Machine-Learning-Predictive-Model.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    " # Data Analysis and Predictive Model Training - OuiBikes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The following notebook is a review of the data quality used for the training of our predictor model, as well as the steps taken to train and improve the model itself. Within our database, the important data to be processed and used is stored in two seperate tables:\n",
    "\n",
    "* dublinbikes.station_status - updated every six minutes and contains all the dynamic data for each bike station\n",
    "* dublinbikes.current_weather - updated every 30 minutes and contains all the data relating to weather in Dublin\n",
    "\n",
    "For the purpose of data processing and model training, it was agreed as a group to write a single SQL query that would load both these tables as a single dataframe for examination and manipulation. The data taken from both tables would be averaged into an hourly sample. This would simplify the process of data processing, especially with the weather and bike information recorded at different time intervals. \n",
    "\n",
    "Within this notebook, all API keys/passwords will be redacted, as the intention is to host this notebook and its code publicly on GitHub. Should a working demonstration of the data processing and model be required, please get in contact with one of the team members on the project."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### SQL Combined Data Query"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "# Library Imports.\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "\n",
    "# Allows plots to appear directly in the notebook.\n",
    "%matplotlib inline\n",
    "\n",
    "from patsy import dmatrices\n",
    "from sqlalchemy import create_engine\n",
    "from sklearn import metrics\n",
    "from sklearn.model_selection import TimeSeriesSplit\n",
    "from sklearn.metrics import r2_score\n",
    "from sklearn.metrics import make_scorer\n",
    "from sklearn.tree import DecisionTreeRegressor\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.model_selection import ShuffleSplit\n",
    "import datetime\n",
    "import pickle\n",
    "import math\n",
    "from IPython.display import Image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "URI = 'REDACTED'\n",
    "DB = ***REMOVED***\n",
    "PORT = \"REDACTED\"\n",
    "USER = \"REDACTED\"\n",
    "PASSWORD = \"REDACTED\"\n",
    "\n",
    "engine = create_engine(\"mysql+pymysql://{}:{}@{}:{}/{}\".format(USER, PASSWORD, URI, PORT, DB), echo=True)\n",
    "\n",
    "def sql_query_model():\n",
    "    ''' SQL Query to retrieve bike station data and weather data from two seperate tables and combine them\n",
    "    into a single dataframe for data processing. The time sample for both data will be set to hourly using\n",
    "    the averages of the data.\n",
    "    '''\n",
    "    \n",
    "    conn = engine.connect()\n",
    "    \n",
    "    sql1 = \"\"\"SELECT number, date_format(CAST(last_update AS DATETIME), '%%Y-%%m-%%d %%H' ) as myDate, AVG(available_bike_stands) as available_bike_stands, AVG(available_bikes) as available_bikes\n",
    "            FROM station_status\n",
    "            GROUP BY number, myDate\n",
    "            ORDER BY myDate ASC;\"\"\"\n",
    "        \n",
    "    df1 = pd.read_sql_query(sql1, conn)\n",
    "    df1['myDate'] = pd.to_datetime(df1.myDate, unit='ns')\n",
    "    df1['myDate'] = df1.myDate.map(lambda x: x.replace(minute=0, second=0))\n",
    "    \n",
    "    sql2 = \"\"\"SELECT date_format(CAST(last_update AS DATETIME), '%%Y-%%m-%%d %%H' ) as myDate, main_weather, (AVG(`main.temp`) -273.15) as main_temp, AVG(rainfall_mm) as rainfall_mm, AVG(`wind.speed`) as wind_speed\n",
    "            FROM current_weather\n",
    "            GROUP BY myDate\n",
    "            ORDER BY myDate ASC;\"\"\"\n",
    "    \n",
    "    df2 = pd.read_sql_query(sql2, conn)\n",
    "    df2['myDate'] = pd.to_datetime(df2.myDate, unit='ns')\n",
    "    df2['myDate'] = df2.myDate.map(lambda x: x.replace(minute=0, second=0))\n",
    "      \n",
    "    df3 = pd.merge(df1, df2, on='myDate')\n",
    "    \n",
    "    for index, row in df3.iterrows():\n",
    "        if df3.loc[index,'myDate'].hour == 0:\n",
    "            df3.loc[index,'myDate'] = df3.loc[index,'myDate'] - datetime.timedelta(minutes=(df3.loc[index,'myDate'].minute + 1))\n",
    "            df3.loc[index,'hour'] = 24\n",
    "        else:\n",
    "            df3.loc[index,'hour'] = (df3.loc[index,'myDate'].hour)\n",
    "            \n",
    "    df3['weekday'] = df3['myDate'].dt.day_name()\n",
    "    df3.set_index('myDate', inplace=True)\n",
    "    \n",
    "    return df3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2019-04-19 07:27:41,356 INFO sqlalchemy.engine.base.Engine SHOW VARIABLES LIKE 'sql_mode'\n",
      "2019-04-19 07:27:41,358 INFO sqlalchemy.engine.base.Engine {}\n",
      "2019-04-19 07:27:41,470 INFO sqlalchemy.engine.base.Engine SHOW VARIABLES LIKE 'lower_case_table_names'\n",
      "2019-04-19 07:27:41,471 INFO sqlalchemy.engine.base.Engine {}\n",
      "2019-04-19 07:27:41,693 INFO sqlalchemy.engine.base.Engine SELECT DATABASE()\n",
      "2019-04-19 07:27:41,695 INFO sqlalchemy.engine.base.Engine {}\n",
      "2019-04-19 07:27:41,917 INFO sqlalchemy.engine.base.Engine show collation where `Charset` = 'utf8mb4' and `Collation` = 'utf8mb4_bin'\n",
      "2019-04-19 07:27:41,918 INFO sqlalchemy.engine.base.Engine {}\n",
      "2019-04-19 07:27:42,031 INFO sqlalchemy.engine.base.Engine SELECT CAST('test plain returns' AS CHAR(60)) AS anon_1\n",
      "2019-04-19 07:27:42,033 INFO sqlalchemy.engine.base.Engine {}\n",
      "2019-04-19 07:27:42,146 INFO sqlalchemy.engine.base.Engine SELECT CAST('test unicode returns' AS CHAR(60)) AS anon_1\n",
