4HiuBsaSPLJ/nFsur2Prah8XhhM1qeGesEpojvdOFtwnmTI4P5xgTZuHXZcFcKjw6yNSWiGEEEIshLKGgGia9iHg9dY/79N1/X9pmnY78K+AD/iOrut/Y+17JfAFoB54HHiPruspTdNWA98A2gAdeIuuXzzFlD95ZoQ6n8pLrig/awxTwfFELENrqLLnPD2QIGNMrzcGcDtVGgIOBqRjxZz991MjqIrCq69r5IVTUVJpg+Fwelr5ylzYnUmGJ9IMT6Roqpt9vLNDCdIZWNPmps7n4Iq1fp4+Eub1NzeVtchTCCGEEHNXMnNsBcEvA64CrgSu0TTtTcCXgFcDW4Bdmqa90nrIN4D36bq+CVCAd1nbPwN8Rtf1zcAe4IO1fCGLqbsvzr7uGK+4JoTHVVkyvs5nBkdzaeeWbzGerTXkZFAyx3NydjDBU4fD3HFVPU11TtoazJHctegaMR6d+pwLZY9P95uf65pW83O9YXOQ4XCao2cLZ5uFEEIIURvlRHK9wJ/pup7QdT0JHAY2Acd0XT+l63oKMyB+naZpawCfrutPW4/9irXdBdwC3Ju7vXYvY3H9ZPcofo/KS3dUmPrFbOUGzGkQyKk+c9FWY9Ax62st9S4GpOZ4Tu59chivW+GuXQ0AtFkLHPtqEBxPxKY+52Pn8y/y6+lP4HUrtDaYz3v1Bj8elyKlFUIIIcQCKBkc67p+0A52NU27DLO8IoMZNNt6gZXAigLbW4BxK5DO3b7snRtKsOdYhNuvrMfvqbyE2y6rmEs7t3yL8WytISfDEynS0uu4IsfPT7L3RJQ7dzZQ5zM/m+Y6J06HQn8Nun/YNccrmlwcP58/E9wzEGd169TEQ49L5ZoNAXYfjZBKy+cphBBCzKeyozlN07YBPwc+AJwEcn9LK5gBs1rmdqzty94Lp6IYwO1X1s/p8XZZRaWZ48lEhvNDybwlFWB2rMgYMDwh2eNKfO/JYer9Dl5+9dRdAFVVaG900z9a/Xs5EU2jADvW+enuj5NITf/cMxmD0wNmp4pcN2wOEoln2CfjpBdELJFhLCI/O0IIcSkqKzjWNO1G4GHgL3Vd/ypwFujM2aUDOF9kez8Q0jTNvv/faW1f9oYnUnhdCg2BuS3UcjgUfG6l4prj7v44BvnrjQFaQmad7GJ1rIglMnzhwYElE2CUk0GPTKY5fGaS23fU43VP/9HobHLXpqxiMkPQp7Kpy0s6A91901u09Y0liScN1rRN/1y3rfFR51OltGKBfP2Xg/zTvb2ldxRCCHHRKWdB3irgh8CbdV3/trX5GfNL2kYr4H0z8ICu6z3ApBVMA7zN2p4EngDeYG1/O/BADV/HohmNpGkMVtfBIOirfIT0c8ejOFTY2Fk4cwwsWseK/d1RHj84wcESwy4WwkQszXs/28NzxyNF97Oz9815OlKsaPbQP5asehjHRDRN0OfgshVeYPaivJ5+M1he0zY9c+x0KFy7KcjzJ6JE45f25MOJWHreL7qOnpvk/FCSRPKiuMElhBCiAuVkjv8c8AL/qmnabzRN+w3wDuu/7wOHgCNMLbZ7C/BxTdOOAEHg363t9wDv1jTtEHAz8Dc1eg2LaiScyrsgrhIBr1pRzXE6Y/D0kTBXrvcT9OV/7qY6J4qyeJnj473mYrOR8OIHcgd6YkTjGXqHi18oRONmIJSvdryz2c1kwpi2oG4uJmJp6n0O6v0O2kLOWXXHPf1xHCp0NbtnPfaGzUESKYNfHxyr6hyWu/94oJ+//ea5qsauFxOZTNM/lsIALtTgbkG1Ygmzn7kQQoiFUTLlqev6nwJ/WuDLO/Ls/wJwbZ7tPcBLKjy/JW8knGZTl7eqYwS9jopqjg/0xBiLprlpa+Geyk6HQlPQuWhT8o5ZQd9IePHLKg70mHW6sUTx99gOjn3uPMFxk5mh7xtNZjuMzMVELE1no1nysnGFl4OnYxiGkV1U2dOfoKvZjdMxe5HlxhUe6v0OfnMizPZVjXM+h+XMMAyO98aJxjN85eFB3ntXW94FqdWwWyQC9A4nWd2a/+7MQvnWo0M8rYf57D1r835fCCGEqC2ZkFcFwzBqkjkOetWKao6fPDRBwKuyY52/6H4t9U4GatBhoVKJVIbuvqWROTYMg4M9ZmlHJF48OLaDZ7939o/FCiuTW22v44lYJtsFY2Onh7FIOnsBYxgGPf3xWSUVNlVRWNns4sxA/hZwl4KBsRTReIbVrW52H43w5OHa12CfyqkD7x1Z3MxxLJ7h10fCxJNGTWrehRBClCbBcRUmYhnSGaqvOfaWX3McjWd47niU67VgySxSa2hxMsc9/eaEN4e6+Jnj3pEkw1aAHi1xAZItq8iTOW5vcqMA/VWUqWQMg3AsnW3fZ9cd2/2OhydSTMQysxbj5epscnOmf7Lq2uflqtsakPL7d7SgdXn52i8Ha34B2N0Xpy3kpKXeWbIUZ749rYdJpMzP+tyQlFYIIcRCkOC4Cnbg1xiovuY4OpkhU0ZHhWePhUmmDW7aGiy5b2vIxWg4TTK1sIGUXVKxdZVv0TPHB6yscZ1PLb+sIk/Nsdup0ljnqCp7F53MkDHIZo5XtrjxuJTsorwT581zndnGLVdnk4toPMNYpDbv62gkVfJ9AYgnM0ticVp3XwKHCqtaPPzRK1tRFYXPPdBf037ep/rirOvw0Nnoond4cQPSxw5M0NnoQlGQumMhhFggEhxXYdQK/GrRrcJgKjgr5slDYToaXazvKF0H2VLvxAAGJxY2+3Xs/CRtISdr2jyMRlJkFjHLebAnRlvIyapWT8n3N1ZkQR5Ae8hVVVmFPQCkzmce36EqbOjwZBfl2cFxsRrXFU1mvfL5GmQ0M4bBR751ns/e3198v4zBh791jns+28MnfnSBJw5OZF/LQuvuj7OyxY3LqdBS7+Ltt7Vw7Hycn+4ercnxJ2Jmmcu6dg+dTS56R6rvUDJXZwYSnLwQ57Yd9bSFXBIcCyHEApHguArDdua4rtqyCvNjKFVaMTCW5MjZSW7aGixrEVKrNfZ4cAE7VtgLpi5b4aUx6CCdMduXLYZU2uDw2RiXr/Hhdyslg+NIPIPHpeBQ87+3bQ2uqsoqxq1OF3U5HUY2rvByeiBBPJnhRG+M9gZn3sy1bUWTmVU+X4OM5oneOEMTKX5zMsqZgcLH23M8wrmhJNvW+Ojui/OfDw7wvs/18I/fPc+BUwvXd9kwDLr74qzNKTt50ZYg12sBfvDrEfQz1Q9IsRfjrW3z0NnkJp40Fu3ux2MHxnE6zNe4ssXFuSGpORZCiIUgwXEVRsIpFCBURfcCMGuOwWwhVcxT1uKjF20pXVIB0FJvDQJZwLrjwfEUY5E0G1d4sxn1xQouTlyIM5kw2LbGj9/jKCtzXGwEeFuDi/FouqwyhHzC2cxxTnDc6SVjwKkLcU6cj5XsjNAYdOBzqzWphd1zLIJDBbdT4f49+TOvhmFw37NjtDe4+NPfaufj71rNR97SxW9d20DvSJJP/vBs1edRrqGJNOHJDGtnDL75vZe20BB08KkanIu9kHRtuyfbVaQWFyKVSqQyPHk4zM6NAep8Drqa3VwYTc6aqCiEEKL2JDiuwmgkTb3fUXV7pYCdOS7SQ9cwDJ48HGbzSm826C2lKejAoZafOZ6IVR9E2/XGl63wZLt4LNaivIM9URQFtq7y4vOopYPjRPHguL3BDPbnWloxbmXQ63IupjZYQ1xe6I5yYThRsFOFTVEUVrZ6qi6rMAyD545H2Lbax0uuqONpPZx38eahM5Oc6otz584QqqqgKArr2j387o1NvOq6Bk73x0ve7j9yNsb9e0Z5dP84u4+GOdAT5eSFOPEKa5h7+u2s7vT3KOB1cNfOBo6fj3G6yk4epy7E6Wh04feo2RKWxViUt/d4lMhkhhdfYY6lX9nixjDgwiJ3zxBCiEuBBMdVqEUbNyA7yKNYO7eTF+JcGElyYxkL8WyqqtBc5yxrSt7JC3He+PcHOTFjYluljp+P43UprGx2L3rm+EBPjPXtHgJeBwGPSjxpFF24FYln8vY4trU1mMHSXINj+/Oty2kVV+dz0Nno4vEDEwBFO1XYVrd5q14odmYwQf9Yims2BnjFNSEMAx7cO3u4yH3PjhLyO/J+3+26LICqwDN64dKKVNrgkz/p49uPD/Olnw/yqZ/289HvX+DD3zrH50rUOs90qi+OqsCqltkXENdvNru3PHmoujKPU/3x7Ej2UMCB160sSju3Rw9M0BpysmWV2dFkpdVKUOqOhRBi/klwXIWRcJqGKhfjQU7muEjN8a8OhXE5FK69rPzgGKAl5Corc9zTHydjmGOpq3Gsd5L1nV5UVSEUcKAoU7XZCykaz3DyQpxta3zA1CK7YtnjUmUV7SErOJ5j3fF41KxpdrumP8fGFZ7s5L1SmWOAla0ehsNzL+8As6RCAa7ZGKCl3sX1WpBH949PK+3p7otzoCfGy68O4XbOfl9CASdXrA/yzNFIwUVre09EmIhl+JO72/j4u1bzf9+2kv/9+k62rvJypsJAr6c/zoom16z3D8yLjF1aHU8dDs+5c8VYJMXwRJp17eZnoCgKKxrdC5457h9Ncuh0jFu21aFaaws6Gl04VKTuWAghFoAEx1WoVeY44FFRKFxznEobPK2HuWajv+hirXxa651l1Rzb2dAXTs09OJ5MZDg9kOAyq1TAoSqE/I5sV4+FdPhMjIwBl1vBsa+M4DgaTxcNjn0elTqfOud2bhOx9LR6Y9vGTjM72Bh00hAofbG1yqpLriZoe+54lE1d3uy0vzt3hYgnDR5+YTy7z317RvG5FW7bUV/wOLdc0cCFkWTBQPfR/RM01Tm4ZmOA5jonq1rdaCt9bOj0MjieqiiQ7e5PzKo3zvXSq5sYi6az7fsq1W0N/8h9js6mhW/n9vjBCRQFbt42NQHT6VDoaJSOFUIIsRAkOJ6jZMpgIpapuo0bmOUPfq9asOb4YE+MyGSGG4uMiy6kNeRkPJouWd9pZ0PPDCYYnphbZvTkhTiGMTXcAsyAbzFqjg/0xPC4lGzgaQe9saLBcabkxUdbw9zbuYVjaerzBMf2+7Vhha+s46xqM/efa9BmB7M7Lwtkt61u9bB9rY+Hnh8nkcrQN5pk99EIt22vL3rB8KJtIau0IjLrawNjSQ72WBnQGR1A2hqcZAwYKnOx6EjYXOi5tkjZyS6tjoBX5VeHJso65kyn+uIoTC9t6WxyMRxOM1lFlr4S6YzBEwcn2L7WT9OMLjhdzW4ZBCKEEAtAguM5Grw2+bEAACAASURBVI3UZgCIrdgIaTsrtykn6CxXtmNFiVKA/rEk7Y3m7eR93XPLHtvDLOxFZmB2V1is4HjzSl92sWSgRObYMIySC/LA6nU817KKWJqgL99oahetISdXbiyvZKazyY2qzL3X8XPHzUD2mo2Badvv3NnAeDTNk4fCPLBnFIcKL786VPRYDUEnW1f7eEYPzyqtsOuob7l89kXdVIlKea/Bnoy3pkjm2O1UuUELWovZKr9bcaovTmeTa1rdud2xYqEWwu3vjjESTvOSK2a/Zyub3fSPpSpeyCiEEKIyEhzP0UiNBoDYio2Q7htNUu93VFxSAVMdFoqVAhiGQf9okl1aHU11jjmXVhw7P0lXs4uAd+qCwcwcL2xZxeB4kr7RZLakAkqXVSRTBql04QEgtrYGJ0PjqTlNHZyIZqZ1qrCpisK/vHMVv31Ta1nHcTlV2hvmfrt/z7EI69o9tNRP/97dssrLunYPP9k9yhMHw9y4pa6smvrrNgXoH0vR3T91PumMweMHJ7hirS9vd5WpxY3lXWj09CXMrG6R6YEAN24Nkkwb7D46O5NdSnff1GI8W2e2r/TCBMcPvzBOyO9gxzr/rK91WQsRpe5YCCHmlwTHc5TNHNcoOA54VSJFgmM7yK2U/cu9WH1qeDJDLGHQ2eRhxzo/B0/HSKUrC/4y1vAPu4zB1hh0EIkv7Ohhu+Y0NzgutSAvmrCn4xW/E9De4DKnDpbRAWSmiQJlFUC2TVq57OltlRqeSHHiQpxrNs4OvhRF4a5dIQbHU6TSBnfuKp41tl2zMYBDhd05XSvsDKjdimymhqADl0Mpu377VH+cjiYX3iLdRADWd5iT7Z48XFnXipFwipFIelZw3N5gjm5eiLrjvpEk+05FuXV7Xd72kCtbzAsKKa0QQoj5JcHxHA1nM8c1KqvwOQqWVZjBcXm9jWfye1RCfge9I4V/odo1tJ3Nbrav9TOZMDh6rrKWbhdGkkQmM2xcMT24WIx2bgd6YjQGHNk+tZAbHOc/D7sWuVgrN6g842mLJzMkUka2bV+1VjS5uTCSrLgzg11SkVtvnGvnxgBdzS6u3xyks7F05wwwv3e3rfZN61rx6P5x6v0Orlo/OwgHM1veGnKWXVbRM2MyXiGKonDT1iBHz01WtHDSHv6xbsZYdpdTobXeuSDt3H7xwhiqCrdtz39B0R5y4XIosihPCCHmmQTHczQykcLlULJt2KoVLJA5jiczjITTdDTOLTgGsw1UsZpJux65o8nNttU+HGrldcdTwz9mZ45h4QaBZAyDQ6djbFvjm5aJtYPegpnjuJ05LhEch0qXqeQzYU3Hq89TczwXnU0u0pnya3Zte45HWNHkyo6hnklVFT7yli7e9fLySjxs12lBBsdTnLwQZzRsjqS+eWuw6IAcc3Fj6e+L8Wia4XCate3lBesv2lKHAjxZwcK8U31xFAVW5ynb6Gwq3M7tob1j/NVXz/DlXwzw9JEwo3P8Po8lMjx+YIJrNwULlrKoqkJnk0syx0IIMc8kOJ6j0UiaxqCjolvhxQS95njjmZlAO3hom2PmGOx2VIWDKDvA6mxy43WraCt9FdcdHz8fJ+BVZwXxduZ4uILMcTpj8L1fDc8p0DjRGyc8meGKtdMzlqqq4HMrVQfH9X4HHpdScVBq9zHOV3M8F3ZW/HwF9acTsTRHzk4WzBrb3E614qmPV2/w41DhmaMRnjg0QcagYEmFrc3KHBfqkWzLjnQuI3MM0FznZMtqH08enr1IsJBTfXG6mlx48vRQXtFkXlxmZvxsptIGP9k9SmQyw6+PhPnM/f38j8+f5i++fIYfPTlQ1vPanjw0QSxh8LKrir9nK5vdkjkWQoh5JsHxHJk9jmtTbwxTg0BmZo8vWBnKjmqC40YX4clMNns5U/9oypoGZgZuO9b5ODeUzDtOuJDjvZNs7PRkhxbY7PfIrtEuR3dfnJ/sHuXpw7MntpWy+6g5LOXKPAuaio2QLjc4VhSF9gYXfRWWVUzYo6NrVFZRTi35THtPRDAMs3Si1gJeB1es9bNbD/PY/gm2rPSWvNvRFnIRTxrZC4dCsp0qygyOAW7eGmRgLFVWeZBhGJzqS8wqqbB1NrlIpg2GZrQ4fOFUlLFomnfc3sJn71nLh9/cxRtvacLjUvj8/efLHtKSMQx+/ptx1nd42NBZvCNNV4ub4XC65Ch0IYQQcyfB8RyNhFM01KjeGAqPkLbrgedacwzQ0VS8HVX/WDJbLgCw3cq6lltaEZlMc24oOaukAsDnVvC4lIrKKuwyj0pLMTKGwbNHI1yx1pe3s0fA4yjY59gONsrpCNJWQa2szb4wqVVw7PeoNASK15LnMgyDZ/QILfXOsqbwzcV1mwIMh9P0j6V4cZ5WZDPZd0NKlaj09Cdob3CWvHDJdc3GAB6XwhMHSy/MGwmnGY/OXoxns9u5zexY8fiBCUIBs7OEQ1VY3+Hhzp0NvPGWZjIZ0M+WV7d/sCdG73CSO64snjUGWNksi/KEEGK+SXBcwLmhBF984DyZPLdlDcNgJJxemMzxyNzbuNlWNBbPMprB8VTwvaLJRUu9s+zSit+cNPfTVs4eYqEoSsXt3AasThAjE5UFoCd64wyH01y7KX9m1F8kcxxLlJc5BjOoGxibfZu9mKnguHY/ciuaXGWVVcSTGT59Xz8HemK8+PK6mpUCzXT1hoBZh+9RS5ZugNkWDyg5VKW7P15R1hjA61a5cUuQJw5NlMwen7IX4xUKju0sfc7F5Wg4xQunoty0NYhjxoCTjZ0eXE6FQ2fKm9T30PNm+7ZrN5Xuc73SaucmpRVCCDF/JDguoLsvzr2PD+QNKKNxs/NArTpVgFlzDOYUtVz9VbRxs7WEnDhU8mYZkymDkYk0rTmZY0VR2L7Oz6HTsbL6+T66f4L2BiebVuQPLiodBGJnjiud1GeXVFy1Pn9g5vOo2ZZtM0XjGRQFvK7SgWN7g4tUenoddcYwODuYIJHKf/yJWAaHWl7wXa7OJje9I8VrdgfGkvzdf53n2WMR3nhLE6+6rqFmzz+Tz6Py2hsaef3NTbidpV9na70LBYoOVQnH0gyMpcquN871hpubaa138tn7+4sOBTl5IY6qwKqW/Bn1Op9KwKtOa+f2xKEwGSP/gBO3S2Xr6gCHTpcOjnPbt7mcpb/3muuduJ2KZI6FEGIeSXBcgF1/eKI3Putr2QEggdpljoNW5njmIJALo8mqFuMBOFTFGhoxO9AfGE9iMLtsY8c6H/GkgX6u+C/43uEE+rlJXnx5fcGMZMWZY7usooLMcamSCiieOY7GM/jdallZ1dxygGPnJ/nmo4P8z/88zV9/7SwP7s1fJz0RSxP01W4BJ5i1sNF4hrFI/vf28JkYH/rmOYYmUvzZazq4c2fDvGWNbXdf28CtBVqRzeRyKjTWOYpmjnsGzCCw3E4VuXwelT++s43RSIov/Xww70XEgZ4oD+4dY1OXF3eexXhgXix2Nk79/BiGweMHJtC6vAXb3e3YEOT0QKJgnb+tVPu2mVRFMRflySAQIYSYNxIcF9DR6CLgVTlxYfYtWTsLWtPMcZ6a42wbtyqDYzDrJvP1ah0o0A1j6yofTgfs6y4eHD+6fwKHCjdvK3xL2M4c5ytRyWdgzC6rKD9zXKqkAqzguMCglVg8U3bpil2f/fEfXuDvv32eh18YZ027h5DfQXdf/ozeRLTwAJC5ynasyHPR88i+cf753l7q/A4+9OYVbM+zQHEpMMdxFw707E4VlZZV2DZ0evndG5t49liER/dPb+32/MkIH/9hH+0NLt57V1vR4+R2fNGtHsr5ssa2HRvMn4cjRUorymnflk9Xi4tzUlYhhBDzRoLjAlRFYdNKf97M8WiktqOjwVy4pirTa47tDGo1i/FsHU0u+kdnD43oswKT3LIKAI9LZXOJlm7JlMGvDk1w9YYAoSJZ9Magk3QGwiW6EgBkMlNdASopxShVUgFWcJzI5M0gRuOZsksemuucbF3lZdsaH3/0ilY+9Z61vP/VHWzs9HC2wO3u8ViaYA3rjYFsr+KZFz0neif5ysODXL7Gx4ff1FX2MI/FUKrXcXd/nJZ6Z1ULGV+5M8S21T6++ehQthxh99Ew//7jPla1uPmr13UW/f4F870ei6aJTKZ57MAEPrdS9EJs00o/HpfCoTOF653t9m3lLMTLtbLZPJdSWWkhhBBzI8FxEZtX+Tk7mCA+Y/SxHbQ1BGqXCVQUc6BIbs2x3V2ivYoBILbORjfpzFTAbRsYS+J2KoTy9N/dvs5P73CyYEeE505EmIhleEmJzgSN1vs0XEawOxxOkc6YFwSTiQyTZbTDsksqLi9SUgFmcGwYMJmsLjhWVYW/fN0K3v/qDm7cWpd9XFeLm76RZN467XAsU7NOFbbGoAOvS+F8TkCeSht88aEBGgIO7rmrvaqFnAuhrcHJWDSd93M2DINj5yZZX6DFWrlUReGPXtGKx6Xwmfv6eWTfOJ++r58NnV7+4nc7y5pa2Gll6U9eiPPs0QjXbw7m7YlsczoUtC5vwUV5hmHwyL4J1ra72dBZ2evrsmqjJXsshBDzY2n/5lxk2qoAGWPq1q5tOJwm4FUL1ijOVdDrmFZz3FeDNm42+5f7zEC3fzRFW8iZtxZ112VmO6wvPTSYtzPDo/snaKl3sm3N7C4VubK9jsuoO7aDd3sM9WiBetpcdknFdUUyeVB8Sl4sUX5ZRSErm91kjPwLH8djtS+rUBQluyjP9tPdo5wdSvKOl7bUdPHffLG7pOQrrRgYSzEcTrN5ZfHev+VoCDp598vbODOY4Mu/GGTrKh8f+O2Osj9zu53bD58eIZEyeHGRkgrb1tU+eoeTee+AHO+Nc2Ywwa1XFK7VL2Rls9WxQhblCSHEvFj6vz0XkbbKrNM8cWF6cDwaTtFUw5IKW9CnzgqO63xqTYIceyDDhRn1qf1jhRf8Ndc5efttLejnJvnx7tFpX+sbTXLodIyXXFE3a/DHTI115ntVTpmEHRzbPZPHyhgeUk5JBUy1y8vX69hekFeNLitoOTdjsVQ6YxCZzNS8rALsdm5mkHR2MMGPnhnhei3AVRtqP+hjPrRawfHMOxoAR6w+wZvztAicix3r/bz+piZu3hbk/a9pL5r5zXeeDhWOnY+zqsVdsO1brq2rzPM+nCd7/Oj+cbwuhes3l27fNlNj0IHfo8qiPCGEmCcSHBfREHTSGnLOqjseCadrWlJhC3gd01pO9Y0ma5I1BnP4RJ1PnZZlNAyDgbFU0W4YN20N8qLNQX7w65Fp/WIf2z+BqsDN20pn0EJ+BwrlBsdJFAXWW8HHWLR45rjckgqYaqMWKRQcV3kR0tHoQlVmD2iw68jrazQ6Oldnkys7Me2LDw3gc6u89daWmj/PfGkv0uv4yNkYQa/Kiuba/AyA2U3jXS9vK6vVXC6nQ8n+nNxSZq/o1a1uAh6VQ6en1x1HJtM8o0e4YXMwezejEoqi0NUsi/KEEGK+SHBcwoYODyd6p/9yq/XoaFvQq05btNY3UrvgGMzgLbed22gkTSJlTJuON5OiKPzeS1um9YtNpQ0ePzjBlev9Zb0PTodCfcBRVju3gbEUzXVOmurssdPFH1NuSQXkllVMP2bGMCrqVlGIy6nQ0eiaNaBh3Arwy6ltrZQ9oOJrvxzkxIU4b721ZV6C8PkS8DoIeFT6CmSOtZXekncmFsqKRhdOB7xoS3nZXlVV2Lxqdt3xU4fDJFJG2S3v8lnZ7KanP87BMnopCyGEqIwExyWs7/QyHE5ns57pjMFYNF3TNm623MxxIplhOJyuyWI8W2eje9oIaXv4Qu50vHxm9ot9/kSE8Wial1xR/i/3cgeBDIwnaa13EvSpOFQK9vC1lVtSAeAvUFYRTxgYQKAG5Stdze5ZmWN7kWWta45hqp3bU4fD7Fjn54bNy6OcIpfZsWJ65nhwPMXgeKpmJRW18OobGvnjO9srWli5dZWPwfFUtj2hYRg8ut9ciLe2jNKMQm7dUU/A6+Cf7+3lX77fy+mB2V11hBBCzI0ExyVsmDEMZCySxjBq28bNVudTmUwapNJGNnCtZea4s8mVbUcFU7eyW8uYwJfbL/YrDw/SFHSwfW35gUu5g0AGxlK0hJyoikJD0FU0OK6kpAIg4DGDmpkL8uypebXo7NDVbLYmS+R0OBnPjo6ufXDc3mCWcnjdCu+4vWXeh3zMh7YG56zgWD9rZkRrsRivVta2edhVxljsXFtXmz8j9rS8EzkL8ao9l39+50redEsTJy/E+eDXz/EfD/TTV6CzjBBCiPJJcFzCmjY3DpXsMJD5GABiy46QnkzXtFOFLbsoz8oe2/W9rfXlPccrd4a4fI2PiViGWy6vQ1XLD8Qag86SrdwSyQyjkXR2kVZTnZPRIgvy+kaTDIfTXL2+vAEXhbpV2INBarHwcWWLGwM4n5Ohn7BKZermYUGe06Fw1y6zjra5rvYXbAuhLeRiaCI1rQf3kbOT+D1qwZHOy8WKJhchvyPb7/iRKhbizeR2qrxyZwMf+4NV3LkzxO6jEd7773rFY9eFEEJMJ8FxCW6nyupWDyetzHF2dPQ8ZI7tbgrhWGaqx3EZWd1yTbVzM4/dP2rW9zod5QW5dr/YV1wT4o6rQhU9d2PAQWQyMy2jOtPghF3mYb7mxqCz6II8e7qfXXdbisup4HIoBTPHtQiOsx0rcuqO7WEN81FzDPC6m5oqzmguJW0N5pCYofGpoO7I2Rhal7eiC7ClSFEUtlh1x9UuxCsk4HXwhlua+fu3dZFMGXzrsaGaHVsIIS5FEhyXYUOHh1N9cTIZI5vJnJfMsRU8RSbT9Ftt3ALe2j1PW8i8BW8vyusfS86ajFdKKODkzS9urrhEINvruEiZhB3s2pnjxrriZRUD4/b+5b8Gv0edFRzbNci1CFjaG8yWX7k9aCeiafweteyLkEtN+4xexyPhFH2jqSVVUlGNrat9jEXSfP/JkaoX4hWzosnNG29rZ/fRCPuKTLYUQghRnATHZdjQ6WEyaXBuKMlIOI1Dhbp56AiQzRxPZugbLdx/eK7sdlR2Vrp/LFVyMV6t2BcTxRblDYxbNdD1Vua4zsl4NJ13AAmYZSEuh0KogrZ6+YJju3tFLTLHTodCZ6NrWq/jiVh6XkoqLhb297k9Rtrub6wtocV41bD7Hf/ihfGqF+KV8js3t9LZ6OJrvxwsepdGCCFEYfIbuwwbOs0M1okLkwyHU4QCjnlpL5Vbc3yhhj2Oc3U2uugdThBLZBiPpmmrYdlGMXbmuNiivIGx1LRgtzHoImMwbTDKzP2b650VfRb5g+PalVWAOd737LSyitqPjr6YNAQduBxKNnN85GwMr1thTdvyrje2tYactFgXfNUuxCvF7VT5vZe20D+WmjW4RwghRHkkOC5De4OTgFflRG+c0XB6XuqNwexzDGYAOTyRpmMeguOORhd9o6nsgr8llTkeS9KSM8q6qd4uxcj/GLvtWyV8RYLjWnSrALMH7eB4ikmrlnkilp6XOw0XC1VRaAlNdaw4cnaSTSu8OJZ5vbFNURSuWOPD567NQrxStq728aItQe57dpTzw9K9QgghKiXBcRkURckOAxkJp2ich+l4AB6XgtMBp6xx1bUuqwBzUV4ybXDEGkwwH8+Rj9+j4nYqDBfLHI+npgW7jUHz3ArVKQ+MpSqumfZ71Fl9jmPxDC6HUvHUtELsRXl2YDIRTVNXw9rxi1FbyEX/WIqxSIre4eSS6m9cC2+4pZm/e+vKmi7EK+ZNtzThcal89eFBDCN/WZIQQoj8JDgu0/oOD+eGkgyOz890PDCD8IDXwQkrOO5orP3zdFrt3F44ZQXHFQaXc6UoitXruFjmeHoNdKPVmizforxYPENkMpNdvFeuQL7McaL66Xi5ulrMczo3lMQwDCYm09T55UetmHar17FujSi/WBbj2fwedV7KpAoJBZy8/qYmDp+Z5JHf1K68Ip7MZC+shRDiYiW/scu0odOLASRSxrwFx2CWVtjjhuel5thqe6afixHw1LYbRinFpuRFJtNE45lpmWD7fc4XHGcX71UY3Bcqq6hlRq895MLlUDg7mGAyYZBKz890vItJW8jFZNJg99EIHpcyr4vWLhUv2V7Hhg4P/3n/eWKJ2izOe+j5Mf7he730yrARIcRFTILjMq3vmPplPR9t3Gz2orygd34C1zqfit+jkkqXNxmvlopNyRuwJgK25GSCfR4HXpfCWHR2QJ3dv8wBJja/RyWZNkikpoKFWDxTk9HRNlVV6GxycW4oMe89ji8WdnnP3hMRLlvhlbZ3NaAqCm+9tZnRcIpf/GasJsc80GNmjfefKp49Ho+m+enu0Wk/Z0IIsVxIcFymOp8jO5BjvjPHMD9ZYzDLG+zSioVajGdrDDoYjaTy1kAOjOXPBIcCzrw1x4X2L8XuSJFbdxyN17asAsxFeWdzgmPJHBdnd01JpS++korFtKHTy7VaHfc/OzbrjkmlEskMx8+bJV/7e4r3UX5gzyjf/dUwn//ZABmpeRZCLDMSHFdgQ4f5S3s+M8d2tng+6xPtSXkLWQMJ5kVFKp2/NVt2oMeM7hMNAUf+soqxFF63kr2YKJcdHEcT04PjWrVxs3W1uBieSNNn9e6VmuPiWutd2Lnii6W/8VLxtjs6iMQzPLi3uuzxsd44ybRBR6OLw2cmi2aF956IEvCo7D4a4btPDFf1vEIIsdDKTrtpmlYPPAXcret6t6ZpXwZuAiLWLh/Rdf0HmqbdDvwr4AO+o+v631iPvxL4AlAPPA68R9f1wquzlqAd6/3s647SVDePmWNrWER74/wHx5VmXauV285tZt/f/rEU/jw10KGAgzMDs+sbzc4Wrmzbt3L5Pebxo5PTyypqHhxbHSv0s+btZ+lzXJzLqdBY52AimmG91BvX1MYuP9ds9POz50Z52VX1cy7XOnQ6hkOF335RI5+5r5+j5ya5fI1/1n7nhxP0jiR5+23NnBtKcv+eMVrqndx+ZWUj54UQYrGUFRFomnYd8CtgU87mncAtuq5faf33A03TfMCXgFcDW4Bdmqa90tr/G8D7dF3fBCjAu2r1IhbK9VqAT/3xGjyu+csCTmWO5y9wXWEtypuPPsrF2OUo9iS0XINjybydM0J+B2PR/GUVcwnu7fKJ3FvMkXkqq4CpaW8SHJe2vt3DFWt9uJxSb1xrr72hkVjC4IHn5p49Png6xvoOD1et9+NyKOzrzl93/Nxxs+Ti6g0B3nprM1et9/P1R4Z4/kQk7/5CCLHUlBsRvAt4L3AeQNM0P7Aa+JKmafs0TfuIpmkqcC1wTNf1U1ZW+BvA6zRNWwP4dF1/2jreV4DX1fB1LAhFUeZlMl6u+a45BrhyvZ/33d3GpgWu7VzT5qbOp/Lo/vFZXzN7Fs9+zaGAg2g8M20UrmEYc+pxDGQX3tmr91Npg0TKqHnmuCXkxO1U6B0xR1x7XBLwlXLPXe287+72xT6Ni9LqVg/Xbgrw0N6xbB18JSKTaU71xdm6yofHpaKt9LK/O3/d8d7jEda3e2iqc+JQFe65q421bR4+fV8/R88Wr1UWQoiloKyIQNf1P9R1/YmcTR3AL4HfB64Hbgb+AFgB9Obs1wusLLJdzLBjnZ+XXVXPmrb5u7XsUBWu3RSc90B/JrdT5Y6rQuzrjk0br5wxDAbHU9kRu7nsUdK52eOJWIZEyqC1wk4VMFVzHLHKKuwg2V/j4QyqorCi2Ty/Op9acfnHpcjpUKRLxTx67Q2NxJMG9z9bed9j/dwkhgHbVpv14Fes8XFuKMnQxPS7QKPhFCcuxLl641S5hcel8v7XtFPvd/Dhr56aNYRHCCGWmjlFBLqun9R1/bW6rvfquh4FPgncaR0vd2myAmSKbBczNNU5eeutLRdtkPDSHfW4nQoPPDf1C3oskiaZNvJmghsC9gjpqeB4rp0qYKqswg6Kaz06Opdddyyjo8VS0NXs5obNQX7+m3HGCoxkL+TQ6Rhup8KGTvNu0xVrzeB3ZvZ478mpkopcDQEnf/iyVkbCqZKdLoQQYrHNKSLQNO0KTdN+J2eTAiSBs0BnzvYOzFKMQtvFJabO5+CWy+t46nCYYSvr1G/1LC5UVgHTB4EMFNm/FK9LQVGmgmI7i1XrsgqYqjuWemOxVLzmhkZSaYOfVpg9Png6xqYub7YevKvZRVPQwf4Zdcd7j0dob3DS1Tz7Z1Nb6SXgVWc9Rgghlpq5RgQK8AlN0xo1TXMB7wZ+ADwDaJqmbdQ0zQG8GXhA1/UeYFLTtButx78NeKDKcxfL1CuvCZEx4OfPm4uDBotkghv8s8sq7Ol4+cowSlEUBb9HzXariEpwLC4hHY0ubtwa5JcvTDBo/RyVMhpJcW4oydZVUy32FEXhirV+DvRESaXNm4KxeIaDp2NcvSGQt4zIoSpctbGOfd3RvL3OhRBiqZhrWcU+4B+BJ4FDwG90Xf8vXdcngXcA37e2HwHutR72FuDjmqYdAYLAv1d36mK5ag252HVZgF/uGycWz+RMu5sd7Nb5HSgw7TbwwFiKOp+Kd451wn6Pmu1zPK9lFS1TNcdCLBWvvaEJRaHs/sOHz5gdV7aunr6Ad/taH7GEwYkL5mCQfd1R0hm4emNg1jFsOzfVMRJOc3awvMBcCCEWQ0WpN13X1+b8/TPAZ/Ls8zCwI8/2FzC7WQjBnTsb2H00wqP7xxkYT9IYcOB2zg4iHapCnd8xq6xiLiUVNr9HnSqrsILkWo6PtjXXOdm22seWVTLUQiwdLfVO7twZ4odPj3LHVZNctqJ415pDp2P4PSprZywS3rrah6qYdcdal5fnjkeo86lc1ll4MfE1m+oBM5Be1equ/sUIIcQ8kJSWWBTrOzxsWenlwb1jXBhJ0lJkcV3I75i1IK+aASZ+91RwnM0c17hbBZi3nv/idzu5pkgmvpcR9QAAIABJREFUTYjFcNeuBhoDDr75yFDJ8c4HT8fYstKLqk4vlQh4HWzo9LC/O0YqbfDCqShXbwjM2i9XS8jFqhZ3wTZwQgixFEhwLBbNnTsbGA6nOXY+XjQTnDtCOpMxGJpIzamNm83vUbML8eazrEKIpcrjUnn9zU2c7Ivz68Phgvv1jyYZHE+xdXX+ux/b1/o51Rdn99EwsYQxrYVbIVes9aGfm2QyIQ2LhBBLk0QEYtFsX+fLrmpvLbK4LhRwMBo1a46Hw2nSmepGX/u9KpG4GWxH4xm8LgVHkWyXEBejG7YEWd/u4btPDBcMVA+dMTtLFAqO7ZZu33l8GLdTYVsZJUTb1/pJZ6aOLYQQS40Ex2LRKIrCnTsbAGgrMhEwZGWODcMo2tmiXD53buY4LVljcUlSFYW33NrMSCTNfQVaux06HaMh4GBFU/6fz7Xt5tTLkUia7Wt9uF2lf5Y2dXnxuAqPnxZCiMUmUYFYVDdsDvL225rZeVnhutxQwEk6Y061Gxi3ehxXW1aRMMhkDGLxzLy0cRNiObhshZfrtQD37xljcHz6YBDDMDh0ZpKtq3wFJzyqisLla8zscbm19U6HwtbVPvadkpZuQoilSaICsaicDoXbrwwVXRBn9zoejaQZGEuiAM1z6HFs83vN48USGaIJQ4JjcUl7/c3NAHz14UH2nohw8HSMY+cn2Xsiyng0XbCkwnbjliBtISdXri9db2zbvtbP4HiKC6Nzb+mWTBk8fSQsAbYQoubmHmEIsUAaAlODQAbGUjTWOaoar+23AvFoPEM0nqZehnSIS1hLvZNXX9/AvU+O8MKp2V0kZvY3nmn7Oj8f+4PVFT3n9rVmwL3vVIzOxrm1dHvswDhf++UQraEV2bHWQghRCxIciyVvaoR0ioHx6jpVwNQ0vGg8QyyeoaNIvbMQl4JXXdfIdVqQaDxDIpkhnjSIpwyCXpWWKn/e8mkNuehsdLG/O8rLrw7N6RjP6BHAHD+/obOWZyeEuNRJcCyWvFDA/DYds8oqtpW4zVtKbnAclZpjIQBoX+CLxCvW+nhk3wSJZKashXy5RsMpjp4zJ/fNrJUWQohqSVQgljyfW8HlUBgcTzEaTlc1HQ/yBcdSViHEQtu+1k8ybXDECnIrsed4BANwqDA4LqOohRC1JcGxWPIURSEUcHDiQhyD6tq4wVRwPBYxeyZLKzchFt7mlV5cDmVO0/Ke0SN0NbtY1epmSDLHQogak6hALAuhgIOe/jhQfGBIOexg2L4dK2UVQiw8t0tlyyov+05V1u/YLqnYdVmA1npXtr2jEELUikQFYlloCDhIW0O8alVWYd+OleBYiMWxfa2f3pEkj+0fL/sxdknFtZuCNNc7GRpPSTs3IURNSVQglgV7UZ7TAQ3B6mqEHaqC16UwNGFljov0WBZCzJ9bLq/j8jU+vvjzQb712BCZTOkgd/dRs6RiZYublnoniZTBRCz/+GshhJgLiQrEsmAPAmmuc6EWmNZVCZ9HzZZVSM2xEIvD61b5s9d2cMeV9fzsuTE+/qML2dHu+YyGU+hnzZIKMHs0AwyMyaI8IUTtSFQglgW713G1i/Fsfo/KaDid/bsQYnE4VIW33dbCO17awv7uGH/37XP0F5icl1tSAVPBsX0XSAghakGiArEszEdwbOT8XQixuG7bUc//+p1ORsNpPvytcxw/P7vFW25JBZAdUDIwJsGxEKJ2JCoQy0LIKquodjqeLTcgluBYiKVh62ofH3pzFwGvyj/d28vzJyPZr41GppdUgPmz688pkRJCiFqQqEAsC51NbrqaXWxZ5a3J8ezBH6oCHlf1NcxCiNroaHTxwTd20dXs5t9+1JftZLHn2PSSCltLvVOCYyFETcn4aLEs+D0q//h7q2p2PHsRns+totRggZ8Qonbq/Q7+6nWdfPInfXzx54OMRNIcOh1jRdNUSYWtpd5JX4EaZSGEmAvJHItLUsAKjv1e+REQYinyulXe/5oObtwa5L+fGuHI2Umu3RSYtV+L9DoWQtSYRAbikuTPyRwLIZYmp0Ph3S9v5e5dDXhdCjdsCc7ap6XeyWTSIDwpvY6FELUhZRXikmSXVchiPCGWNkVReP3NTfzOjY041NklUHbHiqHxFHW+6gYECSEESOZYXKL8EhwLsazkC4wBWqz2jgOyKE8IUSMSGYhLUrasQoJjIZa1ljozOB4cl0V5QojakMhAXJL8Vq1xQIJjIZa1gFfF61YYksyxEKJGJDIQlyS7S4UsyBNieVMUhZY6p5RVCCFqRiIDcUmSVm5CXDxaQi7JHAshakYiA3FJCgWcvP22Zm7QZreGEkIsLy11MiVPCFE70spNXLJuvzK02KcghKiBlpCTaDxDZDJNwCvt3IQQ1ZHMsRBCiGWtpd7M8wxNSPZYCFE9CY6FEEIsa3ZwPDAmwbEQonoSHAshhFjWcqfkCSFEtSQ4FkIIsazV+VTcTkXauQkhakKCYyGEEMuaoii01DsZkil5QogakOBYCCHEstdSL+3chBC1IcGxEEKIZa+lXqbkCSFqQ4JjIYQQy15LvYvIZIZYIrPYpyKEWOYkOBZCCLHsZXsdS/ZYCFElCY6FEEIse3ZwLHXHQohqSXAshBBi2ZsKjqVjhRCiOhIcCyGEWPbqAw5cDkWm5AkhqibBsRBCiGVPVRSa6pwMTUwPjkfDKR56foxU2likMxNCLDfOxT4BIYQQohZaQ04GczLHB3qifO6BAcajaYJeBy/aElzEsxNCLBdlBceaptUDTwF367rerWna7cC/Aj7gO7qu/42135XAF4B64HHgPbqupzRNWw18A2gDdOAtuq6Ha/5qhBBCXLJa6p3sHYiSyRj88OkRfvT0KCuaXRiGwXPHIxIcCyHKUrKsQtO064BfAZusf/uALwGvBrYAuzRNe6W1+zeA9+m6vglQgHdZ2z8DfEbX9c3AHuCDtXwRQgghREu9k/Fomn+6t5cfPj3KjVuD/6+9O4+O46rTPv7tltRSa7NkyZZkW14S2zfeEseOnYCzkT0kkATMhAQIgcnCQOAdJgMv8xKGsA2cwwzLsAwwwMDAZGJIyL6SDZzEcRbHWWznFzvxbnmTte9S9/tHlRJZbjm2lqqW+vmc42N1dUn1qOuq6te3b93i5isnc9KsAl7e0kpnt+ZAFpF3diRjjq8FPgPs8h8vBTaa2WYz68YriD/knJsGxM3sGX+93/jLc4DTgdv6Lh+e+CIiIp7eGSve3N3BtedP4LoLJpKbE2XxzAI6upKs39YeckIRGQ3ecViFmV0D4JzrXTQJqOmzSg0w5TDLy4FGv5Duu1xERGTYLJiez3sWFHHuieOYUh57a/nc6jjxWIQXNrWw8Jj8EBOKyGgwmNkqokDfy34jQOIoluMvFxERGTZF8Sw+ce6EgwpjgOysCCfMyGfNGy0kEpq1QkQObzDF8Q6gqs/jSrwhFwMt3wuMc85l+cureHuIhoiIyIhbNLOAprYEG2s6wo4iImluMMXxasA552b6Be+VwANmthVod84t89f7mL+8C1gJXO4vvwp4YIi5RUREjtgJ0/PJzoI1m1rCjiIiae6oi2MzaweuBm4H1gOv8fbFdh8Bvu+cew0oBP7dX/5p4Drn3HrgNOCmocUWERE5cvHcKHOr47ywqYVkUkMrRGRgR3wTEDOb3ufrR4ETUqzzEt5sFv2XbwXOHFRCERGRYbB4ZgH/9ch+duzvYuLEsNOISLrS7aNFRCQjnHhsPhHghTc0tEJEBqbiWEREMkJJQTbHVuVq3LGIHJaKYxERyRiLZxawZW8ne+s7w44iImlKxbGIiGSMxTMLAFi1riHkJCKSrlQci4hIxqgszWFyWQ6r1jeGHUVE0pSKYxERySiLji3glS3NNLX1hB1FRNKQimMREckoi47NJ5GAdVvbwo4iImlIxbGIiGSUGRW5FOZlsW6bimMROZSKYxERySjRaIQTji3k1a1tulueiBxCxbGIiGSchTMLqW3qZk99d9hRRCTNqDgWEZGMc+LMIgANrRCRQ6g4FhGRjDOpLEZ5cTbrtraGHUVE0oyKYxERyTiRSIR5U+Os395OIqFxxyLyNhXHIiKSkeZNjdPakWDzno6wo4hIGlFxLCIiGWnu1DigcccicjAVxyIikpGK87OYOiGm4lhEDqLiWEREMta8qXE27mqnoysRdhQRSRMqjkVEJGPNmxanuwdsZ3vYUUQkTag4FhGRjOUm55GdBeu2amiFiHhUHIuISMbKzYkyqypP445F5C0qjkVEJKPNmxZn275OGlt7wo4iImlAxbGIiGS0ef6UbuvVeywiqDgWEZEMN6Mil/zcqIZWiAig4lhERDJcNBphTnUer25tI5nUraRFMp2KYxERyXjzp+VT29TN7vqusKOISMhUHIuISMY7YYY37vhZawk5iYiETcWxiIhkvPLiHOZMyePJ9c0aWiGS4VQci4iIAKfOK2JPfRebajrCjiIiIVJxLCIiApw0q4BYdoSV65rCjiIiIVJxLCIiAsRjUZbMKmC1NdPZlQg7joiERMWxiIiI77R5RbR1JlnzRmvYUUQkJCqORUREfMdV51FWlM2T6zW0QiRTqTgWERHxRSMRls0t5JWtbdQ1d4cdR0RCoOJYRESkj2VzC0km4ekNzWFHEZEQqDgWERHpo6o0xsyqXJ5c36Q5j0UykIpjERGRfk6dV8TO2i427+kMO4qIBEzFsYiISD8nzy4gJyuiC/NEMpCKYxERkX4K8rJYNDOfVa8109WtoRUimUTFsYiISAqnzSuipT3Bg2sawo4iIgFScSwiIpLCgmlxls4u4LanDrBhe1vYcUQkICqORUREUohEIvzteROoLMnhJ/ft1bzHIhlCxbGIiMgA4rEon3t/BR1dCX587x66ezT+WGSsU3EsIiJyGJPLYnzy3Als3NXBipUHwo4jIiMseyjf7Jx7HJgIdPmLrgeOBW4CcoAfmNlP/HXPAb4HxIEVZnbTULYtIiISlHcdV8imXe08tKaBWZNyWTq7MOxIIjJCBl0cO+ciwGxgmpl1+8smA7cCi4EO4Gm/gN4M/Bo4A9gO3Oecu9DMHhhifhERkUBccUYZm/d08MuH9rG7roulswupLM0JO5aIDLOh9Bw7//+HnXNlwH8CTcBjZnYAwDl3G7Ac+Auw0cw2+8t/D3wIUHEsIiKjQnZWhBsuruA/7t/LbU/VcdtTdUydEGPp7AJOnl1IhQplkTFhKMVxKfAo8Fm8IRRPACuAmj7r1ABLgUkplk8ZwrZFREQCN74omy9fPonapm6ee72ZZ19v4ban6rj9qTq+8MFK5k/LDzuiiAzRoItjM1sFrOp97Jz7Fd6Y4m/2WS0CJPAu/EumWC4iIjLqlBVlc8HiEi5YXEJtUzff+N+d3Ptcg4pjkTFg0LNVOOdOdc6d3WdRBNgCVPVZVgnsAnYMsFxERGRUKyvK5uyF41i/rY3t+zrDjiMiQzSUqdxKgO865/Kcc0XAx4GPAmc75yY45/KBDwIPAqsB55yb6ZzLAq5E441FRGSMOHNBEbHsCA+/qFtNi4x2gy6Ozexe4D7gReAF4Ndm9hTwZeBxYC1wi5k9a2btwNXA7cB64DXgtqFFFxERSQ9F8SyWzS3k6Q3NNLb2hB1HRIZgSPMcm9lXgK/0W3YLcEuKdR8FThjK9kRERNLVeSeO4/GXm3jilUbef3Jp2HFEZJB0hzwREZFhMLksxvxpcR5d26jbTIuMYiqORUREhsn5i8ZR19LDs6+3hB1FRAZJxbGIiMgwWTA9TmVpDg+vaSCZVO+xyGik4lhERGSYRCMRzjuxmDf3dLCppuOt5a0dCe57rp6b/2cnO/ZrujeRdDakC/JERETkYKfOLeK2p+p4aE0DZUXZPLSmgcdfaaS90+tJXvNGC1PKYyGnFJGBqDgWEREZRnmxKGfML+LBNQ28sKmFZBKWzi7gvSeV8JP79rJ5T8c7/xARCY2KYxERkWF23qJxvLq1jTnVeZy/aBwTxuUAMKMixsZdKo5F0pmKYxERkWFWVpTNt66acsjy6RW5PGMtNLb2UJyfFUIyEXknuiBPREQkIDMqcgHYoqEVImlLxbGIiEhApk30imONOxZJXyqORUREApKfG6WyNEc9xyJpTMWxiIhIgKZPjLF5r4pjkXSl4lhERCRAMypyOdDUQ2NrT9hRRCQFFcciIiIB6r0oT+OORdKTimMREZEA9V6Up3HHIulJxbGIiEiA4v5Feeo5FklPKo5FREQCNqMiVz3HImlKxbGIiEjAplfEONDcQ0NLd9hRRKQfFcciIiIBm/HWuOPOkJOISH8qjkVERAI2bWIuEdB8xyJpSMWxiIhIwHRRnkj6UnEsIiISAl2UJ5KeVByLiIiEYHpFjLrmHup1UZ5IWlFxLCIiEoLpFbooTyQdqTgWEREJQe9FeRpaIZJeVByLiIiEIB6LUjk+RzNWiKQZFcciIiIhmTExly27VRyLpBMVxyIiIiGZXplLXUsPB5q6wo4iIj4VxyIiIiGZMTEGwO8f2U1ja0/IaUQEIDvsACIiIplq5qQ8TptXyIPPHeCxF+s4Z2ExFy4uoTg/K+xoIhlLPcciIiIhyYpGuPb8ifzs7x0nHpPP/c81cOOvtrFiZS0Nmv9YJBTqORYREQnZ1Il5fPqiCi45pZO7nqnj/ucaeHhNI6fOLeSCk8ZRVRoLO6JIxlBxLCIikiYml8X49EUVXPauTh54oYEn1zfzxCtNLJqZz3tPKmFmVS6RSCTsmCJjmopjERGRNFM1PsYnz53AB99dyp/XNvLo2kZe2LSLoniUGRW5HFOZyzGVeUyviJEfO3SEZHZ2hKiKaJFBUXEsIiKSpsYVZLN82XguXlLCamtm464O3tzdzitb20gmD/+9sewIsewIuTlR4rEI5y8ax+nzi9TzLPIOVByLiIikubxYlDMWFHPGAu9xe2eCLXs72Lq3k67ug6vkJNDVnaCjK0lnd5KOrgS7DnTxqz/v5/Vd7Vx1Vjm5OboeX2QgKo5FRERGmbxYlOOmxDluSvyI1k8kktz5TB13PVPP1r2dfPbiCipKc0Y4pcjopLeOIiIiY1w0GuED7x7PjZdVUtvUzT//zw5e2NQSdiyRtKSeYxERkQxx/Ix8vvHRyfzonr388O49TJ0Q8y/u8/5NLsvsKeO6upM8vaGJipIcjqs+sl75TNDdk2Tluiaa2nq4aEkJWdGxPW5dxbGIiEgGKS/O4abLJ/Hgmno2bG/n2ddbeOKVJsC7iG/e9AJmV8VYMD3OlPLYQbNedPrjl3fXd3FSVuywRUTNgU4eerGR4niUpbMLmVyWk7YXA3Z1J/nLq43c82w9dc09ZGfB37+/kuNn5IeaqyeRJBLhqGceqW/u5u7V9bR2JLhoSQnVEwb3pieRSLLqtWb+tKqOfQ3eTWle39nOZy6qIJ576OCDZDLJX19tYuu+Tk6fV8T0itxBbTdsKo5FREQyTE52hPctLeV9S72CZk99N2/u7uCNmnZer+lkxaZmVqyE4vws5lbn0dGdZOf+TvY1dNN7+V/0/r2cNKuA955UwjGVbxdB+xu7uHNVPSvXN5EdjdCdSHLnM/VMGp/D0tkFLJlVCMC2fR1s3dfJtr0d7KztYk51Hh99T/lhb53d1pGgqydJYV6U6FH2XiaSSXp6Dl7Wk0jy5Pqmt4ri2ZPz+PjZ5dyxqo4f3r2Hz19awfxpqQvkji7vosfCeHTYps1rae9hU00HG3e1s3FnO2/s7iA/N8rFS0o4c0ERsXe4kLKtI8F9z9fz4AsN9CSS5GRHWPVaM6ccV8gH3lV6xOPME8kkL2xq5U9PH2BnbRfTJsa48dJKapu7+e9H9/PNFbv4h8sqKSt6u4zc39jFLx/ez/ptbUQj8MjaRtzkPM5fNI5Fx+a/tb8ONHWzYXsbr+1oZ1xBFsuXjR/8CzZCVByLiIhksEgkQmVpDpWlObx7TiETJhRhb9axblsb67a28dqONuK5UaZX5LJsbhGTy3IoL85h3c5O7l21n2dfb2FOdR7nLhzHhu1tPP5KIwDnnVjM+5aWkkgmeX5jC8+93sJdz9Rz5zP1b207JytCdXmM46rzeG5jC+u2tfHxs8tZOrvwoIz1Ld3c/3wDj73USGd3kghQGI9SFM+iOD+LCeOymTQ+xpSyGJPLcxhflE17Z5I3atq9QnOXV/i3d6We/2725Dyuu2Aic6vziEQizJqUx3f+WMMP7trDjZdVMqfPEIvO7gSPrG3kntX1tHQkyIpCaWE2pYVZlBRkUxiPkpsTJdefRi+WEyEei5KfG6UgN0p+nvd1a0eCHfu72HWgk537O9lR28We+i4AohGYOjHGGfOL2L6/k98/Ucu9z9W/VST319GV4C+vNHHX6jqa2hKc4gr44LLxFORGuf/5Bh5+sYHV1szp84tYOruAuuYe9jV0sb+xm/2N3TS09tDZlXyr4O/q8V6nqvE53HDxRE6aVfDWG4AJxdn8+N493HzLTj5/SSXl5YU89nIjt/6lFoCrzynn5NkF/HVdE39+sZF/v2cP5cXZuMl5bNzVzl6/Bzo/N8oZ8w/9XdJBJPlOEyWGbzqwuba2mURieLJOmFDEvn1Nw7ZeEFmCkE5ZIL3ypFNbyLTX5UgpS+rt6nVJLZ2yQHrlOZpj0bYdDTzxSiMPrmmgrrmHaAROn1/EJaeUHtSr2Ku+pZuX3mwllhNl6oQYlaU5b41f3bG/k188tJctezpZOruAq84qp3R8Ab97aCePvdRIV0+Sd88pZEZFLk2tPTS19dDY2kNjW4I99V00tLzdLRzLjtDVnSQJRCJQXR5j1qQ8xhcd2it9bGUec/yiuK/G1h6+/cdd7Gvo5gsfqORdJ0zg7pU13P5UHbVN3Rw/Pc6C6fnUt3RT19xDfbP3f0tHgo6uBJ1dSY6kaolGoKI0hyllMaZO8HIeU5lLXp+bu2zY3sYdq+p4bUc7JQVZLJ5dzJ4D7dQ1d1PvbxNgbnUef3Na2UE9+b2v+z2r63ns5UZ6vFWJAKWFWZSPy2Fcfha5OV4x3/t/VWkOS2YXpBxfvGN/J9+7czeNrT3MnBxn/dZW5k6Nc8155ZQXv9073ZNIsuaNVh5e08CuA53MnJTHnCl5HFcdZ2p5LGXv/1D+Fo7ke6PRCGVlhQAzgC2p1gm059g5dyVwE5AD/MDMfhLk9kVERGT4xHOjXHhSCeeeOI5Xt7ZRVZpz2I/uSwqyOWNBccrnppTH+OoVk7nvuXruWFXH+m1tdPV4czUvm1PI+08upfIwP7u5rYedB7rYWdvJrtpOCvKymDUpl2Mr81KOj30nxflZ/N/lVXz7DzX82x27qfxrHVt2tzO9Isa151cxd+rhL9hLJpN0dSfp6E7S3pmgpSNBa3uC1o4ELR095OZEmVIWo6Ikh5zsww/LmFMdZ051nA3b27hrdT1r32hmXH6UypIc5kyJU1KYxcyq1EU+eK/7x84q56IlJeyu66KsOJuyomyyswY3HMTbV5P4wV172FzTztXnlPOeBYfeYCYrGmHJrAKWzCoY1HbCElhx7JybDHwLWAx0AE875x43s/VBZRAREZHhl50VYeExQ794LSsa4f0nl7Lo2AJWrKxl4vg8zj2+8LBFca/CeBZuchZuct6Qc/QqKcjmSx+q4jt/rKG9M8GnL5rI0tkFRzTGOBKJEMuJEMuBongWE4YhT2+RPNje1fFF2YxP0aM/GOMKsvnKhydROK6A1qbWYfmZ6SLInuNzgMfM7ACAc+42YDnw9QAziIiISJqbUh7jxsuq0mK4SWlhNt+6agoVE4uorW0ONUu6iUYjFORl0ZoeI4KGTZA3AZkE1PR5XANMCXD7IiIiIkctOyty1LNjyOgVZHEchYPGpkeARIDbFxERERE5rCCL4x1AVZ/HlcCuALcvIiIiInJYQY45fgS42Tk3AWgBPghcF+D2RUREREQOK7CeYzPbCXwZeBxYC9xiZs8GtX0RERERkXcS6DzHZnYLcEuQ2xQREREROVJBjjkWEREREUlrKo5FRERERHwqjkVEREREfIGOOR6kLGDYJ98+0p8XxKTf6TSxeDplgfTKk05tIdNelyOlLKm3q9cltXTKAumVJ52ORaPxdQnCWM4ylJ/3Tt/b5/msgdaJJJPJgZ5LF6cCK8MOISIiIiJjxmnAk6meGA3FcS6wBO920z0hZxERERGR0SsL76Z0zwEdqVYYDcWxiIiIiEggdEGeiIiIiIhPxbGIiIiIiE/FsYiIiIiIT8WxiIiIiIhPxbGIiIiIiG803ARkyJxzXweWA0ngV2b2PefcOcD3gDiwwsxuCijLlcBNQA7wAzP7iXPuXcD3gSLgZeDjZtYZUJ5i4GngYjPb4pz7O+AGIALcB3zRzEZ8ShPn3FeBv/Ef3mdmXwxxHx2UBXgC+Jc+q0wGVpvZxQFkSdVeQtlHfp7+7SWUfdQ/CzCX8PbRIW23z3M3AMvN7MyRztFnm6nazHXA5/COgc8D1wdxjBkgy9XAF/Gm5nwMuNHMukc6i5+nf/sN5dg7wPEurGOvzo8D59H58R2yEOL50c9zSJvp89ygj79jvufYOXcGcBZwPHAS8Fnn3AnAr4FLgDnAEufchQFkmQx8C+/GJguB65xzxwN/Aq4zs3n+qn870ln8PCfjTYA92388A/gHYCmwAHg3cG4AOc4BzgNOxHtdFjvnriCcfXRIFiDXzBaa2ULgAqAR+HwAWVK1l3mEsI/8PP3bS5wQ9lGqLGZ2f0j7KFXbvcx/bi7wpZHO0C/PQMeYL+C1lePxjvufCSnLCcA3gbPNbAHeCe1zI53Fz9O//RYTwrF3gDbzecI59ur8OHAenR+PIAshnR/9PKnazFz/uSEdf8d8cWxmfwHe4/dMTMTrLS8BNprZZn/574EPBRDnHOAxMztgZi3AbcBlwCoze9lf57PAHQHduE5LAAAM50lEQVRkAbgW7yS5C8DMNgNz/WwlwDigPoAcNXi9R51m1gVswDsghbGPUmWZ2uf57wI/M7ONAWRJ1V4+SDj7CPq1F7yTRBj7KFWWvoLcRynbi3MuF/g58M8BZOhroGPMp82s0e/leoWD23SQWS7FO97V+Ovc6y8LQv82cy7hHHtTtZkEIfxd6/x4WDo/HlmWsM6PkLrNLB+O429GDKswsy7n3NeAfwT+CEzC28m9aoApAURJtd3/BH7nnLsVOA54CrgxgCyY2TUAzrm+y7qcc9cC/wo8C6wNIMe63q+dc7PwPrL5ESHsowGyLOvz+EzgmpHO4UvVXpaGsY8gZXsJ6+8oZdv1Hwe6jw7TXr6N17OzOYgcfQzUZv7sZ5yA97Hw1SFluRKY5Zyrxis6lgOVAWRJ1WZmAs1BH3sHajMh/l3r/JiCzo9HnCWs8yMMcLxjGI6/Y77nuJeZfRWYAFTjvevqO04ogvfOfaRFU2y3Gzgf+Ce8jygKCPij2P7M7D+BMmA3cHNQ2/WHDPwZ7yPgNwlnHx2Spc+74OuAn5pZyttNjoBU7SUB4e2jfgbMF6Kg9xFwSNudDkw1s/8KMoNvwH3ifwT5KN640idCyvI63vHtbmAl3hjSQMaPppBNiMfeVMeYsP6udX48cjo/ps35EVK3mbMZhuPvmC+OnXPHOecWAphZK974pTPx7qvdq5LUH88Otx0ptvtN4Bn/45Ee4A9473wC55yrds4tA/A/qrkVbyxaENtehnfi/pKZ/ZbUr1UQ+yhVll6X4r0mQUn1GtSEtY9SCG0fHUbQ+yhVe7kCmOecWwv8EjjJObcioDgp94lz7ji8C4t+a2bfCDML8KyZnWhm7wZ2Am8ElKe/3YR07O3fZsI69ur8eOR0fhwwS6/Aj72kfh1mMAzH30wYVnEM8DXn3Kl47zAuwRuL8l3n3Ey8bvcr8brgR9ojwM3+R5steONH/w9wrXOu2sy24115/0IAWVIZB/yPf7BswPvI88mR3qj/EeudwOVm9pi/eLX3VLD7aIAsOOfKgbg/7iwoqdrL5whhHw0glH00kDD2Uar2Ymaf7PP8mcDNZnZ5QJEGOsY8DHzZzH4XUI6BslwPPOr3PHXgjSH9WYCZ+noY79wQ6LF3gGNMKMdedH48Gjo/ptf5EVK3mVPM7Fk/15kM8vg75nuOzex+vOlGXsT7o3razG7FG3N3O7AeeA1vIPdIZ9kJfBl4HG+s0i3+BRHXA/c4514DxuONlwmcmb3qb/tp4CWgFfi3ADb9j0Ae8D3n3Fr/Hd/VhLCPUmVxzn0K7ySyI4Dtv2WA9vJXwtlHqfK1E84+Gkjg+4iB20soUrUZYBFQAdzYJ+PXw8hiZquBrwHPAK/iXUxzy0hnGSDfdsI59qY63p1KCH/XOj8eVT6dH9Po/AgDtplnh+NnR5LJQKZHFRERERFJe2O+51hERERE5EipOBYRERER8ak4FhERERHxjani2Dk33TmXdM79vN/yhf7yqzM1j7Ioi7IoS6bkURZlURZlGYoxVRz7aoELnHNZfZZdDuxTHmVRFmVRlozJoyzKoizKMihjcZ7jZrwpPU7Hm94D4Dy8+fBwzl2MN7F4FO8uM9eb2R7n3Bbgd3h34ykArjKz4ZhPMZ3yKIuyKIuy6HinLMqiLMpyGGOx5xi8u+gsB3DOLeHtW5ROxJvg/FIzOx7vPu0/7vN9tWa2FG9S+v83RvMoi7Ioi7KMVJZ0y6MsyqIsynLUWcZqcXw3cKFzLorX5d5768BWvFuXbvEf/wLvPty9HvT/fxVvsvGxmEdZlEVZlGWksqRbHmVRFmVRlqPOMiaLYzNrxruDzanAWfhd7hz6+0Y4eGhJu/9/0n9uzOVRFmVRFmUZqSzplkdZlEVZlGUwWcZkcez7A/Ad4Hkz6/aXxYFTnHPT/cfX8fa4lUzKoyzKoizKkil5lEVZlEVZjspYLo7vARbydpc7wB68F+wO59w64EzgUxmYR1mURVmUJVPyKIuyKIuyHJVIMpkcSkARERERkTFjLPcci4iIiIgcFRXHIiIiIiI+FcciIiIiIj4VxyIiIiIiPhXHIiIiIiI+FcciIiIiIj4VxyIiIiIiPhXHIiIiIiI+FcciIiIiIj4VxyIiIiIiPhXHIiIiIiI+FcciIiIiIj4VxyIiIiIiPhXHIiIiIiI+FcciIiIiIj4VxyIiIiIivuywA6QL59zVwJlmdnXIUWQMcs5NB14H1vd76n1mtj3F+jcDmNnNI51NRg+/HW0GfmFm1/dZvhB4EfiEmf0mnHQyVjnn5gOvAMvN7Paw88joNVqOYSqORYKzy8wWhh1CRr1a4ALnXJaZ9fjLLgf2hZhJxrZPAn8ErgdUHMtQpf0xTMVxP865M4BvAflACfB5M7vLOfcboAFYDEwGvm5m/xVaUBkTnHMVwM+BaiAB/JOZPeI/vdQ5txooxHuX/cOQYkp6aQbWAqcDj/vLzgMeAXDO3QB8DCgAOoErzMycc1uA1cBC4DQz2xtsbBmNnHM5wEeA04CnnXPHmtkbfnv6A3Cuv+onzexF59wTwAFgHnC5ma0NPrWkuaM+hgFT8OquZf46VwMnm9nfjURAjTk+1GeBa8xsEXAN8M0+z1XjHSDeD/xrCNlkdJvknFvb598XgB8CvzazxXjt6ufOuSJ//SrgLOBdwA3+x04i4BUlywGcc0uAl/FOIsXApXhDxOYD9wI39Pm+B8zMqTCWo3ARsNXMXgfuBK7r81yLmZ0I/DPw2z7LX/bbmQpjGcjRHsMeA6qcc8f6338V8JuRCqfi+FAfBeY7574C3IjXa9frYTNLAq8C48MIJ6PaLjNb2Offd4FzgK8759YCDwA5QO8f/61m1mJmjcA9wBnhxJY0dDdwoXMuivdx5Ap/eSNwJfBh59y3gfdx8DFsdaApZSz4BPC//tcrgE8452L+418AmNk9wBTnXLm/XO1M3slRHcP82uu3wEedc1OBCjMbsXaW0cWxc+4059wk/2EE6AZWAkuBF/CGV0T6fEs7gL+TRIZDFnBWb8EMnIx34Qt47bFXFOgKOpykJzNrBl4CTsX7dKF3KE41sApvSNgDeD0rfY9hbcGllNHOOTcRuBC40R9G8UugFPiAv0r/Y1Tv+FG1MzmsQR7DfgN8GG+YxX+PZL6MLo7xLjK41P/6eOBNYDbeR0QPAJfgFS8iI+Ux4NMAzrm5eJ9K5PvPLXfO5TrnSoGLeXtslgh4H0t+B3jezHqLlBZgk5l9H3gOuAwdw2TwPgY8amZTzGy6mU3D6zT6lP/8hwGcc5cBG8ysLqScMjod1THMzLYCO4C/A343ksEyvTj+NvBx59wGYD7wH8CvgHXABqAIyHfOFYQXUca4zwKnOOdexvtY6aNm1uQ/txV4CngS+Bcz2xBSRklP9+BdXLeiz7JOIOqcWw+sAV4DZoSQTcaGq4Gf9lv2E7xPV/OAZf6QsH8EPh5sNBkDBnMMuxVYb2a7RjJYJJnUCAERERE5cv4wizPNbEu4SSRTOOey8XqM/2hmfxrJbWV6z7GIiIiIpDHnXATYhTfl6Z0jvT31HIuIiIiI+NRzLCIiIiLiy7g75Dnnvgr8jf/wPjP7onPuHOB7QBxYYWY3+eteAnwNbxqRzXj3/K7z59j7PTARMOAj/rQkIgdRe5OhUhuSIKm9yUgYbe0qo3qO/R1xHnAi3hWSi51zVwC/xpu2bQ6wxDl3oXOuGG/2iovM7AS8u7fc7P+onwI/NbPjgOeBrwT6i8iooPYmQ6U2JEFSe5ORMBrbVUYVx0ANcKOZdZpZF950bbOBjWa22Z9n7/fAh/DuVPYZM9vpf+/LwFT/PvOnA7f5y3/jry/Sn9qbDJXakARJ7U1GwqhrVxk1rMLM1vV+7ZybhdfF/yO8HderBphiZrXAHf66ceBL/rrlQGOfCatrgCkjn15GG7U3GSq1IQmS2puMhNHYrjKt5xgA59w84M/AF/Duitd3yo4I3lQhveuOA+4DXjKz3+K9Zv2n+EggMgC1NxkqtSEJktqbjITR1K4yrjh2zi0DHgW+5L/gO4CqPqtU4s2lh3OuCliJ161/jf/8XmCcc673lqxVveuL9Kf2JkOlNiRBUnuTkTDa2lVGFcfOuWq8yaOvNLNb/cWrvafcTP9FvxJ4wP/6HuAPZvb3ZpYE8MfLrAQu97//KuCBIH8PGR3U3mSo1IYkSGpvMhJGY7vKqJuAOOd+CHwSeKPP4p8BG/GmE8kD7gf+AbgUuB3vnUuv583sGufcNOC3eNOJbAOuMLO6kf8NZDRRe5OhUhuSIKm9yUgYje0qo4pjEREREZHDyahhFSIiIiIih6PiWERERETEp+JYRERERMSn4lhERERExKfiWERERETEp+JYRERERMSn4lhERERExPf/AcXOrVvbBQFmAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot of the number of flights each day in France\n",
    "df_flights_airports_groupby = df_flights_airports.groupby('day').count()[\"origin_airport_name\"]\n",
    "fig, ax = plt.subplots(figsize=(10, 5))\n",
    "ax.plot_date(df_flights_airports_groupby.index,\n",
    "             df_flights_airports_groupby.values,\n",
    "             '-')\n",