      "2019-04-19 07:27:42,147 INFO sqlalchemy.engine.base.Engine {}\n",
      "2019-04-19 07:27:42,259 INFO sqlalchemy.engine.base.Engine SELECT CAST('test collated returns' AS CHAR CHARACTER SET utf8mb4) COLLATE utf8mb4_bin AS anon_1\n",
      "2019-04-19 07:27:42,260 INFO sqlalchemy.engine.base.Engine {}\n",
      "2019-04-19 07:27:42,483 INFO sqlalchemy.engine.base.Engine SELECT number, date_format(CAST(last_update AS DATETIME), '%%Y-%%m-%%d %%H' ) as myDate, AVG(available_bike_stands) as available_bike_stands, AVG(available_bikes) as available_bikes\n",
      "            FROM station_status\n",
      "            GROUP BY number, myDate\n",
      "            ORDER BY myDate ASC;\n",
      "2019-04-19 07:27:42,485 INFO sqlalchemy.engine.base.Engine {}\n",
      "2019-04-19 07:27:49,651 INFO sqlalchemy.engine.base.Engine SELECT date_format(CAST(last_update AS DATETIME), '%%Y-%%m-%%d %%H' ) as myDate, main_weather, (AVG(`main.temp`) -273.15) as main_temp, AVG(rainfall_mm) as rainfall_mm, AVG(`wind.speed`) as wind_speed\n",
      "            FROM current_weather\n",
      "            GROUP BY myDate\n",
      "            ORDER BY myDate ASC;\n",
      "2019-04-19 07:27:49,652 INFO sqlalchemy.engine.base.Engine {}\n"
     ]
    }
   ],
   "source": [
    "df_db = sql_query_model()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Quality Report"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A first step for data analysis and processing is to check the quality of the data to be used. This will indentify issues such as NaN entries, outliers and the type of data to be processed (categorical or continuous). "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### - Check how many rows and columns the loaded data has"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(127012, 9)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Print the number of rows and columns\n",
    "df_db.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A total of 127,012 rows of data are present, and there are nine columns/features contained within this dataframe."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### - Print the first and last five rows"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "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>number</th>\n",
       "      <th>available_bike_stands</th>\n",
       "      <th>available_bikes</th>\n",
       "      <th>main_weather</th>\n",
       "      <th>main_temp</th>\n",
       "      <th>rainfall_mm</th>\n",
       "      <th>wind_speed</th>\n",
       "      <th>hour</th>\n",
       "      <th>weekday</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>myDate</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2019-02-21 23:00:00</th>\n",
       "      <td>95</td>\n",
       "      <td>1.2</td>\n",
       "      <td>38.8</td>\n",
       "      <td>Clouds</td>\n",
       "      <td>11.85</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>23.0</td>\n",
       "      <td>Thursday</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2019-02-21 23:00:00</th>\n",
       "      <td>96</td>\n",
       "      <td>17.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>Clouds</td>\n",
       "      <td>11.85</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>23.0</td>\n",
       "      <td>Thursday</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2019-02-21 23:00:00</th>\n",
       "      <td>91</td>\n",
       "      <td>4.5</td>\n",
       "      <td>25.5</td>\n",
       "      <td>Clouds</td>\n",
       "      <td>11.85</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>23.0</td>\n",
       "      <td>Thursday</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2019-02-21 23:00:00</th>\n",
       "      <td>92</td>\n",
       "      <td>9.8</td>\n",
       "      <td>30.2</td>\n",
       "      <td>Clouds</td>\n",
       "      <td>11.85</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>23.0</td>\n",
       "      <td>Thursday</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2019-02-21 23:00:00</th>\n",
       "      <td>87</td>\n",
       "      <td>12.0</td>\n",
       "      <td>26.0</td>\n",
       "      <td>Clouds</td>\n",
       "      <td>11.85</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>23.0</td>\n",
       "      <td>Thursday</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                     number  available_bike_stands  available_bikes  \\\n",
       "myDate                                                                \n",
       "2019-02-21 23:00:00      95                    1.2             38.8   \n",
       "2019-02-21 23:00:00      96                   17.0             13.0   \n",
       "2019-02-21 23:00:00      91                    4.5             25.5   \n",
       "2019-02-21 23:00:00      92                    9.8             30.2   \n",
       "2019-02-21 23:00:00      87                   12.0             26.0   \n",
       "\n",
       "                    main_weather  main_temp  rainfall_mm  wind_speed  hour  \\\n",
       "myDate                                                                       \n",
       "2019-02-21 23:00:00       Clouds      11.85          0.0         3.0  23.0   \n",
       "2019-02-21 23:00:00       Clouds      11.85          0.0         3.0  23.0   \n",
       "2019-02-21 23:00:00       Clouds      11.85          0.0         3.0  23.0   \n",
       "2019-02-21 23:00:00       Clouds      11.85          0.0         3.0  23.0   \n",
       "2019-02-21 23:00:00       Clouds      11.85          0.0         3.0  23.0   \n",
       "\n",
       "                      weekday  \n",
       "myDate                         \n",
       "2019-02-21 23:00:00  Thursday  \n",
       "2019-02-21 23:00:00  Thursday  \n",