    "\n",
    "ax.xaxis.set_minor_locator(dates.WeekdayLocator(byweekday=(0),\n",
    "                                                interval=1))\n",
    "ax.xaxis.set_minor_formatter(dates.DateFormatter('%d\\n%a'))\n",
    "ax.xaxis.grid(True, which=\"minor\")\n",
    "ax.yaxis.grid()\n",
    "ax.xaxis.set_major_locator(dates.MonthLocator())\n",
    "ax.xaxis.set_major_formatter(dates.DateFormatter('\\n\\n\\n%b\\n%Y'))\n",
    "\n",
    "plt.title('Count of daily flights in France', fontsize=16)\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> As expected, the number of flights decreased a lot as soon as the start of the lockdown was announced (March 16)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* Finaly, a quick diving in air quality data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['wind gust', 'wind-gust', 'humidity', 'wind-speed', 'pm10', 'pm25',\n",
       "       'no2', 'dew', 'temperature', 'wind speed', 'pressure', 'co', 'o3',\n",
       "       'so2', 'precipitation'], dtype=object)"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# See all types of indicators measured\n",
    "df_air.Specie.unique()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> We were interested in pollution indicators. According to specialized websites, we then only kept pm10, pm25, no2, co, so2, o3."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "9.452732901463399e-31"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Look at the variance of co:\n",
    "df_air[df_air[\"Specie\"] == \"co\"][\"median\"].var()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> As the co indicators variations are way too small, we choose to remove it."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# <span style=\"color:#990000\"> III - Data Analysis <a class=\"anchor\" id=\"third-part\">"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <span style=\"color:#990000\"> **Below are the functions we have coded to realize our interactive plots. To go to the analysis part and see the result directly, click [here](#data_analysis).**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <span style=\"color:#008000\"> General variables for Bokeh"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "    <div class=\"bk-root\">\n",
       "        <a href=\"https://bokeh.pydata.org\" target=\"_blank\" class=\"bk-logo bk-logo-small bk-logo-notebook\"></a>\n",
       "        <span id=\"1001\">Loading BokehJS ...</span>\n",
       "    </div>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/javascript": [
       "\n",
       "(function(root) {\n",
       "  function now() {\n",
       "    return new Date();\n",
       "  }\n",
       "\n",
       "  var force = true;\n",
       "\n",
       "  if (typeof root._bokeh_onload_callbacks === \"undefined\" || force === true) {\n",
       "    root._bokeh_onload_callbacks = [];\n",
       "    root._bokeh_is_loading = undefined;\n",
       "  }\n",
       "\n",
       "  var JS_MIME_TYPE = 'application/javascript';\n",
       "  var HTML_MIME_TYPE = 'text/html';\n",
       "  var EXEC_MIME_TYPE = 'application/vnd.bokehjs_exec.v0+json';\n",
       "  var CLASS_NAME = 'output_bokeh rendered_html';\n",
       "\n",
       "  /**\n",
       "   * Render data to the DOM node\n",
       "   */\n",
       "  function render(props, node) {\n",
       "    var script = document.createElement(\"script\");\n",
       "    node.appendChild(script);\n",
       "  }\n",
       "\n",
       "  /**\n",
       "   * Handle when an output is cleared or removed\n",
       "   */\n",
       "  function handleClearOutput(event, handle) {\n",
       "    var cell = handle.cell;\n",
       "\n",
       "    var id = cell.output_area._bokeh_element_id;\n",
       "    var server_id = cell.output_area._bokeh_server_id;\n",
       "    // Clean up Bokeh references\n",
       "    if (id != null && id in Bokeh.index) {\n",
       "      Bokeh.index[id].model.document.clear();\n",
       "      delete Bokeh.index[id];\n",
       "    }\n",
       "\n",
       "    if (server_id !== undefined) {\n",
       "      // Clean up Bokeh references\n",
       "      var cmd = \"from bokeh.io.state import curstate; print(curstate().uuid_to_server['\" + server_id + \"'].get_sessions()[0].document.roots[0]._id)\";\n",
       "      cell.notebook.kernel.execute(cmd, {\n",
       "        iopub: {\n",
       "          output: function(msg) {\n",
       "            var id = msg.content.text.trim();\n",
       "            if (id in Bokeh.index) {\n",
       "              Bokeh.index[id].model.document.clear();\n",
       "              delete Bokeh.index[id];\n",
       "            }\n",
       "          }\n",
       "        }\n",
       "      });\n",
       "      // Destroy server and session\n",
       "      var cmd = \"import bokeh.io.notebook as ion; ion.destroy_server('\" + server_id + \"')\";\n",
       "      cell.notebook.kernel.execute(cmd);\n",
       "    }\n",
       "  }\n",
       "\n",
       "  /**\n",
       "   * Handle when a new output is added\n",
       "   */\n",
       "  function handleAddOutput(event, handle) {\n",
       "    var output_area = handle.output_area;\n",
       "    var output = handle.output;\n",
       "\n",
       "    // limit handleAddOutput to display_data with EXEC_MIME_TYPE content only\n",
       "    if ((output.output_type != \"display_data\") || (!output.data.hasOwnProperty(EXEC_MIME_TYPE))) {\n",
       "      return\n",
       "    }\n",
       "\n",
       "    var toinsert = output_area.element.find(\".\" + CLASS_NAME.split(' ')[0]);\n",
       "\n",
       "    if (output.metadata[EXEC_MIME_TYPE][\"id\"] !== undefined) {\n",
       "      toinsert[toinsert.length - 1].firstChild.textContent = output.data[JS_MIME_TYPE];\n",
       "      // store reference to embed id on output_area\n",
       "      output_area._bokeh_element_id = output.metadata[EXEC_MIME_TYPE][\"id\"];\n",
       "    }\n",
       "    if (output.metadata[EXEC_MIME_TYPE][\"server_id\"] !== undefined) {\n",
       "      var bk_div = document.createElement(\"div\");\n",
       "      bk_div.innerHTML = output.data[HTML_MIME_TYPE];\n",
       "      var script_attrs = bk_div.children[0].attributes;\n",
       "      for (var i = 0; i < script_attrs.length; i++) {\n",
       "        toinsert[toinsert.length - 1].firstChild.setAttribute(script_attrs[i].name, script_attrs[i].value);\n",
       "      }\n",
       "      // store reference to server id on output_area\n",
       "      output_area._bokeh_server_id = output.metadata[EXEC_MIME_TYPE][\"server_id\"];\n",
       "    }\n",
       "  }\n",
       "\n",
       "  function register_renderer(events, OutputArea) {\n",
       "\n",
       "    function append_mime(data, metadata, element) {\n",
       "      // create a DOM node to render to\n",
       "      var toinsert = this.create_output_subarea(\n",
       "        metadata,\n",
       "        CLASS_NAME,\n",
       "        EXEC_MIME_TYPE\n",
       "      );\n",
       "      this.keyboard_manager.register_events(toinsert);\n",
       "      // Render to node\n",
       "      var props = {data: data, metadata: metadata[EXEC_MIME_TYPE]};\n",
       "      render(props, toinsert[toinsert.length - 1]);\n",
       "      element.append(toinsert);\n",
       "      return toinsert\n",
       "    }\n",
       "\n",
       "    /* Handle when an output is cleared or removed */\n",
       "    events.on('clear_output.CodeCell', handleClearOutput);\n",
       "    events.on('delete.Cell', handleClearOutput);\n",
       "\n",
       "    /* Handle when a new output is added */\n",
       "    events.on('output_added.OutputArea', handleAddOutput);\n",
       "\n",
       "    /**\n",
       "     * Register the mime type and append_mime function with output_area\n",
       "     */\n",
       "    OutputArea.prototype.register_mime_type(EXEC_MIME_TYPE, append_mime, {\n",
       "      /* Is output safe? */\n",
       "      safe: true,\n",
       "      /* Index of renderer in `output_area.display_order` */\n",
       "      index: 0\n",
       "    });\n",
       "  }\n",
       "\n",
       "  // register the mime type if in Jupyter Notebook environment and previously unregistered\n",
       "  if (root.Jupyter !== undefined) {\n",
       "    var events = require('base/js/events');\n",
       "    var OutputArea = require('notebook/js/outputarea').OutputArea;\n",
       "\n",
       "    if (OutputArea.prototype.mime_types().indexOf(EXEC_MIME_TYPE) == -1) {\n",
       "      register_renderer(events, OutputArea);\n",
       "    }\n",
       "  }\n",
       "\n",
       "  \n",
       "  if (typeof (root._bokeh_timeout) === \"undefined\" || force === true) {\n",
       "    root._bokeh_timeout = Date.now() + 5000;\n",
       "    root._bokeh_failed_load = false;\n",
       "  }\n",
       "\n",
       "  var NB_LOAD_WARNING = {'data': {'text/html':\n",
       "     \"<div style='background-color: #fdd'>\\n\"+\n",
       "     \"<p>\\n\"+\n",
       "     \"BokehJS does not appear to have successfully loaded. If loading BokehJS from CDN, this \\n\"+\n",
       "     \"may be due to a slow or bad network connection. Possible fixes:\\n\"+\n",
       "     \"</p>\\n\"+\n",
       "     \"<ul>\\n\"+\n",
       "     \"<li>re-rerun `output_notebook()` to attempt to load from CDN again, or</li>\\n\"+\n",
       "     \"<li>use INLINE resources instead, as so:</li>\\n\"+\n",
       "     \"</ul>\\n\"+\n",
       "     \"<code>\\n\"+\n",
       "     \"from bokeh.resources import INLINE\\n\"+\n",
       "     \"output_notebook(resources=INLINE)\\n\"+\n",
       "     \"</code>\\n\"+\n",
       "     \"</div>\"}};\n",
       "\n",
       "  function display_loaded() {\n",
       "    var el = document.getElementById(\"1001\");\n",
       "    if (el != null) {\n",
       "      el.textContent = \"BokehJS is loading...\";\n",
       "    }\n",
       "    if (root.Bokeh !== undefined) {\n",
       "      if (el != null) {\n",
       "        el.textContent = \"BokehJS \" + root.Bokeh.version + \" successfully loaded.\";\n",
       "      }\n",
       "    } else if (Date.now() < root._bokeh_timeout) {\n",
       "      setTimeout(display_loaded, 100)\n",
       "    }\n",
       "  }\n",
       "\n",
       "\n",
       "  function run_callbacks() {\n",
       "    try {\n",
       "      root._bokeh_onload_callbacks.forEach(function(callback) {\n",
       "        if (callback != null)\n",
       "          callback();\n",
       "      });\n",
       "    } finally {\n",
       "      delete root._bokeh_onload_callbacks\n",
       "    }\n",
       "    console.debug(\"Bokeh: all callbacks have finished\");\n",
       "  }\n",
       "\n",
       "  function load_libs(css_urls, js_urls, callback) {\n",
       "    if (css_urls == null) css_urls = [];\n",
       "    if (js_urls == null) js_urls = [];\n",
       "\n",
       "    root._bokeh_onload_callbacks.push(callback);\n",
       "    if (root._bokeh_is_loading > 0) {\n",
       "      console.debug(\"Bokeh: BokehJS is being loaded, scheduling callback at\", now());\n",
       "      return null;\n",
       "    }\n",
       "    if (js_urls == null || js_urls.length === 0) {\n",
       "      run_callbacks();\n",
       "      return null;\n",
       "    }\n",
       "    console.debug(\"Bokeh: BokehJS not loaded, scheduling load and callback at\", now());\n",
       "    root._bokeh_is_loading = css_urls.length + js_urls.length;\n",
       "\n",
       "    function on_load() {\n",
       "      root._bokeh_is_loading--;\n",
       "      if (root._bokeh_is_loading === 0) {\n",
       "        console.debug(\"Bokeh: all BokehJS libraries/stylesheets loaded\");\n",
       "        run_callbacks()\n",
       "      }\n",
       "    }\n",
       "\n",
       "    function on_error() {\n",
       "      console.error(\"failed to load \" + url);\n",
       "    }\n",
       "\n",
       "    for (var i = 0; i < css_urls.length; i++) {\n",
       "      var url = css_urls[i];\n",
       "      const element = document.createElement(\"link\");\n",
       "      element.onload = on_load;\n",
       "      element.onerror = on_error;\n",
       "      element.rel = \"stylesheet\";\n",
       "      element.type = \"text/css\";\n",
       "      element.href = url;\n",
       "      console.debug(\"Bokeh: injecting link tag for BokehJS stylesheet: \", url);\n",
       "      document.body.appendChild(element);\n",
       "    }\n",
       "\n",
       "    for (var i = 0; i < js_urls.length; i++) {\n",
       "      var url = js_urls[i];\n",
       "      var element = document.createElement('script');\n",
       "      element.onload = on_load;\n",
       "      element.onerror = on_error;\n",
       "      element.async = false;\n",
       "      element.src = url;\n",
       "      console.debug(\"Bokeh: injecting script tag for BokehJS library: \", url);\n",
       "      document.head.appendChild(element);\n",
       "    }\n",
       "  };var element = document.getElementById(\"1001\");\n",
       "  if (element == null) {\n",
       "    console.error(\"Bokeh: ERROR: autoload.js configured with elementid '1001' but no matching script tag was found. \")\n",
       "    return false;\n",
       "  }\n",
       "\n",
       "  function inject_raw_css(css) {\n",
       "    const element = document.createElement(\"style\");\n",
       "    element.appendChild(document.createTextNode(css));\n",
       "    document.body.appendChild(element);\n",
       "  }\n",
       "\n",
       "  var js_urls = [\"https://cdn.pydata.org/bokeh/release/bokeh-1.3.4.min.js\", \"https://cdn.pydata.org/bokeh/release/bokeh-widgets-1.3.4.min.js\", \"https://cdn.pydata.org/bokeh/release/bokeh-tables-1.3.4.min.js\", \"https://cdn.pydata.org/bokeh/release/bokeh-gl-1.3.4.min.js\"];\n",
       "  var css_urls = [];\n",
       "\n",
       "  var inline_js = [\n",
       "    function(Bokeh) {\n",
       "      Bokeh.set_log_level(\"info\");\n",
       "    },\n",
       "    \n",
       "    function(Bokeh) {\n",
       "      \n",
       "    },\n",
       "    function(Bokeh) {} // ensure no trailing comma for IE\n",
       "  ];\n",
       "\n",
       "  function run_inline_js() {\n",
       "    \n",
       "    if ((root.Bokeh !== undefined) || (force === true)) {\n",
       "      for (var i = 0; i < inline_js.length; i++) {\n",
       "        inline_js[i].call(root, root.Bokeh);\n",
       "      }if (force === true) {\n",
       "        display_loaded();\n",
       "      }} else if (Date.now() < root._bokeh_timeout) {\n",
       "      setTimeout(run_inline_js, 100);\n",
       "    } else if (!root._bokeh_failed_load) {\n",
       "      console.log(\"Bokeh: BokehJS failed to load within specified timeout.\");\n",
       "      root._bokeh_failed_load = true;\n",
       "    } else if (force !== true) {\n",
       "      var cell = $(document.getElementById(\"1001\")).parents('.cell').data().cell;\n",
       "      cell.output_area.append_execute_result(NB_LOAD_WARNING)\n",
       "    }\n",
       "\n",
       "  }\n",
       "\n",
       "  if (root._bokeh_is_loading === 0) {\n",
       "    console.debug(\"Bokeh: BokehJS loaded, going straight to plotting\");\n",
       "    run_inline_js();\n",
       "  } else {\n",
       "    load_libs(css_urls, js_urls, function() {\n",
       "      console.debug(\"Bokeh: BokehJS plotting callback run at\", now());\n",
       "      run_inline_js();\n",
       "    });\n",
       "  }\n",
       "}(window));"
      ],
      "application/vnd.bokehjs_load.v0+json": "\n(function(root) {\n  function now() {\n    return new Date();\n  }\n\n  var force = true;\n\n  if (typeof root._bokeh_onload_callbacks === \"undefined\" || force === true) {\n    root._bokeh_onload_callbacks = [];\n    root._bokeh_is_loading = undefined;\n  }\n\n  \n\n  \n  if (typeof (root._bokeh_timeout) === \"undefined\" || force === true) {\n    root._bokeh_timeout = Date.now() + 5000;\n    root._bokeh_failed_load = false;\n  }\n\n  var NB_LOAD_WARNING = {'data': {'text/html':\n     \"<div style='background-color: #fdd'>\\n\"+\n     \"<p>\\n\"+\n     \"BokehJS does not appear to have successfully loaded. If loading BokehJS from CDN, this \\n\"+\n     \"may be due to a slow or bad network connection. Possible fixes:\\n\"+\n     \"</p>\\n\"+\n     \"<ul>\\n\"+\n     \"<li>re-rerun `output_notebook()` to attempt to load from CDN again, or</li>\\n\"+\n     \"<li>use INLINE resources instead, as so:</li>\\n\"+\n     \"</ul>\\n\"+\n     \"<code>\\n\"+\n     \"from bokeh.resources import INLINE\\n\"+\n     \"output_notebook(resources=INLINE)\\n\"+\n     \"</code>\\n\"+\n     \"</div>\"}};\n\n  function display_loaded() {\n    var el = document.getElementById(\"1001\");\n    if (el != null) {\n      el.textContent = \"BokehJS is loading...\";\n    }\n    if (root.Bokeh !== undefined) {\n      if (el != null) {\n        el.textContent = \"BokehJS \" + root.Bokeh.version + \" successfully loaded.\";\n      }\n    } else if (Date.now() < root._bokeh_timeout) {\n      setTimeout(display_loaded, 100)\n    }\n  }\n\n\n  function run_callbacks() {\n    try {\n      root._bokeh_onload_callbacks.forEach(function(callback) {\n        if (callback != null)\n          callback();\n      });\n    } finally {\n      delete root._bokeh_onload_callbacks\n    }\n    console.debug(\"Bokeh: all callbacks have finished\");\n  }\n\n  function load_libs(css_urls, js_urls, callback) {\n    if (css_urls == null) css_urls = [];\n    if (js_urls == null) js_urls = [];\n\n    root._bokeh_onload_callbacks.push(callback);\n    if (root._bokeh_is_loading > 0) {\n      console.debug(\"Bokeh: BokehJS is being loaded, scheduling callback at\", now());\n      return null;\n    }\n    if (js_urls == null || js_urls.length === 0) {\n      run_callbacks();\n      return null;\n    }\n    console.debug(\"Bokeh: BokehJS not loaded, scheduling load and callback at\", now());\n    root._bokeh_is_loading = css_urls.length + js_urls.length;\n\n    function on_load() {\n      root._bokeh_is_loading--;\n      if (root._bokeh_is_loading === 0) {\n        console.debug(\"Bokeh: all BokehJS libraries/stylesheets loaded\");\n        run_callbacks()\n      }\n    }\n\n    function on_error() {\n      console.error(\"failed to load \" + url);\n    }\n\n    for (var i = 0; i < css_urls.length; i++) {\n      var url = css_urls[i];\n      const element = document.createElement(\"link\");\n      element.onload = on_load;\n      element.onerror = on_error;\n      element.rel = \"stylesheet\";\n      element.type = \"text/css\";\n      element.href = url;\n      console.debug(\"Bokeh: injecting link tag for BokehJS stylesheet: \", url);\n      document.body.appendChild(element);\n    }\n\n    for (var i = 0; i < js_urls.length; i++) {\n      var url = js_urls[i];\n      var element = document.createElement('script');\n      element.onload = on_load;\n      element.onerror = on_error;\n      element.async = false;\n      element.src = url;\n      console.debug(\"Bokeh: injecting script tag for BokehJS library: \", url);\n      document.head.appendChild(element);\n    }\n  };var element = document.getElementById(\"1001\");\n  if (element == null) {\n    console.error(\"Bokeh: ERROR: autoload.js configured with elementid '1001' but no matching script tag was found. \")\n    return false;\n  }\n\n  function inject_raw_css(css) {\n    const element = document.createElement(\"style\");\n    element.appendChild(document.createTextNode(css));\n    document.body.appendChild(element);\n  }\n\n  var js_urls = [\"https://cdn.pydata.org/bokeh/release/bokeh-1.3.4.min.js\", \"https://cdn.pydata.org/bokeh/release/bokeh-widgets-1.3.4.min.js\", \"https://cdn.pydata.org/bokeh/release/bokeh-tables-1.3.4.min.js\", \"https://cdn.pydata.org/bokeh/release/bokeh-gl-1.3.4.min.js\"];\n  var css_urls = [];\n\n  var inline_js = [\n    function(Bokeh) {\n      Bokeh.set_log_level(\"info\");\n    },\n    \n    function(Bokeh) {\n      \n    },\n    function(Bokeh) {} // ensure no trailing comma for IE\n  ];\n\n  function run_inline_js() {\n    \n    if ((root.Bokeh !== undefined) || (force === true)) {\n      for (var i = 0; i < inline_js.length; i++) {\n        inline_js[i].call(root, root.Bokeh);\n      }if (force === true) {\n        display_loaded();\n      }} else if (Date.now() < root._bokeh_timeout) {\n      setTimeout(run_inline_js, 100);\n    } else if (!root._bokeh_failed_load) {\n      console.log(\"Bokeh: BokehJS failed to load within specified timeout.\");\n      root._bokeh_failed_load = true;\n    } else if (force !== true) {\n      var cell = $(document.getElementById(\"1001\")).parents('.cell').data().cell;\n      cell.output_area.append_execute_result(NB_LOAD_WARNING)\n    }\n\n  }\n\n  if (root._bokeh_is_loading === 0) {\n    console.debug(\"Bokeh: BokehJS loaded, going straight to plotting\");\n    run_inline_js();\n  } else {\n    load_libs(css_urls, js_urls, function() {\n      console.debug(\"Bokeh: BokehJS plotting callback run at\", now());\n      run_inline_js();\n    });\n  }\n}(window));"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "\"\"\" Definition of general variables used in all bokeh Maps/Plots \"\"\"\n",
    "\n",
    "# Chose color palette\n",
    "import bokeh.palettes as palettes\n",
    "\n",
    "from bokeh.palettes import Viridis256 as palette\n",
    "palette = tuple(reversed(palette))\n",
    "\n",
    "# Other palettes\n",
    "from bokeh.palettes import Inferno256 as palette_inferno\n",
    "palette_inferno = tuple(reversed(palette_inferno))\n",
    "\n",
    "from bokeh.palettes import RdYlGn as palette_11\n",
    "palette_11 = palette_11[11]\n",
    "\n",
    "\n",
    "# List of tools used in all Bokeh plots/maps\n",
    "TOOLS = \"pan,wheel_zoom,reset,hover,save\"\n",
    "\n",
    "\n",
    "\"\"\" To see the plot in Jupyter Notebook\"\"\"\n",
    "output_notebook()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <span style=\"color:#008000\"> General function for interactive multi-line plot in Bokeh"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" -------------------- Funtion to plot multi lines --------------------  \"\"\"\n",
    "\n",
    "def plot_bokeh_multi_lines(dataframes, title, \n",
    "                           dates_str,\n",
    "                           label_modes,\n",
    "                           line_width=1, size=8, plot_width=300, plot_height=220, \n",
    "                           tools=\"pan,wheel_zoom,reset,save\", toolbar_location='below'):\n",
    "    \n",
    "    \"\"\" ----------- Dictionnary of data to pass to bokeh as argument and variables -----------  \"\"\"\n",
    "    df=dataframes[0]\n",
    "    source = ColumnDataSource(df)\n",
    "    col_names = source.column_names[1:-1] #not to take the index or integer column\n",
    "    \n",
    "    # Secondary source for visible circles\n",
    "    visible_circles = df[df.integer == 0]\n",
    "    source_visible=ColumnDataSource(visible_circles)\n",
    "\n",
    "    \"\"\" ----------- Create a palette of colors -----------  \"\"\"\n",
    "    colors = palettes.all_palettes['Category10'][max(3,len(col_names))][:len(col_names)]\n",
    "    \n",
    "    \"\"\" ----------- Create the figure with lines and circles -----------  \"\"\"\n",
    "    p = figure(title=title,\n",
    "               x_axis_type=\"datetime\",\n",
    "               plot_width=plot_width, plot_height=plot_height,\n",
    "                   sizing_mode=\"scale_width\",\n",
    "               toolbar_location=toolbar_location, \n",
    "               tools=tools)\n",
    "    \n",
    "    # Create dictionary to store lines\n",
    "    p_dict = dict() \n",
    "    \n",
    "    for col, c, col_name in zip(df.columns[:-1], #not to take integer column\n",
    "                                colors, col_names):\n",
    "        # Add lines\n",
    "        p_dict[col_name] = p.line('date', col, source=source, \n",
    "                                  color=c, line_width=line_width,\n",
    "                                  muted_alpha=0.2)\n",
    "        # Add circles\n",
    "        p.circle('date', col, source=source_visible, \n",
    "                color=c, size=size,\n",
    "                muted_alpha=0.2)\n",
    "        \n",
    "        # Add hover effect\n",
    "        p.add_tools(HoverTool(\n",
    "                            renderers=[p_dict[col_name]],\n",
    "                            tooltips=[('date','@date{%Y-%m-%d}'),(col, f'@{col}')],\n",
    "                            formatters={'@date': 'datetime'}))\n",
    "\n",
    "    # Add legends to items as tuple containing both the name of the column and its line\n",
    "    legend = Legend(items=[(x, [p_dict[x]]) for x in p_dict]) \n",
    "    p.add_layout(legend,'right')\n",
    "    \n",
    "    # Assign the click policy \n",
    "    p.legend.click_policy=\"mute\" \n",
    "        \n",
    "  \n",
    "    \"\"\" ----------- Set up callbacks -----------  \"\"\"\n",
    "    # Update Map while changing feature or mode\n",
    "    def update_map(attrname, old, new):\n",
    "        # Switch current dataframe\n",
    "        df=dataframes[button_mode.active]\n",
    "        source.data = df\n",
    "        \n",
    "        int_date = date_slider.value  #Retrieve current date\n",
    "        visible_circles = df[(df.integer == int_date)]\n",
    "        source_visible.data = visible_circles\n",
    "            \n",
    "            \n",
    "    # Update the slider on a click of the user\n",
    "    def slider_update(attrname, old, new):\n",
    "        # Retrieve current date\n",
    "        int_date = date_slider.value\n",
    "        date_slider.title = \"Date :\" + dates_str[int_date]\n",
    "        \n",
    "        # Change data\n",
    "        df=dataframes[0]\n",
    "        if len(label_modes)>0:\n",
    "            df=dataframes[button_mode.active]\n",
    "        visible_circles = df[(df['integer'] == int_date)]\n",
    "        source_visible.data = visible_circles\n",
    "        \n",
    "    \n",
    "    # Update the slider automatically when clicking on the animation button\n",
    "    def animate_update():\n",
    "        int_date = date_slider.value + 1\n",
    "        # Go back to the beginning if slider reach the end\n",
    "        if int_date > len(dates_str) - 1 :\n",
    "            int_date = 0\n",
    "        date_slider.value = int_date\n",
    "\n",
    "\n",
    "    # Change the button label depending if it is playing or not\n",
    "    callback_id = None\n",
    "    def animate():\n",
    "        global callback_id\n",
    "        if button_animation.label == '► Play':\n",
    "            button_animation.label = '❚❚ Pause'\n",
    "            callback_id = curdoc().add_periodic_callback(animate_update, 200)\n",
    "        else:\n",
    "            button_animation.label = '► Play'\n",
    "            curdoc().remove_periodic_callback(callback_id)\n",
    "            \n",
    "            \n",
    "    \"\"\" ----------- Set up widgets and events -----------  \"\"\"\n",
    "    # Slider for the date\n",
    "    date_slider = Slider(title=\"Date \"+ dates_str[0], start = 0, end = len(dates_str)-1, value = 0, step=1, show_value=False)\n",
    "    date_slider.on_change('value', slider_update)\n",
    "            \n",
    "    # Button to animate the slider\n",
    "    button_animation = Button(label='► Play', width=60)\n",
    "    button_animation.on_click(animate)\n",
    "\n",
    "    # Set up inputs for layout\n",
    "    inputs = widgetbox(date_slider, button_animation)\n",
    "    \n",
    "    # Set up Button to choose the displayed mode if they exists\n",
    "    if len(label_modes)>0:\n",
    "        button_mode = RadioButtonGroup(labels= label_modes, active=0)\n",
    "        button_mode.on_change('active', update_map)\n",
    "        # As the RadioButton isn't implemented with a title option, we add a PreText object before it\n",
    "        title_button_mode = PreText(text= \"Choose feature\", style={'font-size':'12pt', \n",
    "                         'color': 'black', \n",
    "                         'font-family': 'sans-serif'})\n",
    "        # update inputs\n",
    "        inputs = widgetbox(title_button_mode, button_mode, date_slider, button_animation)\n",
    "        \n",
    "    # Set up layouts and add to document\n",
    "    layout_doc = layout(p, inputs, sizing_mode=\"scale_width\")\n",
    "     \n",
    "\n",
    "    \n",
    "    \"\"\" Run the interaction in the bokeh Map in the Jupyter Notebook \"\"\"\n",
    "    # All server related code is added in the modify_doc function\n",
    "    def modify_doc(doc): # To run the code in a bokeh server, simply replace calls to modify_doc to calls to curdoc() \n",
    "        doc.add_root(layout_doc)\n",
    "        doc.title = title\n",
    "\n",
    "    handler = FunctionHandler(modify_doc)\n",
    "    app = Application(handler)\n",
    "    show(app, notebook_url = notebook_url)\n",
    "    \n",
    "    \"\"\"\n",
    "    # If you want to display this code on a web browser with a bokeh server, replace the previous seciton with this one\n",
    "    # and remove the output_notebook() at the beginning of the code\n",
    "    \n",
    "    curdoc().add_root(layout)\n",
    "    curdoc().title = title\n",
    "    \"\"\" "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <span style=\"color:#008000\"> Function to compute Bokeh Map for each set of data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <span style=\"color:#008000\">a- Hospital Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" -------------------- Funtion to handle the plot of all bokeh maps --------------------  \"\"\"\n",
    "def plot_all_bokeh_map_hospital(title, \n",
    "                                data_rate, data_min_max, \n",
    "                                name_tooltips,\n",
    "                               TypeColorMapper=LinearColorMapper,\n",
    "                               plot_width=450, plot_height=400, \n",
    "                               tools=TOOLS, palette=palette):\n",
    "        \n",
    "    \n",
    "    \"\"\" ----------- Create the color mapper -----------  \"\"\"\n",
    "    color_mapper = TypeColorMapper(palette=palette, low=data_min_max['synthesis']['hospitalized'][\"min\"], high=data_min_max['synthesis']['hospitalized'][\"max\"])\n",
    "\n",
    "    \n",
    "    \"\"\" ----------- Dictionnary of data to pass to bokeh as argument and variables -----------  \"\"\"\n",
    "    \n",
    "    data=dict(x = lon_by_department,\n",
    "              y = lat_by_department,\n",
    "              name = name_department,\n",
    "              rate = data_rate[\"synthesis\"]['hospitalized'])\n",
    "    \n",
    "    source=ColumnDataSource(data)\n",
    "    \n",
    "    \"\"\" ----------- Create the figure -----------  \"\"\"\n",
    "    p = figure(title=title, tools=tools,\n",
    "               x_axis_location=None, y_axis_location=None,\n",
    "               plot_width=plot_width, plot_height=plot_height,\n",
    "               tooltips=[(\"Name\", \"@name\"), (name_tooltips, \"@rate\"), (\"(Long, Lat)\", \"($x, $y)\")]\n",
    "              )\n",
    "    p.grid.grid_line_color = None\n",
    "    p.hover.point_policy = \"follow_mouse\"\n",
    "\n",
    "    p.patches('x', 'y', source=source,\n",
    "              fill_color={'field': 'rate', 'transform': color_mapper},\n",
    "              fill_alpha=0.7, line_color=\"black\", line_width=0.5)\n",
    "\n",
    "    #Add color bar to figure\n",
    "    color_bar = ColorBar(color_mapper=color_mapper, ticker=LogTicker(),\n",
    "                         label_standoff=12, border_line_color=None, location=(0,0))\n",