       "2019-02-21 23:00:00  Thursday  \n",
       "2019-02-21 23:00:00  Thursday  \n",
       "2019-02-21 23:00:00  Thursday  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Print the first five rows of data\n",
    "df_db.head(5)"
   ]
  },
  {
   "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>number</th>\n",
       "      <th>available_bike_stands</th>\n",
       "      <th>available_bikes</th>\n",
       "      <th>main_weather</th>\n",
       "      <th>main_temp</th>\n",
       "      <th>rainfall_mm</th>\n",
       "      <th>wind_speed</th>\n",
       "      <th>hour</th>\n",
       "      <th>weekday</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>myDate</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2019-04-19 06:00:00</th>\n",
       "      <td>95</td>\n",
       "      <td>8.8</td>\n",
       "      <td>31.2</td>\n",
       "      <td>Mist</td>\n",
       "      <td>6.08</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.6</td>\n",
       "      <td>6.0</td>\n",
       "      <td>Friday</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2019-04-19 06:00:00</th>\n",
       "      <td>93</td>\n",
       "      <td>6.0</td>\n",
       "      <td>34.0</td>\n",
       "      <td>Mist</td>\n",
       "      <td>6.08</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.6</td>\n",
       "      <td>6.0</td>\n",
       "      <td>Friday</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2019-04-19 06:00:00</th>\n",
       "      <td>90</td>\n",
       "      <td>33.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>Mist</td>\n",
       "      <td>6.08</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.6</td>\n",
       "      <td>6.0</td>\n",
       "      <td>Friday</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2019-04-19 06:00:00</th>\n",
       "      <td>91</td>\n",
       "      <td>30.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Mist</td>\n",
       "      <td>6.08</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.6</td>\n",
       "      <td>6.0</td>\n",
       "      <td>Friday</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2019-04-19 06:00:00</th>\n",
       "      <td>89</td>\n",
       "      <td>32.6</td>\n",
       "      <td>6.4</td>\n",
       "      <td>Mist</td>\n",
       "      <td>6.08</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.6</td>\n",
       "      <td>6.0</td>\n",
       "      <td>Friday</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                     number  available_bike_stands  available_bikes  \\\n",
       "myDate                                                                \n",
       "2019-04-19 06:00:00      95                    8.8             31.2   \n",
       "2019-04-19 06:00:00      93                    6.0             34.0   \n",
       "2019-04-19 06:00:00      90                   33.0              7.0   \n",
       "2019-04-19 06:00:00      91                   30.0              0.0   \n",
       "2019-04-19 06:00:00      89                   32.6              6.4   \n",
       "\n",
       "                    main_weather  main_temp  rainfall_mm  wind_speed  hour  \\\n",
       "myDate                                                                       \n",
       "2019-04-19 06:00:00         Mist       6.08          0.0         2.6   6.0   \n",
       "2019-04-19 06:00:00         Mist       6.08          0.0         2.6   6.0   \n",
       "2019-04-19 06:00:00         Mist       6.08          0.0         2.6   6.0   \n",
       "2019-04-19 06:00:00         Mist       6.08          0.0         2.6   6.0   \n",
       "2019-04-19 06:00:00         Mist       6.08          0.0         2.6   6.0   \n",
       "\n",
       "                    weekday  \n",
       "myDate                       \n",
       "2019-04-19 06:00:00  Friday  \n",
       "2019-04-19 06:00:00  Friday  \n",
       "2019-04-19 06:00:00  Friday  \n",
       "2019-04-19 06:00:00  Friday  \n",
       "2019-04-19 06:00:00  Friday  "
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Print the last five rows of data\n",
    "df_db.tail(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Converting features to their appropriate data types (which features are more appropriate as continuous features, which are more appropriate as categorical features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "number                     int64\n",
       "available_bike_stands    float64\n",
       "available_bikes          float64\n",
       "main_weather              object\n",
       "main_temp                float64\n",
       "rainfall_mm              float64\n",
       "wind_speed               float64\n",
       "hour                     float64\n",
       "weekday                   object\n",
       "dtype: object"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check each features current data type \n",
    "df_db.dtypes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As it stands currently, all objects in our dataframe should be converted to categories. It is also important to note for the purpose of our modelling that the index is a datetime stamp. This will become important when we perform further sampling operations, as we are working with time series data. For the model training and cross validation, we must respect data entries and their chronological order. No training data should come *after* any of our testing or validation data. This would not make sense, as bike availability rises and falls based on the cycle of the working day. For example, I would expect less bikes available during the morning and evening on a work day, as this is rush hour. I might also expect a slight decline during lunch time, as people look to get places quickly for lunch."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Select columns containing the aforementioned to be converted to categorical data type\n",