    "\n",
    "    p.add_layout(color_bar, 'right')\n",
    "    \n",
    "    \n",
    "    \"\"\" ----------- Set up callbacks -----------  \"\"\"\n",
    "    # Update Map while changing feature or mode\n",
    "    def update_map(attrname, old, new):\n",
    "        \n",
    "        # Retrieve the new feature and mode to use\n",
    "        feature = label_features[button_choice.active] \n",
    "        mode = label_modes[button_mode.active] \n",
    "        \n",
    "        # Update the title of the graph\n",
    "        title_update(feature, mode)\n",
    "        \n",
    "        # Update the color map\n",
    "        color_mapper_update(feature, mode)\n",
    "        \n",
    "        # Update map\n",
    "        if mode == \"synthesis\":\n",
    "            source.data['rate'] = data_rate[mode][feature]\n",
    "            date_slider.visible = False # Hide timeline slider\n",
    "            button_animation.visible = False # Hide animation button\n",
    "        else:\n",
    "            int_date = date_slider.value # Get integer for the currently chosen date in the slider\n",
    "            source.data['rate'] = data_rate[mode][feature][int_date]\n",
    "            date_slider.visible = True   # Show timeline\n",
    "            button_animation.visible = True # Show animation button\n",
    "    \n",
    "    \n",
    "    # Update title of map when feature or mode is changed\n",
    "    def title_update(feature, mode):\n",
    "        if mode == \"synthesis\" or mode == \"by_date\":\n",
    "            if feature == \"hospitalized\":\n",
    "                first_word = \"Hospitalized\"\n",
    "            if feature == \"intensive_care\":\n",
    "                first_word = \"Intensive-cares\"\n",
    "            if feature == \"death\":\n",
    "                first_word = \"Deaths\"\n",
    "            if mode == \"by_date\":\n",
    "                new_title = first_word + \" by day since the beginning of Covid-19 spread\"\n",
    "            else:\n",
    "                new_title = first_word + \" since the beginning of Covid-19 spread\"\n",
    "        else:\n",
    "            if feature == \"hospitalized\":\n",
    "                second_word = \"hospitalized\"\n",
    "            if feature == \"intensive_care\":\n",
    "                second_word = \"intensive-cares\"\n",
    "            if feature == \"death\":\n",
    "                second_word = \"deaths\"\n",
    "            new_title = \"Cumulated \" + second_word +  \" since the beginning of Covid-19 spread\"\n",
    "        p.title.text = new_title\n",
    "                \n",
    "      \n",
    "        \n",
    "    # Update color mapper of map when features or mode displayed are changed\n",
    "    def color_mapper_update(feature, mode):\n",
    "        # Change the limits of the Color Mapper\n",
    "        color_mapper.update(low = data_min_max[mode][feature][\"min\"], high = data_min_max[mode][feature][\"max\"])\n",
    "            \n",
    "    \n",
    "    # Update the slider on a click of the user\n",
    "    def slider_update(attrname, old, new):\n",
    "        # Retrieve current date\n",
    "        int_date = date_slider.value \n",
    "        date_slider.title = \"Date :\" + hospital_dates_str[int_date]\n",
    "        \n",
    "        # Retrieve current feature and mode label\n",
    "        mode = label_modes[button_mode.active]\n",
    "        feature = label_features[button_choice.active]\n",
    "        \n",
    "        # Change data\n",
    "        source.data['rate'] = data_rate[mode][feature][int_date]\n",
    "    \n",
    "    \n",
    "    # Update the slider automatically when clicking on the animation button\n",
    "    def animate_update():\n",
    "        int_date = date_slider.value + 1\n",
    "        # Go back to the beginning if slider reach the end\n",
    "        if int_date > len(hospital_dates_str) - 1 :\n",
    "            int_date = 0\n",
    "        date_slider.value = int_date\n",
    "\n",
    "\n",
    "    # Change the button label depending if it is playing or not\n",
    "    callback_id = None\n",
    "    def animate():\n",
    "        global callback_id\n",
    "        if button_animation.label == '► Play':\n",
    "            button_animation.label = '❚❚ Pause'\n",
    "            callback_id = curdoc().add_periodic_callback(animate_update, 200)\n",
    "        else:\n",
    "            button_animation.label = '► Play'\n",
    "            curdoc().remove_periodic_callback(callback_id)\n",
    "        \n",
    "        \n",
    "    \"\"\" ----------- Set up widgets and events -----------  \"\"\"\n",
    "    # Button to choose the feature to display in the map\n",
    "    label_features = [\"hospitalized\", \"intensive_care\", \"death\"] # labels of possible features\n",
    "    button_choice = RadioButtonGroup(labels=[\"Hospitalized\", \"Intensive Care\", \"Death\"], active=0)\n",
    "    button_choice.on_change('active', update_map)\n",
    "    # As the RadioButton isn't implemented with a title option, we add a PreText object before it\n",
    "    title_button_choice = PreText(text= \"Choose the feature to display\", style={'font-size':'12pt', \n",
    "                     'color': 'black', \n",
    "                     'font-family': 'sans-serif'})\n",
    "    \n",
    "    \n",
    "    # Button to choose the displayed mode: synthesis, by_date, cumulated\n",
    "    label_modes = [\"synthesis\", \"by_date\", \"cumulated\"] # labels of possible features\n",
    "    button_mode = RadioButtonGroup(labels= [\"Synthesis\", \"By Day\", \"Cumulated\"], active=0)\n",
    "    button_mode.on_change('active', update_map)\n",
    "    # As the RadioButton isn't implemented with a title option, we add a PreText object before it\n",
    "    title_button_mode = PreText(text= \"Choose display mode \", style={'font-size':'12pt', \n",
    "                     'color': 'black', \n",
    "                     'font-family': 'sans-serif'})\n",
    "    \n",
    "    # Slider for the date\n",
    "    date_slider = Slider(title=\"Date \"+ hospital_dates_str[0], start = 0, end = len(hospital_dates_str)-1, value = 0, step=1, show_value=False)\n",
    "    date_slider.on_change('value', slider_update)\n",
    "    date_slider.visible = False # Initially the slider is not visible\n",
    "\n",
    "    # Button to animate the slider\n",
    "    button_animation = Button(label='► Play', width=60)\n",
    "    button_animation.on_click(animate)\n",
    "    button_animation.visible = False # Initially the slider is not visible\n",
    "       \n",
    "\n",
    "    # Set up layouts and add to document\n",
    "    inputs = widgetbox(title_button_choice, button_choice, title_button_mode, button_mode, date_slider, button_animation)\n",
    "    layout = row(p, inputs)\n",
    "    \n",
    "\n",
    "    \n",
    "    \"\"\" Run the interaction in the bokeh Map in the Jupyter Notebook \"\"\"\n",
    "    \n",
    "    # All server related code is added in the modify_doc function\n",
    "    def modify_doc(doc): # To run the code in a bokeh server, simply replace calls to modify_doc to calls to curdoc() \n",
    "        doc.add_root(layout)\n",
    "        doc.title = title\n",
    "\n",
    "\n",
    "    handler = FunctionHandler(modify_doc)\n",
    "    app = Application(handler)\n",
    "    show(app, notebook_url = notebook_url)\n",
    "    \n",
    "    \n",
    "    \"\"\"\n",
    "    # If you want to display this code on a web browser with a bokeh server, replace the previous seciton with this one\n",
    "    # and remove the output_notebook() at the beginning of the code\n",
    "    \n",
    "    curdoc().add_root(layout)\n",
    "    curdoc().title = title\n",
    "    \"\"\" "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <span style=\"color:#008000\">b- Energy data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" -------------------- Funtion to handle the plot of all bokeh maps --------------------  \"\"\"\n",
    "def plot_all_bokeh_map_energy(title, \n",
    "                                data_rate, data_min_max, \n",
    "                                name_tooltips,\n",
    "                               TypeColorMapper=LinearColorMapper,\n",
    "                               plot_width=450, plot_height=400, \n",
    "                               tools=TOOLS, palette=palette_inferno):\n",
    "        \n",
    "    \n",
    "    \"\"\" ----------- Create the color mapper -----------  \"\"\"\n",
    "    color_mapper = TypeColorMapper(palette=palette, low=data_min_max['by_date']['general_consumption']['min'], high=data_min_max['by_date']['general_consumption']['max'])\n",
    "\n",
    "    \n",
    "    \"\"\" ----------- Dictionnary of data to pass to bokeh as argument and variables -----------  \"\"\"\n",
    "    mode = 'by_date' #mode by default and the only one implemented yet\n",
    "    \n",
    "    data=dict(x = lon_by_department,\n",
    "              y = lat_by_department,\n",
    "              name = name_department,\n",
    "              rate = data_rate[\"by_date\"]['general_consumption'][0])\n",
    "    \n",
    "    source=ColumnDataSource(data)\n",
    "    \n",
    "    \"\"\" ----------- Create the figure -----------  \"\"\"\n",
    "    p = figure(title=title, tools=tools,\n",
    "               x_axis_location=None, y_axis_location=None,\n",
    "               plot_width=plot_width, plot_height=plot_height,\n",
    "               tooltips=[(\"Name\", \"@name\"), (name_tooltips, \"@rate\"), (\"(Long, Lat)\", \"($x, $y)\")]\n",
    "              )\n",
    "    p.grid.grid_line_color = None\n",
    "    p.hover.point_policy = \"follow_mouse\"\n",
    "\n",
    "    p.patches('x', 'y', source=source,\n",
    "              fill_color={'field': 'rate', 'transform': color_mapper},\n",
    "              fill_alpha=0.7, line_color=\"black\", line_width=0.5)\n",
    "\n",
    "    #Add color bar to figure\n",
    "    color_bar = ColorBar(color_mapper=color_mapper, ticker=LogTicker(),\n",
    "                         label_standoff=12, border_line_color=None, location=(0,0))\n",
    "\n",
    "    p.add_layout(color_bar, 'right')\n",
    "    \n",
    "    \n",
    "    \"\"\" ----------- Set up callbacks -----------  \"\"\"\n",
    "    # Update Map while changing feature or mode\n",
    "    def update_map(attrname, old, new):\n",
    "        \n",
    "        # Retrieve the new feature to use\n",
    "        feature = label_features[button_choice.active] \n",
    "        \n",
    "        # Update the title of the graph\n",
    "        title_update(feature, mode)\n",
    "        \n",
    "        # Update the color map\n",
    "        color_mapper_update(feature, mode)\n",
    "        \n",
    "        # Update map\n",
    "        int_date = date_slider.value # Get integer for the currently chosen date in the slider\n",
    "        source.data['rate'] = data_rate[mode][feature][int_date]\n",
    "\n",
    "    \n",
    "    \n",
    "    # Update title of map when feature or mode is changed\n",
    "    def title_update(feature, mode):\n",
    "        if mode == 'by_date':\n",
    "            if feature == \"general_consumption\":\n",
    "                new_title = \"General energy consumption per Day\"\n",
    "            if feature == \"thermal_power\":\n",
    "                new_title = \"Thermal power production per Day\"\n",
    "            if feature == \"nuclear_power\":\n",
    "                new_title = \"Nuclear power production per Day\"\n",
    "        p.title.text = new_title\n",
    "                \n",
    "      \n",
    "        \n",
    "    # Update color mapper of map when features or mode displayed are changed\n",
    "    def color_mapper_update(feature, mode):\n",
    "        # Change the limits of the Color Mapper\n",
    "        color_mapper.update(low = data_min_max[mode][feature][\"min\"], high = data_min_max[mode][feature][\"max\"])\n",
    "            \n",
    "    \n",
    "    # Update the slider on a click of the user\n",
    "    def slider_update(attrname, old, new):\n",
    "        # Retrieve current date\n",
    "        int_date = date_slider.value \n",
    "        date_slider.title = \"Date :\" + energy_dates_str[int_date]\n",
    "        \n",
    "        # Retrieve current feature Label\n",
    "        feature = label_features[button_choice.active]\n",
    "        \n",
    "        # Change data\n",
    "        source.data['rate'] = data_rate[mode][feature][int_date]\n",
    "    \n",
    "    \n",
    "    # Update the slider automatically when clicking on the animation button\n",
    "    def animate_update():\n",
    "        int_date = date_slider.value + 1\n",
    "        # Go back to the beginning if slider reach the end\n",
    "        if int_date > len(energy_dates_str) - 2 :\n",
    "            int_date = 0\n",
    "        date_slider.value = int_date\n",
    "\n",
    "\n",
    "    # Change the button label depending if it is playing or not\n",
    "    callback_id = None\n",
    "    def animate():\n",
    "        global callback_id\n",
    "        if button_animation.label == '► Play':\n",
    "            button_animation.label = '❚❚ Pause'\n",
    "            callback_id = curdoc().add_periodic_callback(animate_update, 200)\n",
    "        else:\n",
    "            button_animation.label = '► Play'\n",
    "            curdoc().remove_periodic_callback(callback_id)\n",
    "        \n",
    "        \n",
    "    \"\"\" ----------- Set up widgets and events -----------  \"\"\"\n",
    "    # Button to choose the feature to display in the map\n",
    "    label_features = [\"general_consumption\", \"thermal_power\", \"nuclear_power\"] # labels of possible features\n",
    "    button_choice = RadioButtonGroup(labels=[\"General Consumption\", \"Thermal Power\", \"Nuclear Power\"], active=0)\n",
    "    button_choice.on_change('active', update_map)\n",
    "    # As the RadioButton isn't implemented with a title option, we add a PreText object before it\n",
    "    title_button_choice = PreText(text= \"Choose the feature to display\", style={'font-size':'12pt', \n",
    "                     'color': 'black', \n",
    "                     'font-family': 'sans-serif'})\n",
    "    \n",
    "    \n",
    "    # Slider for the date\n",
    "    date_slider = Slider(title=\"Date \"+ energy_dates_str[0], start = 0, end = len(energy_dates_str)-2, value = 0, step=1, show_value=False)\n",
    "    date_slider.on_change('value', slider_update)\n",
    "\n",
    "    # Button to animate the slider\n",
    "    button_animation = Button(label='► Play', width=60)\n",
    "    button_animation.on_click(animate)\n",
    "    button_animation.visible = False # Initially the slider is not visible\n",
    "    \n",
    "    # Special mention \n",
    "    title_explanation = PreText(text= \"Departments in yellow have an unknown consumption\", style={'font-size':'9pt', \n",
    "                     'color': 'black', \n",
    "                     'font-family': 'sans-serif'})\n",
    "       \n",
    "\n",
    "    # Set up layouts and add to document\n",
    "    inputs = widgetbox(title_button_choice, button_choice, date_slider, button_animation, title_explanation)\n",
    "    layout = row(p, inputs)\n",
    "    \n",
    "\n",
    "    \n",
    "    \"\"\" Run the interaction in the bokeh Map in the Jupyter Notebook \"\"\"\n",
    "    \n",
    "    # All server related code is added in the modify_doc function\n",
    "    def modify_doc(doc): # To run the code in a bokeh server, simply replace calls to modify_doc to calls to curdoc() \n",
    "        doc.add_root(layout)\n",
    "        doc.title = title\n",
    "\n",
    "\n",
    "    handler = FunctionHandler(modify_doc)\n",
    "    app = Application(handler)\n",
    "    show(app, notebook_url = notebook_url)\n",
    "    \n",
    "    \n",
    "    \"\"\"\n",
    "    # If you want to display this code on a web browser with a bokeh server, replace the previous seciton with this one\n",
    "    # and remove the output_notebook() at the beginning of the code\n",
    "    \n",
    "    curdoc().add_root(layout)\n",
    "    curdoc().title = title\n",
    "    \"\"\" "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <span style=\"color:#008000\">c- Airline traffic"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" -------------------- Funtion to handle the plot the air pollution bokeh Map --------------------  \"\"\"\n",
    "def plot_bokeh_map_flights(title, \n",
    "                                data_rate, \n",
    "                                data_size, \n",
    "                               plot_width=450, plot_height=400, \n",
    "                               tools=TOOLS):\n",
    "        \n",
    "\n",
    "    \"\"\" ----------- Dictionnary of data to pass to bokeh as argument and variables -----------  \"\"\"\n",
    "    \n",
    "    # Data for the Shape of France on Map\n",
    "    data_shape =dict(x = lon_by_department,\n",
    "              y = lat_by_department,\n",
    "              name = name_department)\n",
    "    \n",
    "    source = ColumnDataSource(data_shape)\n",
    "    \n",
    "    \n",
    "    # Data for the Shape of France on Map\n",
    "    data_cities1 = dict(x = lon_city_airport,\n",
    "              y = lat_city_airport,\n",
    "              sizes = data_size[\"intranational\"][0],\n",
    "              name = airport_cities,\n",
    "              rate = data_rate[\"intranational\"][0])\n",
    "    data_cities = ColumnDataSource(data_cities1)\n",
    "    \n",
    "    \n",
    "\n",
    "    \"\"\" ----------- Create the figure -----------  \"\"\"\n",
    "    \n",
    "    # General render of the shape of France and its departments\n",
    "    p = figure(title=title, \n",
    "               # tools=tools, # Hide tooltip for departments\n",
    "               toolbar_location = 'below',\n",
    "               x_axis_location=None, y_axis_location=None,\n",
    "               plot_width=plot_width, plot_height=plot_height,\n",
    "               # tooltips=[(\"Department\", \"@name\"), (\"(Long, Lat)\", \"($x, $y)\")] \n",
    "              )\n",
    "    \n",
    "    p.xgrid.grid_line_color = None\n",
    "    p.ygrid.grid_line_color = None\n",
    "    # p.hover.point_policy = \"follow_mouse\"\n",
    "\n",
    "    p.patches('x', 'y', source=source,\n",
    "              fill_color=None, \n",
    "              line_color = 'gray', \n",
    "              line_width = 0.25, \n",
    "              fill_alpha = 1)\n",
    "\n",
    "    \n",
    "    # Add circle at the position of cities\n",
    "    # Their size depends of the air pollution rate\n",
    "    cities_points = p.circle('x', 'y', size = 'sizes', source = data_cities, \n",
    "                     color = 'red',  \n",
    "                     alpha = 0.3, \n",
    "                     name=\"circle\")\n",
    "    \n",
    "    \n",
    "    p.add_tools(HoverTool(renderers = [cities_points],\n",
    "                      tooltips = [('City', '@name'),\n",
    "                                  (\"(Long, Lat)\", \"($x, $y)\"),\n",
    "                                  ('Number of flights', '@rate'),]))\n",
    "    \n",
    "    \"\"\" ----------- Set up callbacks -----------  \"\"\"          \n",
    "    \n",
    "    # Update Map while changing feature or mode\n",
    "    def update_map(attrname, old, new):\n",
    "        \n",
    "        # Retrieve the new feature to use\n",
    "        feature = label_features[button_choice.active] \n",
    "        \n",
    "                \n",
    "        # Update map\n",
    "        int_date = date_slider.value # Get integer for the currently chosen date in the slider\n",
    "        data_cities.data['rate'] = data_rate[feature][int_date]\n",
    "        \n",
    "        # Change color of points in figure\n",
    "        if button_choice.active == 1:\n",
    "            cities_points.glyph.fill_color = 'blue'\n",
    "            cities_points.glyph.line_color = 'blue'\n",
    "        else:\n",
    "            cities_points.glyph.fill_color = 'red'\n",
    "            cities_points.glyph.line_color = 'red'\n",
    "        \n",
    "        \n",
    "    # Update the slider on a click of the user\n",
    "    def slider_update(attrname, old, new):\n",
    "        # Retrieve current date\n",
    "        int_date = date_slider.value \n",
    "        date_slider.title = \"Date :\" + flights_dates_str[int_date]\n",
    "        \n",
    "        # Retrieve current feature Label\n",
    "        feature = label_features[button_choice.active]\n",
    "        \n",
    "        # Change data\n",
    "        data_cities.data['rate'] = data_rate[feature][int_date]\n",
    "        data_cities.data['sizes'] = data_size[feature][int_date]\n",
    "        \n",
    "        # Change size of points in figure\n",
    "        cities_points.glyph.size = 'sizes'\n",
    "\n",
    "                \n",
    "    \n",
    "    # Update the slider automatically when clicking on the animation button\n",
    "    def animate_update():\n",
    "        int_date = date_slider.value + 1\n",
    "        # Go back to the beginning if slider reach the end\n",
    "        if int_date > len(flights_dates_str) - 1 :\n",
    "            int_date = 0\n",
    "        date_slider.value = int_date\n",
    "\n",
    "\n",
    "    # Change the button label depending if it is playing or not\n",
    "    callback_id = None\n",
    "    def animate():\n",
    "        global callback_id\n",
    "        if button_animation.label == '► Play':\n",
    "            button_animation.label = '❚❚ Pause'\n",
    "            callback_id = curdoc().add_periodic_callback(animate_update, 200)\n",
    "        else:\n",
    "            button_animation.label = '► Play'\n",
    "            curdoc().remove_periodic_callback(callback_id)\n",
    "\n",
    "        \n",
    "    \"\"\" ----------- Set up widgets and events -----------  \"\"\"\n",
    "    \n",
    "    # Button to choose the feature to display in the map\n",
    "    label_features = [\"intranational\", \"international\"] # labels of possible features\n",
    "    button_choice = RadioButtonGroup(labels=[\"Intranational\", \"International\"], active=0)\n",
    "    button_choice.on_change('active', update_map)\n",
    "    # As the RadioButton isn't implemented with a title option, we add a PreText object before it\n",
    "    title_button_choice = PreText(text= \"Choose the type of flights to visualize\", style={'font-size':'12pt', \n",
    "                     'color': 'black', \n",
    "                     'font-family': 'sans-serif'})\n",
    "    \n",
    "    # Description for Slider\n",
    "    slider_explanation = PreText(text= \"Slider per day\", style={'font-size':'9pt', \n",
    "                     'color': 'black', \n",
    "                     'font-family': 'sans-serif'})\n",
    "    \n",
    "    # Slider for the date\n",
    "    date_slider = Slider(title=\"Date \"+ flights_dates_str[0], start = 0, end = len(flights_dates_str)-1, value = 0, step=1, show_value=False)\n",
    "    date_slider.on_change('value', slider_update)\n",
    "\n",
    "    # Button to animate the slider\n",
    "    button_animation = Button(label='► Play', width=60)\n",
    "    button_animation.on_click(animate)\n",
    "    \n",
    "    title_lockdown = PreText(text= \"Lockdown begins in France on March 16\", style={'font-size':'12pt', \n",
    "                     'color': 'black', \n",
    "                     'font-family': 'sans-serif'})\n",
    "\n",
    "    # Set up layouts and add to document\n",
    "    inputs = widgetbox(title_button_choice, button_choice, slider_explanation, date_slider, button_animation, title_lockdown)\n",
    "    layout = row(p, inputs)\n",
    "    \n",
    "    \n",
    "    \n",
    "    \n",
    "    \"\"\" Run the interaction in the bokeh Map in the Jupyter Notebook \"\"\"\n",
    "    \n",
    "    # All server related code is added in the modify_doc function\n",
    "    def modify_doc(doc): # To run the code in a bokeh server, simply replace calls to modify_doc to calls to curdoc() \n",
    "        doc.add_root(layout)\n",
    "        doc.title = title\n",
    "\n",
    "\n",
    "    handler = FunctionHandler(modify_doc)\n",
    "    app = Application(handler)\n",
    "    show(app, notebook_url = notebook_url)\n",
    "    \n",
    "    \n",
    "    \"\"\"\n",
    "    # If you want to display this code on a web browser with a bokeh server, replace the previous section with this one\n",
    "    # and remove the output_notebook() at the beginning of the code\n",
    "    \n",
    "    curdoc().add_root(layout)\n",
    "    curdoc().title = title\n",
    "    \"\"\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <span style=\"color:#008000\">d- Air quality"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" -------------------- Funtion to handle the plot the air pollution bokeh Map --------------------  \"\"\"\n",
    "def plot_bokeh_map_air_polution_comparison(title,\n",
    "                                title1, title2, \n",
    "                                data_rate1,\n",
    "                                data_rate2,\n",
    "                                data_min_max,\n",
    "                                air_dates_str1,\n",
    "                                air_dates_str2,\n",
    "                                palette=palette_11, \n",
    "                                tools=TOOLS,\n",
    "                               plot_width=380, plot_height=400):\n",
    "        \n",
    "    \n",
    "    \"\"\" ----------- Create the color mapper -----------  \"\"\"\n",
    "    color_mapper = linear_cmap(field_name='rate', palette=palette ,low=data_min_max[\"pm10\"][\"min\"] ,high=data_min_max[\"pm10\"][\"max\"])\n",
    "\n",
    "\n",
    "    \"\"\" ----------- Dictionnary of data to pass to bokeh as argument and variables -----------  \"\"\"\n",
    "    \n",
    "    # Data for the Shape of France on Map\n",
    "    data_shape =dict(x = lon_by_department,\n",
    "              y = lat_by_department,\n",
    "              name = name_department)\n",
    "    \n",
    "    source = ColumnDataSource(data_shape)\n",
    "    \n",
    "    \n",
    "    # Data concerning corona map\n",
    "    data_cities1 = dict(x = lon_city,\n",
    "              y = lat_city,\n",
    "              name = air_cities,\n",
    "              rate = data_rate1[\"pm10\"][0])\n",
    "    data_cities_source1 = ColumnDataSource(data_cities1)\n",
    "    \n",
    "    \n",
    "    # Data concerning corona map\n",
    "    data_cities2 = dict(x = lon_city,\n",
    "              y = lat_city,\n",
    "              name = air_cities,\n",
    "              rate = data_rate2[\"pm10\"][0])\n",
    "    data_cities_source2 = ColumnDataSource(data_cities2)\n",
    "    \n",
    "    \n",
    "\n",
    "    \"\"\" ----------- Create the figures -----------  \"\"\"\n",
    "    \n",
    "    # General render of the shape of France and its departments\n",
    "    def create_figure_departments(title, data_cities, source = source, color_mapper = color_mapper):\n",
    "        p = figure(title=title, \n",
    "                   # tools=tools, # Hide tooltip for departments\n",
    "                   toolbar_location = 'below',\n",
    "                   x_axis_location=None, y_axis_location=None,\n",
    "                   plot_width=plot_width, plot_height=plot_height,\n",
    "                   sizing_mode=\"scale_width\",\n",
    "                   # tooltips=[(\"Department\", \"@name\"), (\"(Long, Lat)\", \"($x, $y)\")] \n",
    "                  )\n",
    "\n",
    "        p.xgrid.grid_line_color = None\n",
    "        p.ygrid.grid_line_color = None\n",
    "        # p.hover.point_policy = \"follow_mouse\"\n",
    "\n",
    "        p.patches('x', 'y', source=source,\n",
    "                  fill_color=None, \n",
    "                  line_color = 'gray', \n",
    "                  line_width = 0.25, \n",
    "                  fill_alpha = 1) \n",
    "        \n",
    "        # Add circle at the position of cities\n",
    "        # Their size depends of the air pollution rate\n",
    "        cities_points = p.circle('x', 'y', size = 25, source = data_cities, \n",
    "                         line_color = color_mapper, color = color_mapper, \n",
    "                         alpha = 0.45, \n",
    "                         name=\"circle\")\n",
    "        \n",
    "        p.add_tools(HoverTool(renderers = [cities_points],\n",
    "                      tooltips = [('City', '@name'),\n",
    "                                  (\"(Long, Lat)\", \"($x, $y)\"),\n",
    "                                  ('Pollutant concentration', '@rate')]))\n",
    "    \n",
    "        return(p, cities_points)\n",
    "    \n",
    "    \n",
    "    figure1, cities_points1 = create_figure_departments(title1, data_cities_source1)\n",
    "    figure2, cities_points2 = create_figure_departments(title2, data_cities_source2)\n",
    "    \n",
    "    \n",
    "    #Add color to only one of the figures\n",
    "    color_bar = ColorBar(color_mapper=color_mapper['transform'],\n",
    "                         label_standoff=12, border_line_color=None, location=(0,0))\n",
    "\n",
    "    figure2.add_layout(color_bar, 'right')\n",
    "      \n",
    "    \n",
    "    \n",
    "    \n",
    "    \"\"\" ----------- Set up callbacks -----------  \"\"\"          \n",
    "    \n",
    "    # Update Map while changing feature or mode\n",
    "    def update_map(attrname, old, new):\n",
    "        \n",
    "        # Retrieve the new feature to use\n",
    "        feature = label_features[button_choice.active] \n",
    "        \n",
    "        # Update color mapper\n",
    "        color_mapper_update(feature)\n",
    "        \n",
    "        # Update maps\n",
    "        int_date = date_slider.value # Get integer for the currently chosen date in the slider\n",
    "        data_cities_source1.data['rate'] = data_rate1[feature][int_date]\n",
    "        data_cities_source2.data['rate'] = data_rate2[feature][int_date]\n",
    "        \n",
    "        \n",
    "    # Update color mapper of map when features displayed are changed\n",
    "    def color_mapper_update(feature):\n",
    "        # Change the limits of the Color Mapper\n",
    "        color_mapper['transform'].low=data_min_max[feature][\"min\"]\n",
    "        color_mapper['transform'].high=data_min_max[feature][\"max\"]\n",
    "    \n",
    "    \n",
    "    # Update the slider on a click of the user\n",
    "    def slider_update(attrname, old, new):\n",
    "        # Retrieve current date\n",
    "        int_date = date_slider.value \n",
    "        date_slider.title = \"Date Corona: \" + air_dates_str1[int_date] + \"    - Date 2019: \" + air_dates_str2[int_date]\n",
    "        \n",
    "        # Retrieve current feature Label\n",
    "        feature = label_features[button_choice.active]\n",
    "        \n",
    "        # Change data\n",
    "        data_cities_source1.data['rate'] = data_rate1[feature][int_date]\n",
    "        data_cities_source2.data['rate'] = data_rate2[feature][int_date]\n",
    "        \n",
    "        # Change size of points in figure\n",
    "        # cities_points.glyph.size = 'sizes'\n",
    "\n",
    "                \n",
    "    \n",
    "    # Update the slider automatically when clicking on the animation button\n",
    "    def animate_update():\n",
    "        int_date = date_slider.value + 1\n",