    "categorical_columns = df_db[['main_weather', 'weekday']].columns\n",
    "\n",
    "# For each data type in the list, convert it to categorical data\n",
    "for column in categorical_columns:\n",
    "    df_db[column] = df_db[column].astype('category')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "number                      int64\n",
       "available_bike_stands     float64\n",
       "available_bikes           float64\n",
       "main_weather             category\n",
       "main_temp                 float64\n",
       "rainfall_mm               float64\n",
       "wind_speed                float64\n",
       "hour                      float64\n",
       "weekday                  category\n",
       "dtype: object"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check desired data types have been correctly converted\n",
    "df_db.dtypes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### - Drop duplicate rows and columns, if any present"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Duplicate rows: 0\n",
      "Duplicate columns: 0\n"
     ]
    }
   ],
   "source": [
    "# Duplicate row check\n",
    "print('Duplicate rows:', df_db.duplicated()[df_db.duplicated() == True].shape[0])\n",
    "\n",
    "# Duplicate column check\n",
    "print('Duplicate columns:', df_db.columns.size - df_db.columns.unique().size)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "No duplicate rows or columns are present. This means that our data scraping process is robust, and also our merging of data by hour sample using the SQL query presented no issues. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### - Drop constant columns, if any"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Select columns containing continuous data\n",
    "continuous_columns = df_db.select_dtypes(include=['float64']).columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "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>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>available_bike_stands</th>\n",
       "      <td>127012.0</td>\n",
       "      <td>21.080381</td>\n",
       "      <td>11.634433</td>\n",
       "      <td>0.0</td>\n",
       "      <td>12.33330</td>\n",
       "      <td>20.5000</td>\n",
       "      <td>29.9000</td>\n",
       "      <td>41.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>available_bikes</th>\n",
       "      <td>127012.0</td>\n",
       "      <td>10.703650</td>\n",
       "      <td>10.576528</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.50000</td>\n",
       "      <td>7.1000</td>\n",
       "      <td>17.8000</td>\n",
       "      <td>40.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>main_temp</th>\n",
       "      <td>127012.0</td>\n",
       "      <td>7.612343</td>\n",
       "      <td>3.103932</td>\n",
       "      <td>0.1</td>\n",
       "      <td>5.56375</td>\n",
       "      <td>7.5125</td>\n",
       "      <td>9.7275</td>\n",
       "      <td>16.455</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rainfall_mm</th>\n",
       "      <td>127012.0</td>\n",
       "      <td>0.098482</td>\n",
       "      <td>0.351068</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.0100</td>\n",
       "      <td>3.300</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>wind_speed</th>\n",
       "      <td>127012.0</td>\n",
       "      <td>5.423212</td>\n",
       "      <td>2.876910</td>\n",
       "      <td>0.5</td>\n",
       "      <td>3.35000</td>\n",
       "      <td>4.8500</td>\n",
       "      <td>7.2000</td>\n",
       "      <td>15.650</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hour</th>\n",
       "      <td>127012.0</td>\n",
       "      <td>14.500000</td>\n",
       "      <td>5.781097</td>\n",
       "      <td>5.0</td>\n",
       "      <td>9.00000</td>\n",
       "      <td>14.5000</td>\n",
       "      <td>20.0000</td>\n",
       "      <td>24.000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                          count       mean        std  min       25%      50%  \\\n",
       "available_bike_stands  127012.0  21.080381  11.634433  0.0  12.33330  20.5000   \n",
       "available_bikes        127012.0  10.703650  10.576528  0.0   1.50000   7.1000   \n",
       "main_temp              127012.0   7.612343   3.103932  0.1   5.56375   7.5125   \n",
       "rainfall_mm            127012.0   0.098482   0.351068  0.0   0.00000   0.0000   \n",
       "wind_speed             127012.0   5.423212   2.876910  0.5   3.35000   4.8500   \n",
       "hour                   127012.0  14.500000   5.781097  5.0   9.00000  14.5000   \n",
       "\n",
       "                           75%     max  \n",
       "available_bike_stands  29.9000  41.000  \n",
       "available_bikes        17.8000  40.000  \n",
       "main_temp               9.7275  16.455  \n",
       "rainfall_mm             0.0100   3.300  \n",
       "wind_speed              7.2000  15.650  \n",
       "hour                   20.0000  24.000  "
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Print table\n",
    "df_db[continuous_columns].describe().T"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As no features standard deviation = 0, none of the continuous features present in the data are constant. We can also check this is the case for the bike station number, which is of type int64."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Select columns containing continuous data\n",
    "continuous_columns = df_db.select_dtypes(include=['int64']).columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "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>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>number</th>\n",