    "        # Go back to the beginning if slider reach the end\n",
    "        if int_date > len(air_dates_str1) - 1 :\n",
    "            int_date = 0\n",
    "        date_slider.value = int_date\n",
    "\n",
    "\n",
    "    # Change the button label depending if it is playing or not\n",
    "    callback_id = None\n",
    "    def animate():\n",
    "        global callback_id\n",
    "        if button_animation.label == '► Play':\n",
    "            button_animation.label = '❚❚ Pause'\n",
    "            callback_id = curdoc().add_periodic_callback(animate_update, 200)\n",
    "        else:\n",
    "            button_animation.label = '► Play'\n",
    "            curdoc().remove_periodic_callback(callback_id)\n",
    "\n",
    "        \n",
    "    \"\"\" ----------- Set up widgets and events -----------  \"\"\"\n",
    "    \n",
    "    # Button to choose the feature to display in the map\n",
    "    label_features = [\"pm10\", \"pm25\", \"no2\", \"o3\", \"so2\"] # labels of possible features\n",
    "    button_choice = RadioButtonGroup(labels=label_features, active=0)\n",
    "    button_choice.on_change('active', update_map)\n",
    "    # As the RadioButton isn't implemented with a title option, we add a PreText object before it\n",
    "    title_button_choice = PreText(text= \"Choose the concentration of the pollutants to visualize\", \n",
    "                                  style={'font-size':'12pt', \n",
    "                     'color': 'black', \n",
    "                     'font-family': 'sans-serif'})\n",
    "    \n",
    "    # Description for Slider\n",
    "    slider_explanation = PreText(text= \"Slider per Week\", style={'font-size':'9pt', \n",
    "                     'color': 'black', \n",
    "                     'font-family': 'sans-serif'})\n",
    "    \n",
    "    # Slider for the date\n",
    "    date_slider = Slider(title=\"Date Corona: \" + air_dates_str1[0] + \"    - Date 2019: \" + air_dates_str2[0], \n",
    "                         start = 0, end = len(air_dates_str1)-1, value = 0, step=1, show_value=False)\n",
    "    date_slider.on_change('value', slider_update)\n",
    "\n",
    "    # Button to animate the slider\n",
    "    button_animation = Button(label='► Play', width=60)\n",
    "    button_animation.on_click(animate)\n",
    "    \n",
    "    \n",
    "    title_containment = PreText(text= \"Lockdown begins in France on March 16\", style={'font-size':'12pt', \n",
    "                     'color': 'black', \n",
    "                     'font-family': 'sans-serif'})\n",
    "\n",
    "    # Set up layouts and add to document\n",
    "    inputs = widgetbox(title_button_choice, button_choice, slider_explanation, date_slider, button_animation, title_containment)\n",
    "    layout_doc = layout([figure1, figure2], inputs, sizing_mode=\"stretch_both\")  \n",
    "    \n",
    "    \n",
    "    \n",
    "    \"\"\" Run the interaction in the bokeh Map in the Jupyter Notebook \"\"\"\n",
    "    \n",
    "    # All server related code is added in the modify_doc function\n",
    "    def modify_doc(doc): # To run the code in a bokeh server, simply replace calls to modify_doc to calls to curdoc() \n",
    "        doc.add_root(layout_doc)\n",
    "        doc.title = title\n",
    "\n",
    "\n",
    "    handler = FunctionHandler(modify_doc)\n",
    "    app = Application(handler)\n",
    "    show(app, notebook_url = notebook_url)\n",
    "    \n",
    "    \n",
    "    \"\"\"\n",
    "    # If you want to display this code on a web browser with a bokeh server, replace the previous section with this one\n",
    "    # and remove the output_notebook() at the beginning of the code\n",
    "    \n",
    "    curdoc().add_root(layout_doc)\n",
    "    curdoc().title = title\n",
    "    \"\"\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='data_analysis'></a> \n",
    "### **Description of the data analysis and what we've learned about the dataset.**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <span style=\"color:#008000\"> III.1 Hospital data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.bokehjs_exec.v0+json": "",
      "text/html": [
       "\n",
       "<script src=\"http://localhost:57656/autoload.js?bokeh-autoload-element=1053&bokeh-absolute-url=http://localhost:57656&resources=none\" id=\"1053\"></script>"
      ]
     },
     "metadata": {
      "application/vnd.bokehjs_exec.v0+json": {
       "server_id": "82cf1b3cb01c430a983756e38a9ecea7"
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_all_bokeh_map_hospital(\"Hospitalized since the beginning of Covid-19\", \n",
    "                            hospital_data_by_feature ,\n",
    "                           hospital_data_min_max,\n",
    "                           \"Cases\",\n",
    "                           TypeColorMapper=LogColorMapper)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> Regarding the **synthesis display mode**:\n",
    "* Not all departments are affected in the same way. Although the departments around Paris have the smallest area, they are by far the most affected. The southern regions are generally less affected. This can be explained by their lower population density.\n",
    "* When switching between the features **Hospitalized** and **Intensive Care**, the gradient remains roughly the same, so this features seems proportional in each department. In the least affected departments, the gradient is darker for the number of cases in intensive care: the hospitals of these departments probably had to have more space and be able to accommodate more people in intensive care.\n",
    "* The observation is not the same between the features **Intensive Care** and **Death**. The map is much darker for the first one which is encouraging. Hover the cursor on the most affected departments (in the North / Center of France) and compare the number of cases in intensive care and the number of cases deceased lets get an idea of the hospital response. For instance in the \"Moselle\", there have been 6963 cases in intensive care against 718 deceased since the beginning whereas in the \"North\", there have been 9836 cases in intensive care against 510 eceased. This shows that hospital capacities differ depending of the department. They are less adapted in the most affected departments. It would have been interesting to zoom in even more and see the distribution within the cases in the hospitals of a department. For instance, we know from friends working in different hospitals from the \"Nord\" department that some hospitals are completely overloaded, while others have unused intensive care rooms. This could have allowed us to highlight logistical problems in the management of the virus.\n",
    "\n",
    "> Regarding the **By Day** and **Cumulated** display modes:\n",
    "* Hospitalizations cases mainly start increasing in border regions and in Paris (where there are the main airports). So it seems to come first through exchanges with foreign countries.\n",
    "* It is interesting that it takes about ten days for the hospital and intensive care gradients to stick together. This shows that hospitals have been slow to prepare.\n",
    "* Even during the lockdown (starting March 16), the virus continued to spread. The lockdown is not sufficient to totaly limite its spread.\n",
    "* The number of cases decreased slightly and death toll increased in the days before the lockdown was lifted (May 11). The lifting of the lockdown may therefore be premature regarding only hospital data."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Another interesting comparison can be made with the deconfinement map provided by the government on May 11th. \n",
    "On this map are represented in red the departments the most at risk, which have known a deconfinement with more strict rules:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![Image](https://www.francetvinfo.fr/image/75rzgbycs-735f/1200/450/21476471.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> This choice does not seem to be only correlated with the number of cases: certain regions of the South have a greater gradient of hospitalized than some of the regions chosen (for instance: \"Bouches-du-Rhônes\"). The deconfinement rule is to prohibit travels of more than 100 km around either to avoid a new spread of the virus. The government may have thought that this rule would not be well respected, and therefore chose to delimit an area in the broad sense, instead of continuing to think in terms of department. The chosen area hosts some of the departments with the largest numbers of cases, as it can bee seen on our Maps."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's now have a look at general curves in France:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.bokehjs_exec.v0+json": "",
      "text/html": [
       "\n",
       "<script src=\"http://localhost:57658/autoload.js?bokeh-autoload-element=1136&bokeh-absolute-url=http://localhost:57658&resources=none\" id=\"1136\"></script>"
      ]
     },
     "metadata": {
      "application/vnd.bokehjs_exec.v0+json": {
       "server_id": "5ad2d64cf5a84ebca84d3c0440946f90"
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_bokeh_multi_lines([df_hospital_France],\"Evolution of the situation in France\", \n",
    "                       dates_str=hospital_dates_str, \n",
    "                       label_modes=[])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* For the line plot, we decided to show the cumulative deaths, because otherwise the data would be too low to view it at the same time as the hospitalization and intensive care data.\n",
    "* This curve makes it possible to visualize the decrease in the number of people hospitalized and in intensive care since the beginning of the lockdown. The peak has been reached around mid April, and the number of death seems to be stabilized starting in the beginning of May."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <span style=\"color:#008000\"> III.2 Energy data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Above all, keep in mind that when we talk about energy, we are talking about electrical energy here."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.bokehjs_exec.v0+json": "",
      "text/html": [
       "\n",
       "<script src=\"http://localhost:57660/autoload.js?bokeh-autoload-element=1207&bokeh-absolute-url=http://localhost:57660&resources=none\" id=\"1207\"></script>"
      ]
     },
     "metadata": {
      "application/vnd.bokehjs_exec.v0+json": {
       "server_id": "6c9e71f1ce5948dc9596d9b299c84362"
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_bokeh_multi_lines([df_consumption, df_thermic, df_nuclear],\n",
    "                       \"French energy consumption and production\", \n",
    "                       dates_str=energy_dates_str, \n",
    "                       label_modes=['Global consumption','Thermic production','Nuclear production']\n",
    "                      )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> * The global consumption is lower than last year, which should be taken into account in the analysis.\n",
    "* Overall consumption and all types of production fell sharply during the April lockdown. The decline occurred gradually in March, and the lockdown does not seem to have been respected immediately. We see that overall consumption starts to rise again in May because the lockdown is lifted.\n",
    "* Nuclear energy provides more than 70% of electricity in France. Its production has decreased little compared to that of thermal energy. The latter is used as a back-up source and was not required during the lockdown ([french source](https://www.edf.fr/groupe-edf/espaces-dedies/l-energie-de-a-a-z/tout-sur-l-energie/produire-de-l-electricite/le-thermique-a-flamme-en-chiffres))."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.bokehjs_exec.v0+json": "",
      "text/html": [
       "\n",
       "<script src=\"http://localhost:57662/autoload.js?bokeh-autoload-element=1258&bokeh-absolute-url=http://localhost:57662&resources=none\" id=\"1258\"></script>"
      ]
     },
     "metadata": {
      "application/vnd.bokehjs_exec.v0+json": {
       "server_id": "9cbb1095959346dca83332f37fceb391"
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_all_bokeh_map_energy(\"General Energy Consumption per Day\", \n",
    "                        energy_data_by_feature,\n",
    "                           energy_data_min_max_without_0,\n",
    "                           \"Energy consumption in the region\",\n",
    "                           TypeColorMapper=LogColorMapper,\n",
    "                         palette = palette_inferno)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* From the beginning of April, several regions stopped producing thermal energy, because the other productions (nuclear, green energies) were sufficient. \n",
    "* Regarding electricity consumption, it drops drastically at the start of the lockdown. This shows that companies were not prepared for telework, and that it took time to set up. There was thus a much less active period of two weeks starting at the announce of the lockdown.\n",
    "* There is a very strange jump on the first of April, where consumption is drastically increased compared to the end of March. Regarding nuclear power, production is increasing. Even if nuclear production was slowed down at the beginning due to organisational difficulties and a drop in energy consumption, this increase of production seems too huge. Perhaps the choice was to focus on nuclear power and reduce other energy production (thermal power decreases drastically at the same time) because it requires fewer human resources and therefore limits the spread of the virus among staff. It is also possible that the electrical readings were limited to the end of March due to staff limitations."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <span style=\"color:#008000\"> III.3 Airline traffic data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.bokehjs_exec.v0+json": "",
      "text/html": [
       "\n",
       "<script src=\"http://localhost:57664/autoload.js?bokeh-autoload-element=1315&bokeh-absolute-url=http://localhost:57664&resources=none\" id=\"1315\"></script>"
      ]
     },
     "metadata": {
      "application/vnd.bokehjs_exec.v0+json": {
       "server_id": "b192b7f38e034a40bd330ee6a934cacd"
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\"\"\" -------------------- Plot with a linear mapping --------------------  \"\"\"\n",
    "\n",
    "plot_bokeh_map_flights('Flights per airport per day in France', \n",
    "                           flight_by_feature, \n",
    "                            size_airport_city, \n",
    "                            plot_width=450, plot_height=500, \n",
    "                            tools=TOOLS)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.bokehjs_exec.v0+json": "",
      "text/html": [
       "\n",
       "<script src=\"http://localhost:57666/autoload.js?bokeh-autoload-element=1398&bokeh-absolute-url=http://localhost:57666&resources=none\" id=\"1398\"></script>"
      ]
     },
     "metadata": {
      "application/vnd.bokehjs_exec.v0+json": {
       "server_id": "ad9fba4e4a8a4a59bb0e8253c02b13e7"
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_bokeh_multi_lines([df_air_traffic],\"French airline traffic\", \n",
    "                       dates_str=air_dates_str, \n",
    "                       label_modes=[]\n",
    "                      )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> * The number of trips is already decreasing before the announcement of the lockdown (March 16).\n",
    "* Most airports closed when the lockdown was announced. Only Paris still operates a decent amount of flights during the lockdown.\n",