       "      <td>127012.0</td>\n",
       "      <td>58.840708</td>\n",
       "      <td>32.852323</td>\n",
       "      <td>2.0</td>\n",
       "      <td>31.0</td>\n",
       "      <td>59.0</td>\n",
       "      <td>87.0</td>\n",
       "      <td>115.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           count       mean        std  min   25%   50%   75%    max\n",
       "number  127012.0  58.840708  32.852323  2.0  31.0  59.0  87.0  115.0"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Print table\n",
    "df_db[continuous_columns].describe().T"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This feature also has a standard deviation greater than 0, so it is a non-constant feature."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### - Table with descriptive statistics for all the continuous features"
   ]
  },
  {
   "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>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>available_bike_stands</th>\n",
       "      <td>127012.0</td>\n",
       "      <td>21.080381</td>\n",
       "      <td>11.634433</td>\n",
       "      <td>0.0</td>\n",
       "      <td>12.33330</td>\n",
       "      <td>20.5000</td>\n",
       "      <td>29.9000</td>\n",
       "      <td>41.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>available_bikes</th>\n",
       "      <td>127012.0</td>\n",
       "      <td>10.703650</td>\n",
       "      <td>10.576528</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.50000</td>\n",
       "      <td>7.1000</td>\n",
       "      <td>17.8000</td>\n",
       "      <td>40.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>main_temp</th>\n",
       "      <td>127012.0</td>\n",
       "      <td>7.612343</td>\n",
       "      <td>3.103932</td>\n",
       "      <td>0.1</td>\n",
       "      <td>5.56375</td>\n",
       "      <td>7.5125</td>\n",
       "      <td>9.7275</td>\n",
       "      <td>16.455</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rainfall_mm</th>\n",
       "      <td>127012.0</td>\n",
       "      <td>0.098482</td>\n",
       "      <td>0.351068</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.0100</td>\n",
       "      <td>3.300</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>wind_speed</th>\n",
       "      <td>127012.0</td>\n",
       "      <td>5.423212</td>\n",
       "      <td>2.876910</td>\n",
       "      <td>0.5</td>\n",
       "      <td>3.35000</td>\n",
       "      <td>4.8500</td>\n",
       "      <td>7.2000</td>\n",
       "      <td>15.650</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hour</th>\n",
       "      <td>127012.0</td>\n",
       "      <td>14.500000</td>\n",
       "      <td>5.781097</td>\n",
       "      <td>5.0</td>\n",
       "      <td>9.00000</td>\n",
       "      <td>14.5000</td>\n",
       "      <td>20.0000</td>\n",
       "      <td>24.000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                          count       mean        std  min       25%      50%  \\\n",
       "available_bike_stands  127012.0  21.080381  11.634433  0.0  12.33330  20.5000   \n",
       "available_bikes        127012.0  10.703650  10.576528  0.0   1.50000   7.1000   \n",
       "main_temp              127012.0   7.612343   3.103932  0.1   5.56375   7.5125   \n",
       "rainfall_mm            127012.0   0.098482   0.351068  0.0   0.00000   0.0000   \n",
       "wind_speed             127012.0   5.423212   2.876910  0.5   3.35000   4.8500   \n",
       "hour                   127012.0  14.500000   5.781097  5.0   9.00000  14.5000   \n",
       "\n",
       "                           75%     max  \n",
       "available_bike_stands  29.9000  41.000  \n",
       "available_bikes        17.8000  40.000  \n",
       "main_temp               9.7275  16.455  \n",
       "rainfall_mm             0.0100   3.300  \n",
       "wind_speed              7.2000  15.650  \n",
       "hour                   20.0000  24.000  "
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Display a table with descriptive statistics for all the continuous features\n",
    "# Bike Station number will be excluded from this analysis as it has little useful information\n",
    "df_db.select_dtypes(['float64']).describe().T"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "All of these values look reasonable and valid. The most important value to note is the feature 'hour'. This is a continous feature representing the hour of the day that will be passed to the model for training and eventually prediction. The hour ranges from a min of 5 up to a maximum of 24. This is as between the hours of 01:00 and 05:00 no bike station information is being transmitted by the API. So here, the important takeaway for our time series validation is that a day, which is one sample or split in time series cross validation is taken to be **20 hours long**."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### - Table with descriptive statistics for all categorical features"
   ]
  },
  {
   "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>count</th>\n",
       "      <th>unique</th>\n",
       "      <th>top</th>\n",
       "      <th>freq</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>main_weather</th>\n",
       "      <td>127012</td>\n",
       "      <td>8</td>\n",
       "      <td>Clouds</td>\n",
       "      <td>69608</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>weekday</th>\n",
       "      <td>127012</td>\n",
       "      <td>7</td>\n",
       "      <td>Thursday</td>\n",
       "      <td>18306</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               count unique       top   freq\n",
       "main_weather  127012      8    Clouds  69608\n",
       "weekday       127012      7  Thursday  18306"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Display a table with descriptive statistics for all the categorical features\n",
    "df_db.select_dtypes(['category']).describe().T"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Our database has recorded 8 unique weather types, and a total of 7 days in a week (reassuring). With respect to weather, it should be noted that 8 is not an exhaustive list of all the types of weather that may be reported by Open Weather Map API. In fact, 14 different types of weather conditions may be reported, as can be seen on the documentation page [here](https://openweathermap.org/weather-conditions). When training our model, we should keep this in mind, and protect for this."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### - Plot Histograms for all the continuous features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1800x1800 with 6 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Histogram plotting of all the continous features \n",
    "# Assigned to variable to suppress matlibplot messages\n",
    "continous_hist_plot = df_db[continuous_columns].hist(figsize=(25,25))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Points to note from continuous charts**\n",
    "\n",
    "* available_bike_stands is a uniform distribution\n",
    "* available_bikes and rainfall_mm look closest to an exponential shape\n",
    "* main_temp is a normal distribution\n",
    "* wind_speed looks to be a righ skewed unimodel distribution"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### - Plot bar plots for all the categorical features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "for column in categorical_columns:\n",
    "    f = df_db[column].value_counts().plot(kind='bar', title=column,figsize=(10,6))\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Points to note from categorical charts**\n",
    "* Clouds is the most common weather, followed by rain.\n",
    "* Days of the week are distributed close to evenly. This would depend on the way of the week the datascraping started on, and what it is currently on since the latest data pull."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Check for null values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "number                   0\n",
       "available_bike_stands    0\n",
       "available_bikes          0\n",
       "main_weather             0\n",
       "main_temp                0\n",
       "rainfall_mm              0\n",
       "wind_speed               0\n",
       "hour                     0\n",
       "weekday                  0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check if any null values are present in the dataframe\n",
    "df_db.isnull().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This is good news, we have no missing data/null values present in our data. Overall I am pleased with the quality of our data, and will now begin the process of model creation. The next step is check how our data looks. This is important with respect to linear or non-linear relationships, and what model is best suited for training on this data."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### - Exploring data relationships"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For data exploration, I have chosen two random stations on two random days to get an idea of how the shape of the relationship may look for bikes based on time of the day. I previously attempted to plot all the days and stations, but due to the volume, this produced meaningless charts. As such two scenarios will be investigated:\n",
    "* Station 2 on a Wednesday\n",
    "* Station 100 on a Saturday"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create new dataframe \n",
    "wednesday_2 = df_db[df_db['weekday'] == 'Wednesday']\n",
    "wednesday_2 = wednesday_2[wednesday_2['number'] == 2]\n",
    "\n",
    "# Resample each time to the hour, and set values to be the average of that hour\n",
    "wednesday_2 = wednesday_2.resample('H').mean()\n",
    "\n",
    "# Drop NAN rows (which in this case are the hours between 01:00 and 05:00)\n",
    "wednesday_2 = wednesday_2.dropna()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1a26e39f28>"
      ]
     },
     "execution_count": 77,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Nicer plots\n",
    "plt.style.use('ggplot')\n",
    "\n",
    "# Scatter plot the features which displayed a high correlation from above\n",
    "\n",
    "wednesday_2.plot(kind='scatter', x='hour', y='available_bikes')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 87,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Display image with trend line\n",
    "Image(\"image_1.png\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As may be seen in the above graph the data is not linear. It roughly follows the pattern I would expect, that is from about 7am to 4pm, bike availability is low, due to workers commuting. As workers return home, we can see bike availability gradually increasing again."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create new dataframe \n",
    "saturday_100 = df_db[df_db['weekday'] == 'Saturday']\n",
    "saturday_100 = saturday_100[saturday_100['number'] == 100]\n",
    "\n",
    "# Resample each time to the hour, and set values to be the average of that hour\n",
    "saturday_100 = saturday_100.resample('H').mean()\n",
    "\n",
    "# Drop NAN rows (which in this case are the hours between 01:00 and 05:00)\n",
    "saturday_100 = saturday_100.dropna()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1a26f01c88>"
      ]
     },
     "execution_count": 79,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Nicer plots\n",
    "plt.style.use('ggplot')\n",
    "\n",
    "# Scatter plot the features which displayed a high correlation from above\n",
    "\n",