    "* There are many international flights (around 500 per day) at the two Paris airports at the start of the lockdown. This is in line with the encouragement of the various embassies for people to return to their home country.\n",
    "* Surprisingly, there are still some intranational flights during the lockdown."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <span style=\"color:#008000\"> III.4 Air quality data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.bokehjs_exec.v0+json": "",
      "text/html": [
       "\n",
       "<script src=\"http://localhost:57668/autoload.js?bokeh-autoload-element=1469&bokeh-absolute-url=http://localhost:57668&resources=none\" id=\"1469\"></script>"
      ]
     },
     "metadata": {
      "application/vnd.bokehjs_exec.v0+json": {
       "server_id": "be552354f5e7419f907fc2b15b5b4d55"
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_bokeh_multi_lines([df_no2,df_so2,df_o3,df_pm10,df_pm25],\n",
    "                       \"Emission of pollutants in France\", \n",
    "                       dates_str=quality_dates_str, \n",
    "                       label_modes=['no2','so2','o3','pm10','pm25']\n",
    "                      )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We displayed a lot of different indicators. Here is a short description we found on Wikipedia for each of them:\n",
    "* **pm10** and **pm25** are particulate matters, mainly produced through industrial processes (bulk material handling, combustion and minerals processing), natural sources (bushfires, dust storms, pollens and sea spray), vehicles \n",
    "* **so2** is a gas, coming 99% from human sources: industrial activity (generation of electricity, process of some minerals, fossil fuels burning) and motor vehicle emissions\n",
    "* **o3** is formed due to industrial activity and various constituents released into the air, mainly from road traffic. Its rate also changes naturally according to the seasons and increases in spring / summer\n",
    "* **no2** forms from emissions from cars, trucks and buses, power plants, and off-road equipment"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> It is quite difficult to draw conclusions from the single curves for non specialists, since most indicators also depend on the seasons. However, we have seen a very marked decrease in the level of so2 (the only indicator that is almost completely linked to human activity) since the start of the spread of the virus, which has halved nationwide."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Further analysis can be carried out by comparing maps from this year and last year:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.bokehjs_exec.v0+json": "",
      "text/html": [
       "\n",
       "<script src=\"http://localhost:57670/autoload.js?bokeh-autoload-element=1576&bokeh-absolute-url=http://localhost:57670&resources=none\" id=\"1576\"></script>"
      ]
     },
     "metadata": {
      "application/vnd.bokehjs_exec.v0+json": {
       "server_id": "859f0db419a943699993d52a6ceaf047"
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_bokeh_map_air_polution_comparison('Comparison air pollution before and during Covid-19',\n",
    "                        'Air Pollution in France - Beginning of 2020', \n",
    "                            'Air Pollution in France at the same period in 2019',\n",
    "                           air_pollution_by_pollutant_corona,\n",
    "                            air_pollution_by_pollutant_2019,\n",
    "                            air_data_min_max,\n",
    "                            air_dates_str_corona,\n",
    "                            air_dates_str_2019)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> * We note that the emissions before lockdown are generally higher this year than last year.\n",
    "* The largest emissions of pollutants are not all located in Paris as one would expect from the density of pollution and the number of vehicles in circulation, but rather in the cities of North and North West. These are indeed cities close to industrial zones.\n",
    "* In some cities, the pollutants measurements were not made during all days of the lockdown (example: Lille). This result in a value of 0 for the concentration onc some pollutants.\n",
    "* Generally, we see a decrease in pollution gradients for all species compared to last year during the lockdown period. This indicates a decrease in road traffic as well as a slowdown in the industry."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### **Description of the machine learning: Comparison with Italy and Spain through a polynomial regression**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "How is the situation in France compared to its neighbours?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "number_min_days = np.min([len(dict_death['Italy']),len(dict_death['Spain'])])\n",
    "number_min_days\n",
    "\n",
    "# Fill X and y features \n",
    "X=[]\n",
    "y=[]\n",
    "for i in range(number_min_days):\n",
    "    X.append(i)\n",
    "    X.append(i)\n",
    "    y.append(dict_death['Italy'][i])\n",
    "    y.append(dict_death['Spain'][i])\n",
    "X = np.array(X).reshape(-1,1)\n",
    "y = np.array(y).reshape(-1,1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Regression with 1 degree, RMSE: 1605.405559041005\n",
      "Regression with 5 degrees, RMSE: 820.2051102566404\n"
     ]
    }
   ],
   "source": [
    "# sklearn Polynomial regression\n",
    "poly = PolynomialFeatures(degree=5, include_bias=False)\n",
    "X_polynomial = poly.fit_transform(X) # We don't actually fit anything here but fitting is required when a data transformer used for the first time\n",
    "\n",
    "model = LinearRegression()\n",
    "\n",
    "model.fit(X, y) \n",
    "y_pred = model.predict(X)\n",
    "print(\"Regression with 1 degree, RMSE:\", np.sqrt(mean_squared_error(y, y_pred)))\n",
    "\n",
    "model.fit(X_polynomial, y) \n",
    "y_pred_polynomial = model.predict(X_polynomial)\n",
    "print(\"Regression with 5 degrees, RMSE:\", np.sqrt(mean_squared_error(y, y_pred_polynomial)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plot\n",
    "plt.figure(figsize=(12, 5))\n",
    "plt.scatter(X, y, s=20, label='samples')\n",
    "plt.plot(X, y_pred, c='g', label=\"Regression with 1 degree\")\n",
    "plt.plot(X, y_pred_polynomial, c='r', label=\"Regression with 5 degrees\")\n",
    "plt.xlabel(\"days since 400th death\")\n",
    "plt.ylabel(\"deaths\")\n",
    "plt.legend(loc=\"best\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> It's better to do the regression with 5 degrees. <br>\n",
    "Now, let's apply this regression with 5 degrees to the 9 most affected countries excluding France."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**ARIMA forecasting:**\n",
    "\n",
    "Let's do dome forecasting for the death toll.<br>\n",
    "\n",
    "The Autoregressive Integrated Moving Average (ARIMA) method models the next step in the sequence as a linear function of the differenced observations and residual errors at prior time steps. <br>\n",
    "\n",
    "It combines both Autoregression (AR) and Moving Average (MA) models as well as a differencing pre-processing step of the sequence to make the sequence stationary, called integration (I).\n",
    "\n",
    "To determine parameters value of the ARIMA model, we plot ACF and PACF."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_acf(df_hospital_France['Cumulated_deceased'])\n",
    "plot_pacf(df_hospital_France['Cumulated_deceased']);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the autocorrelation plot, the first two sticks above the confidence interval. So 2 will be the moving average value. <br>\n",
    "Regarding auto regression, the partial autocorrelation plot indicates a value of 1 is the maximum."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Fit ARIMA model\n",
    "model = ARIMA(df_hospital_France['Cumulated deceased'], order=(1,1,2))\n",
    "arima = model.fit()\n",
    "\n",
    "y_pred_arima=arima.predict(max(df_hospital_France.index)+datetime.timedelta(days=1), \n",
    "                     max(df_hospital_France.index)+datetime.timedelta(days=10), #10 days forecasting\n",
    "                     typ='levels'\n",
    "                    )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Plot\n",
    "plt.figure(figsize=(12, 5))\n",
    "plt.scatter(df_hospital_France['integer'],df_hospital_France['Cumulated deceased'], label='France cumulated deceased')\n",
    "plt.scatter(range(max(df_hospital_France['integer'])+1,\n",
    "                  max(df_hospital_France['integer'])+1+len(y_pred_arima)),y_pred_arima,\n",
    "           label='France deaths forecasted')\n",
    "plt.plot(X, y_pred_polynomial, c='r', label=\"Spain and Italy polynomial regression\")\n",
    "plt.legend(loc='best')\n",
    "plt.xlabel(\"Days since 400th death\")\n",
    "plt.ylabel(\"Cumulated deceased\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> France has suffered from fewer deaths than Italy and Spain. Perharps the lockdown decided earlier, stricter health measures and a bigger health system capacity has helped the country to stem the tide of the outbreak. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# <span style=\"color:#990000\"> IV - Genre <a class=\"anchor\" id=\"fourth-part\">"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### **Which genre of data story did you use?**\n",
    "> We chose an interactive slideshow, with a reader-driven approach. The maps we have done are rich in information, and each user will probably draw different conclusions in the light of his/her interests and knowledges of the situation."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### **Which tools did you use from each of the 3 categories of Visual Narrative (Figure 7 in Segal and Heer). Why?**\n",
    "> From the Visual Narrative categories, we used:\n",
    "* A consistent visual platform, to give a constant frame for the reader to navigate into\n",
    "* Feature distinction, for the highlighting process\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### **Which tools did you use from each of the 3 categories of Narrative Structure (Figure 7 in Segal and Heer). Why?**\n",
    "> From the Narrative Structure categories, we used:\n",
    "* A user directed path, as we wanted to give the user freedom upon the order in which to view the topics\n",
    "* Hover highlighting, navigation buttons and filtering for the interactivity, as well as explicit instructions.\n",
    "* Captions and annotations, for the context"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# <span style=\"color:#990000\"> V - Vizualizations <a class=\"anchor\" id=\"fifth-part\">"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###  **Explain the visualizations you've chosen.**\n",
    "> We chose to display **interactive maps and line plots**. \n",
    "\n",
    "> The maps appeared as the most suited visualizations we could use to display the geographical evolution of our data: cities in the air quality dataset, airports in the flights dataset, and region or department data for the two others. With maps, we could for instance highlight how differently the east of France and the Paris region are affected differently from the rest of the country. For some of them, the user can also select a specific date on a slider, to be able to compare at their own pace the evolution of the dataset on both a geographical basis and a temporal basis.\n",
    "\n",
    "> The line plots are used in our website to show the temporal evolution of the data: they highlight especially how much of a difference the beginning of the lockdown made in some of our datasets, and also give us the opportunity to visually compare the 2020 data with the 2019 data, to give the reader an idea of how the situation could be without the pandemic occurring. \n",
    "  \n",
    "> Both our maps and line plots are interactive, meaning that they provide additional information to the reader upon hovering. Some also provide a \"Play\" function, to allow for an automatic geographical and temporal visualization.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###  **Why are they right for the story you want to tell?**\n",
    "> Interactive maps and line plots, as explained in the previous section, give to the reader a precise idea of both the geographical and temporal evolution of the data, which is specifically important to tell our story: we want the user to be able to discover the different impact of the pandemic depending on the region of France, but also the evolution before/after the lockdown."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# <span style=\"color:#990000\"> VI - Discussion <a class=\"anchor\" id=\"sixth-part\">"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###  What went well? What is still missing? What could be improved?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> There are some things that could have been added to the graphs to make them even more meaningful that we thought about but that we didn't have time to put in place:\n",
    "* Regarding **hospital data**, it could have been interesting to put the data in relation to other data such as the population density or the number of inhabitants for each region. \n",
    "* For the **energy data**: another feature to implement could be green energies. It would also have been interesting to add the consumption shares (industry, household, ...). We would expect to see more energy consumption in households and less in factories. \n",
    "* For the **flights data**: with more time, we would have liked to compare with last year's flights as for air quality, and to allow the user to observe departure and arrival flights differently. There are probably more return flights abroad to France than flights in the other direction. It would also have been interesting to visualize the data of French expatriates in the main countries and compare it with the number of flights which still ensure the transition.\n",
    "* Concerning **air quality data**, it would have probably been more useful to show the curves to compare last year and this year instead of the min/median/max.\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "> Concerning the tools we chose to use:\n",
    "* Navigation could have been improved. For example, it could have been interesting to offer the user to display the maps he wants at the same time. It would have been much easier to implement it in javascript if we had worked with a library like d3. Implementing this option under Bokeh is possible but would have involved even more code redundancy.\n",
    "* As we wanted to use interactive Bokeh charts, we needed to use a Bokeh server. The best way to display Bokeh server plots on a website we found is to use the Django framework, with embedded plots. But hosting a Django website with such plots did not work with a AWS host, nor with a Heroku host. We ended up having to host the website locally, which provided us with additional flexibility. An improvement would be to be able to host this website on an online host."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# <span style=\"color:#990000\"> VII - Contributions <a class=\"anchor\" id=\"seventh-part\">"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<table style=\"width:100%; text-align:left;\">\n",
    "  <tr>\n",
    "    <th style=\"text-align:left\">Gaétan Moisson-Franckhauser</th>\n",
    "    <th style=\"text-align:left\">Brian Tang</th>\n",
    "    <th style=\"text-align:left\">Tanguy Navez</th>\n",
    "  </tr>\n",
    "  <tr>\n",
    "    <td style=\"text-align:left\">Website, Explainer notebook</td>\n",
    "    <td style=\"text-align:left\">Bokeh line plots, Explainer notebook</td>\n",
    "    <td style=\"text-align:left\">Bokeh Maps, Explainer Notebook</td>\n",
    "  </tr>\n",
    "</table> "
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}