    "saturday_100.plot(kind='scatter', x='hour', y='available_bikes')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAbgAAAEeCAYAAAD4jwBAAAABfGlDQ1BJQ0MgUHJvZmlsZQAAKJFjYGAqSSwoyGFhYGDIzSspCnJ3UoiIjFJgv8PAzcDDIMRgxSCemFxc4BgQ4MOAE3y7xsAIoi/rgsxK8/x506a1fP4WNq+ZclYlOrj1gQF3SmpxMgMDIweQnZxSnJwLZOcA2TrJBUUlQPYMIFu3vKQAxD4BZIsUAR0IZN8BsdMh7A8gdhKYzcQCVhMS5AxkSwDZAkkQtgaInQ5hW4DYyRmJKUC2B8guiBvAgNPDRcHcwFLXkYC7SQa5OaUwO0ChxZOaFxoMcgcQyzB4MLgwKDCYMxgwWDLoMjiWpFaUgBQ65xdUFmWmZ5QoOAJDNlXBOT+3oLQktUhHwTMvWU9HwcjA0ACkDhRnEKM/B4FNZxQ7jxDLX8jAYKnMwMDcgxBLmsbAsH0PA4PEKYSYyjwGBn5rBoZt5woSixLhDmf8xkKIX5xmbARh8zgxMLDe+///sxoDA/skBoa/E////73o//+/i4H2A+PsQA4AJHdp4IxrEg8AAAIEaVRYdFhNTDpjb20uYWRvYmUueG1wAAAAAAA8eDp4bXBtZXRhIHhtbG5zOng9ImFkb2JlOm5zOm1ldGEvIiB4OnhtcHRrPSJYTVAgQ29yZSA1LjQuMCI+CiAgIDxyZGY6UkRGIHhtbG5zOnJkZj0iaHR0cDovL3d3dy53My5vcmcvMTk5OS8wMi8yMi1yZGYtc3ludGF4LW5zIyI+CiAgICAgIDxyZGY6RGVzY3JpcHRpb24gcmRmOmFib3V0PSIiCiAgICAgICAgICAgIHhtbG5zOmV4aWY9Imh0dHA6Ly9ucy5hZG9iZS5jb20vZXhpZi8xLjAvIgogICAgICAgICAgICB4bWxuczp0aWZmPSJodHRwOi8vbnMuYWRvYmUuY29tL3RpZmYvMS4wLyI+CiAgICAgICAgIDxleGlmOlBpeGVsWURpbWVuc2lvbj4yODY8L2V4aWY6UGl4ZWxZRGltZW5zaW9uPgogICAgICAgICA8ZXhpZjpQaXhlbFhEaW1lbnNpb24+NDQwPC9leGlmOlBpeGVsWERpbWVuc2lvbj4KICAgICAgICAgPHRpZmY6T3JpZW50YXRpb24+MTwvdGlmZjpPcmllbnRhdGlvbj4KICAgICAgPC9yZGY6RGVzY3JpcHRpb24+CiAgIDwvcmRmOlJERj4KPC94OnhtcG1ldGE+CoPHh0IAAEAASURBVHgB7V0JvE3V919vnrzJezxzZfhJfklEqYh+oYFUUigqRAiVpFBJPZFISYYiSSmk4lX4qVTUjz8aRJMSxfO80Zvn+9/r6Dz37HPudM65Z7h37c/nvXPWHtf+7nPPOnvtvdcKcbAAFAgBQoAQIAQIgQBDIDTA+kPdIQQIAUKAECAEBARIwNGDQAgQAoQAIRCQCJCAC8hhpU4RAoQAIUAIkICjZ4AQIAQIAUIgIBEgAReQw0qdIgQIAUKAECABR88AIUAIEAKEQEAiQAIuIIeVOkUIEAKEACFAAo6eAUKAECAECIGARIAEXEAOK3WKECAECAFCgAQcPQOEACFACBACAYkACbiAHFbqFCFACBAChAAJOHoGCAFCgBAgBAISARJwATms1ClCgBAgBAgBEnD0DBAChAAhQAgEJAIk4AJyWKlThAAhQAgQAiTg6BkgBAgBQoAQCEgESMAF5LBSpwgBQoAQIARIwNEzQAgQAoQAIRCQCJCAC8hhpU4RAoQAIUAIkICjZ4AQIAQIAUIgIBEgAReQw0qdIgQIAUKAECABR88AIUAIEAKEQEAiEB6QvfKxU8eOHfOxxJnsaWlpkJWVpaqsUYWIR/2QJiz1wZJwDAwcW7RooU9H/FgLzeD8CC5VTQgQAoQAIWAeAiTgzMOeWiYECAFCgBDwIwIk4PwILlVNCBAChAAhYB4CJODMw55aJgQIAUKAEPAjAiTg/AguVU0IEAKEACFgHgIk4MzDnlomBAgBQoAQ8CMCdExABbh/n66A57/8C0qqDkNcBMCUns2haXyUipqoiFYExLEoKq+B+OgwW46FHn0Q61D7TIrlteCoRx1anwcqTwg4I0AzOGc0vLxH4XY4uxwyC8qF67wdf3lZkrLpjYA4FllFVbYdCz36INah9pkUy2vBUY869H4+qL7gRoBmcCrGH79ynQNPO6fRvWsExC9+tbMOrJnHnqddt26dFJ5nnvaG0/ySKkk2npYkKhB8mzytUEQWxZfhaVkBiiAE/IwACTgVAKMqDL90xYA0Bd8REL/4xZI4E17Yv7VIenWNCguR5ONpSaICIQpZfBmrVXFqrSNU2gXgaQW2ZVHFlbWSOJ6WJCoQ0eFSZQ5PKxSRRdHvQgYJRZiMgPSpNpkZuzSPa26tG0RD46Ro4Yo0Bd8R4L/wedqbGqsd0lw8LU2VU7M/PSqoNkXV3Oz/HpVn8hAjCmqxDl9V1rkl1ZIWeFqS6IKIj5J+ZPG0i2J10Q4HByRP1+V0fTO8cxpEhYcAfnPgFWkKhICZCNAMTgX6uKEEZxp2sKmnonuGFdHji/9UYaWEX56WJCoQmVz5ExytUEQWxQtmnpYV4CJqaqXChae57IpkUmw45DgJSqR9CRU1Uh7KOdqbulbvy4KKf74watgV6U5N4r0pSnkIAb8g4NuvwC8sBGelWtVaVkBNax9uOD8ZXsrOBHy1opYOaV8DP2PjaV/rU5OfV+fxtKc6w9mUBwWCGJD2NaAWAWeOuBQn7uz1pQ49PjZ4wc7TvvBDeQkBPRAgFaUeKKqoQ6taS0WTuhfR2oel35wUhBsyhq93pH0NUsUcAE97qq9JovR4B097Ko/pWtV7069pIVHtIe1rELUKm8ZeIWgXfD22Iqrd0+IjVKvdUUg6B552TqN7QsAIBGgGZwTKCm3wX7c8rVDEclE8zzztieFqTg3G057KY3qz5Cg4ml9RlxVpX8KIro1g9qfHANvGmdPISxv5UlzIq1W91zAuEpozvhE/FApp8ZE+8yDOptXuSHVI96j43D4WEGeRYj98XZsW++Bc3ldBrYpxKhSwCNAMzqSh5b9uedoktnxqlueZpz1VxqvieNpTeUyfxmY7uOFHnHkg7UtYueeksG6EshbXj1bs9n0Wyfebpz3xo3UmjPU/9d8/JWczZ27901OzknQ9eBBnkSsGtVU1i9SDB0mniAh6BGgGZ9IjoPVr1yS2Jc0O65Qmmf34umvuvm6N4KWvzq7BIe1rEF+qvpYT859gVmmcA087p7m61zqW/MyXp1216xyfWXj22ArG87RzXqV7vk2eVirDx+07XiR5HlDV6ssmE75NnubbI5oQ8ISALQUcetGeP38+FBQUQEhICPTp0wcGDhxY19d3330X3nzzTVizZg0kJSXVxVvpRuuL2Qp9eXO/tl1zH/yYJ1mDQ7p3mxQrdM0nHrSOJc74zD5XyW+M4WlvAEFVr/MuyvTtx+C94e29KSrk4dvkaa8rooyEwD8I2FLAhYeFwciRI6Ft27ZQWlIKEydNhM6dO8O5554LKPy+//57SEmx34vSbk8l/4XN0576w8+WeNpTeT3SGydEwrGCs0cNkDY6aJ0B6sGv1o0yyAO/hsrTnvjUygOt4XlCOPjSbbkGl5KaKgg3HK7YuFho2rQp5OTkCKO3bNkyuOeee4SZXfANp7E95teaeNpYbtS1Nr33OZI1PKSNDuIMUO3aFfLbOIFZ/XYKPO2UpHirdaMMVsqvofK0YsNOkVp5oDU8JzDpVkDAljM457HLzMyEI0eOQLvz28GuXbsgpX4KtGnTxjmL7D5jcwZs3bZViF+0aJFwYFuWyYuIiIgI1WW9qF6XLP7kce7AeJj+4UE4XVYJiTGRkD6gPaQlx3rNd/OUI/BHdkld/uYpcYbjmcaMbaxt7d3GFFdY/u+PXJiy8Qeoqq6FCGby6vlbO8Cl5xqrQXh56JmxKCyrhoSYcJ/HIrneMYmaNLlejM9jMX9QODy8QYpDWpocB1c4auUBd5A6BzwTiMYY1ARXPKqpy19l7MCjv/rubb22FnBlZWWQnp4Oo0aNgrDwMFi3bh3MTp/tse/9+vcD/BMDqjXVBDtYMvEnj7gh//nrnIRDZRFTERd5DeXUnk24w8lNBBWz1xUYnNEVlg+/d/Ds2lNVLUxe/71Pa096dCObbZaprq4U1jTxmpObA1GV3h+ZePDKNDYWlXVHFZD29XcRXlkBzZIi6+qIqC5hdUjNkGFfXeGolYcIqJVAibSvfRArcMWjmG6Fq9k8tmjh9Nu3AiAKPNhWwFVVVcHTTz8NV111FfTo0QP++OMPOHXqFIy/f7zQzby8PJg4cSIsfGEh1E+pr9B1ijIbAVE1Z/YPVSsO/FoTT3uqX4+1I1E9J7aFVk18MVwtjoVYXs3VmQfcNGM0D7jhzDnwtHMa3QcHArYUcLgYvWDBAmjWrBkMGjRIGKmWLVvC2rVr60bt7rvvhoULF1p2F2Udo3RjewRwrUmLqS2tggEB5Df48LQnkPUQsnybPO2JB63p5UxF7Bx42jmN7oMDAVtuMjlw4AB89dVXgNdx48YJf998801wjBj10nIIaDW1xQsCnvamw/wGH572VIcoZNV6RMD6+TZ52hMPWtP59nhaa/1U3n4I2HIG16FDB8jIyHCL9qpVq9ym2z1Rjy9urRhYgQetfdCjPB5m9uW8F98mvoi1noMTjxqoNbbMC1We5nlWokUesCz2CWkjg9ntG9lXass7BGwp4LzrWmDnEr+4sZdq1jv0QMcKPGjthxWEtB4vZnENTe16ph5CVg97llrGU8RASx1UNrAQIAFn0nhqfbHyX9g8bUS3+DZ52ggetLZhBSFthRezHkLWClhqfR6ofGAhQALOpPHU+jKI4nyG8bQR3eJNKfG0ETxobYMXyjyttX67lNdDyPLY8bRdsCA+AwcBEnAmjSX/4+dpT2zxW6B52lN5PdK1mlbSgwetdeihmtPKg9bZvNb29SpvBSz16gvVExgI2HIXZSBAjy8D58DTzmlK9/wWaJ5WKqN3nFbTSnrzo6Y+VM05u9tB2uggzua17GA0mmel9qyApRJfFBe8CNAMTsXYi1/cap1LYpNa1zys8LVsBR5UDJ+kiB6qOfF5wFk4YoJji/V6G/jZO097W4/Z+fTAUksftI4Dti3WoeW3raUPVFZfBGgGpwJP8Ys7s6BccDKJFht8DeLLQK2BXSt8LVuBB19x90d+8XlQOwPjZ+887Q+eA7FOreOAmIh1aPltByK2du0TzeBUjBz/hc3TKqr0uYgoIH0uqGMBK/CgY3dUV8WPP097qljrbN5T/XZJt4LD1IJSqe1MnrYLlsTnGQRIwKl4EgJBNaei27YvElJcCKElxRDCjAILf1XMDxz+RUSCA/8io9gfu8bFQ229BK/7q/V5oA+FM1BrdZiqdRyQi6KKGsm487QkkQjLI0ACTsUQiV/caq1GqGiSiniBQGheDkT8/QeEH/sdIv5i17+OQFjeKQjLzxH+QirPOjb1VJ2DuUKqSawPtfUbQE1yKlQ3bg5h7TpAdEIKVDc7T6Ah/MzPR3wecOaGL1mk7RbEtSfnPqDgNTLwRqp52hMvwzunAXoRx3JoHxRpX0O9yFDmGeKskEOagn0RIAGnYuzEL261ViNUNElFOARCykoh8ufvIeqn7yDy4H7hGpaXzeVST4YwbxXhOcyNEv6J4QOAhv/cowCsbHk+VP7rQohreyEsZn9VrdoBMG/zdgzi2hPybpZlHK1Gq1fvyzrrtqjaAUijGTVfQnJcBOSWnhVwSFOwLwIk4Ow7dkHHecTvP0HMni8gevcOiPpxL4RUS9dLjAQEBWDULweEP9h8puXamFiouLALlF/cDSo6XQGVbdoDhNpjBsCvG/K0Edje160RvPRVpuDTDh3fIO1L4HnmaW/qEmfjpJ3xBi3r5yEBZ/0x8huHWtVSWst707FINkOL2/4hxOz4CMJzT3lTxLQ8oWxWiQIY/zDU1ouHskt6QNmVfaDssl7g8GFdz+hO6LF+pZXnj37OF4Qb1uNgf0j3biP3CO6qHT36QNoZV+jaM54EnIpxE1/sdj8ro1UtpbW8K+jDjx+FuC3rIe7TTRB+4pirbD7H10bHQG1C0j+bSXBDCVtjYptLcKOJuOkEr6FFpyG0tMTn+vkCocVFEMcEM/452HpdxUWXQukVfaC05w1sbS+Vz24qLc5ccNZj1joiP+PiaU8AWaEPnnikdGMRIAGnAm/nFzsW99VzsYom/VKEf4HwtKdG+fw87am8JJ05sY3+5jOI/2A1RLMZkNQ3sySnSwLXxapwA8g5raGqeSuoatEKaho1gxrcKML+HEyF6G0QBB1b0xM2qGSfhIjjf0J8TiZU/XqQbWRhm1fyc72tSsiH6tTofbuEv+TFs6D8ku5Q0vtmKOveFxxR0T7V5Y/MZnsCwD5pnYGJsy9/4EN12hMBWwq4rKwsmD9/PhQUFADaYOzTpw8MHDgQli1bBnv37mWb28IBN4BMnjwZ4uN9W2T2Zhj5FzlPe1OHFfLwxpF52hOPWl9IWH9IRTmEvbEImqx6CcJP/u2pSUl6dWoaVPz7Eqj8d2eouOBitub1bzYj02dTAM7uBOHIBCSwvSNlrOUY9kydYs8ehlAm/KJ+/gEif/nnj6lSw07nC2me/oXU1EAMW0fEP1y3K73qeigecCdUtuvoqajf0p0/2szaZEIzML8Nb9BWbEsBF852qo0cORLatm0LpSWlMHHSROjcuTN0urgTjBo5CsLCw2D58uWwdu1aGD16tO6Dq8eLXXemVFSo1ViylhcS7oLE2Vr8u8shrCDPK+4dbNwr2neG8m5XQ9mlvaCqZVuvyvkjEx4fKLv8P8KfWD9ugone/zX72wVR3+/2Ss2J63b1tmwQ/nBTStGAYVDa+ybDZ3X8RxpPi33055VmYP5ENzjrtqWAS0lNBfzDEBsXC02bNoWcnBzo0rVL3Si2a9cOdu7cWUfreSO+2LXstBLX8fBFIq554A/c26C1PLaj1ViymhcSqv7i1712RrCxtS5PATcb4NpVCXvpl+HalYU3auAxAfwrGjQSgM3Son7YA7E7t0HMzq0QnnXCU1ch8reDkPL8o5C45Bkove42Vs8oqElr4rGcHhkC5aNNDyyojsBBwJYCzhn+zMxMOHLkCLQ7n+mRnMK2bduge/fuTjH63Yovdi3n4LSqhLSWRzSMfqnFbnsfkl57DsJPZXocjGqmGiy+8Q4oueYmqGnY2GN+y2XA2SYeF2B/+ROehIhff4TYr7ayHaEfQHime9ul4czaSsKGlRD//htQ+p8boXDwfX6frYofbc4fXJbDlBgiBHxEIISpqfAj2ZahrKwMpkyZArfddhv06NGjrg9r1qyBw4cPw5NPPims0dUl/HOTsTkDtm7bKlCLFi2CiooKPotXdARuamDnodSEG5fsAjToKobGSdGwaewVIunx6m15dzweyy+F6R8ehNNllZAYEwnpA9pDi2TvN2IczSuFGZs8lw/Z9zWEz5kCoT/u99iv2m69oObOcVDb6wa2QOd5q4m3PHhs2IsM7rD0onhdlpC9uyBs01sQuuU9CCkqrIt3d1PTvQ/UjH4EHJe4f0b04tEdL67SvB0LM3l0xTsfTzzyiMjpqCjvNU7y0sbE2FbAoWBBAXbxxRfDoEGD6tDasmUL4N+cOXMgOtq73WnHjqnbiq5lBvfA5sOCJwKRcfRJtrB/a5H0ePW2vDsetao5PfGAth+Tlz4LcRlr3e6KdDBBVtqrHxQOmwBV5/3LY9+dM3jiwTmv1nt3WKqqm6lrY7/4BHJeXw7nnzjkVRXll1wJBSMmQyXbVKMUdOdRqREXcd6OhZk8umBdFk08yiCRRbRo0UIWZ7UIe5hZ4FDDSeeCBQugWbNmEuG2Z/ce2LhxI8ycOdNr4cZVbRiJKiEtjja1lseOimpOtW5e+I0IznTsjo+hyfD/QD03wg0FWwlTwVVlfAu5TyzyWbhhH5zbVKIxzrKB7dTEDSX5y96Hh4Yvg48vuBYqwt1/FUfv3QmNxt0MDR4bIag9rdQ3W4+FlYAkXnRDwJZrcAcOHICvvvoKmjdvDuPGjRPAGDZsGLz66qtQzc4bTZs2TYhr06YNPPjgg7qBpWdF4jqe2jq1lsd283GXjFPgaackxVulNbxQdj6s/rypEPv1dsUyYmTZpT0hf9wM4cwafi3DP9vvxXRvr/zRBp72th4z8+FYPjCiLwD7y2az3npMfRm/YQUzFJ3jkq0YdmYQ/0qvug4KxjwG1U3M/5pWeh5cdoASCAEDELClgOvQoQNkZGTI4OnWrZssjiJcI1BcWStJ5GlJogLBb0xIjz8CjUcOdvtirjq3jSDYyrtepVCj71GyJWT7LikLncddooVDx0LhrSOEowMJ7yxza80FVZwx7GOi6KbhUHjXJGAHQH0HUacS/POANAVCwEwEbCngzAQskNqOj2I7/ZwMFiPtS6ibRbK1JFxri9+4ymVxPNBcMGoKFLMXsZ4W97UedXDJsNkJTH2Ju0iL+w2B2M8zIHH1SxBx9LAiV2j4OWH9CmbebAM4JjwOcPVNILryUSzgp8i658FP9duhWq3r2nboo514tOUanJ0AtjKvcZFSgcbT3vCOdiMb3XejW+GG6sjMVduheOA9ugo35A/VYs6Bp53TbHnPvBHgUYHM17dB7rQFZ/zQuehIGDtXGD77YWh8Tx+IYmbBgi2gcMGNLiPX/yJcjxep2x2tBTet69pa2qaycgRIwMkxsUWMHj9mreo9tB3ZaEx/iPzjF0XMaphh45xpL0D23FV+O7Csx2YbReatFskEXUmfW+DEm59D3kPpgGbKXAV09po2+Q5ImTUBQi3ugcFVH9TEW0G40EYbNSPnvzKkovQftn6tWfwxYyNqbQdqUe8lrFoIiezP1Uk13M6e89gCqE0RXYT6B46gU4sxO6vCAfhrbxWswSS8vQTQ3JdSiPtsM7N3+TmcZscKim6+yza+6ZT64k2cFYQLbbTxZqSMy0MzOOOw1rUlPX7MvDqPp5UYDikvg9RpIyHJhXBDtzD5Yx6FU8+v8btwU+IvWOLQGDSeGzyxZgcUXzcI8MiFUghlVlGSFz0FjUb3gwhmCiyQA//88rQRfQ8ajYIRYOrQBs3gdADRjCrwx4szNzGo+TH7uusNLeg3ZOevIpkna6VQxbaq5z65GCrbXqiU7Je4YF/Uxxly3tR5UHTL3YIgi2b2L5VC5OFDwlpp4ZD74PRwttsykvnBs1jQOpa+Ps9K3Rd5UOvrMeg0CkogWijOtpZM9MTQDEsmWvnHBXT0Q4czORRu+OPGHxcf9LLIEP7nb9Bw6t3MaPBxvgmBLuvSgwm3l1UZQ9bCo7fWMxSZ9jFSC58+NqU6e+Pdn0LonEfc+qurYv7ycplQdGUNRXXjXhZ0haORY+mKVSvw4Io3Pt4Vjnw+f9F2sGRCMzh/jb6f6zXySzHq22+gweOjAT1UKwU8t1Vw7yNe2Y5UKq8lTg9VrZb2rVa29sahkNW+CyQunwv1Nr8N7AtWxiIeN0gbf4vg9QCPbgBTd1ohWGEsrcCDFcYiUHigNbhAGUk/9SP660+FmZuScKtlnqhzZi6GgtFTTRFu2GVeNcvTfoLF0tXiYfF8ttMy65X3obL1BYq8ouBLYG6LhLU5pr60QuDHjqeN4JFvk6c98aDH7mZPbVC69wiQgPMeq6DLGbvjI2jwxBhAH258qEmqD6cWvgOlzEebmYEW9V2jjx7CTy7bDAUjHwaHC0/nkUz13GjsAIhfuxRAYbbnunb9U6wwliIP6N0DbcUi7UsQdzerte/qS1uU1zMCpKL0jFFQ5oj7ZL1gUzKkVmrOC8GoanoOZD+3GqrZ1exgpKrW7L6qap/5pSscdj+Ude8L9edMhqiff5BVg5ZQkpfNgZj/fc4Ok7/gtzOLsoa5CCuMpciD2vUtUnFyg2oySTM4kwZAqypDa3l33UZjv/XnTgEl4VbBZgVZi9+3hHBz1wdKkyKANkCzXvmAGWZ+FPCIgVKI/n43syXaF9ATBAV1CPAqTZ5WVyuVUosACTi1yGksp1WVobW8K/bRnmHygumKB7jLO18Bp15YC7VMPUnBhggwayh4TCDz1Y+gsk17xQ7gWmvqzHGQ/MIMgMpKxTyBHCl+OKJDYdxR6au5L1HFmRYfoUrFGcjYmtE3UlGagTprU6sqQ2t5pW7HMssXwsxNIRHtSWY/vcwyO+4UWKQoLxGoZscETrLZXNLK+RDPvBUo7bSM/3ANRB3czzYRvQLVzc71smb7ZxM/HMWe4FEcXxwRiypOsTxdzUXA1Bnc9u3b4c8//xQQwLNoDz/8MDzyyCNw9OhRc1ExoHVedcHTnljg8/O0p/J8esxXWyEl/QHFl13plX0g+5lXSbjxoNmZZptOUF2JM/Lqho0VeyIcDh99A8R+ukkxPRAj/fHhGIg42aVPpgq4t99+GxLiEwSs0Flp69at4YILLoDFixe7xS+LOcdEQTh69GgYM2YMvPfee0L+wsJCePTRR2HkyJHCFWmrBq2qDK3lnXFBL9Gps+6HkJoa52jhvoTtksx5agmAi114sgI2ixBVUmZaoDcTsoqOl0Hmyq2A46wUQktLIPXpiWdUlmwzSqAH/kORpwO9/4HWP1NVlEVMANVPqQ8VFRXwyy+/wOOPP87cWIXD4MGD3eIcznaGoRBr27YtlJaUwsRJE6Fz586wbds2uKjDRTBk6BBY+/ZaeOeddwQh6LYykxK1qjK0lsdu48v9vQ2fw9PvPgi4k44PpZf/B3Iff0l3Fzd8O2bSzioptUarzeRfj7Yd7NxcLjvPWLH5CsHcl9KxEFRZ4owuZ9ZSqPGzAW09+qS2DvxwRLUkOruPiwCfjwmobZfK+QcBU2dw8QkJcPzv47Bnzx5o1aoVM48Xyda1PS9sp6SmCsINIYmNi4WmTZtCTk6OUE+fPr0FpPCK9fojiF/9aheikSexDjNnDm9k7IXH1j8GsZVlMpgEbwAz2cyNfUwEcigorZZ0j6cliQFOFPcfCieXfABVLtbccE2u0b03QNQP/xewSIgfjpvGXiGsvSFNwb4ImCrgbrvtNnjggUmwaNEiGDhwoIDivn374JxzvD9flZmZCUeOHIF257eD06dPAwo/DHj1l4pS/OrPLCiHw9nlwhef0KgP/8Q6zDoQGlpYADPWT4XU8nwZ1xUXXvLPmpv1DPLKmNUYUVQhVcvytMbqbVe8qlU7OLn8Iyjp1U+R9zA0uP3gYKj33uuK6RRJCFgJAVNVlNdeey307NlTwCM6Olq44hocCitvQllZGaSnp8OoUaMgrl6cN0WEPBmbM2Drtq3CPQpXPNTpS0BL484B1Rlm1OHMg6v7CLZ2JuONzZIj2EsqtOBvWbFa9ASw4iNo+M/aqCyDHyIUefRDO0pVJsT8CtlFZ7UGCTEKeP1T0Ew+lXhXitONx1fWQ9Xa5RD+7MMy9TWu1dZnLngSj/4G1U+zWb6Pngl041EJAJ3izOTxaF4pzNh0EE6XVUJiTCSkD2gPLZJjZT0zk0cZMxaNMFXAISa4/rZ7927Iy8sT1t6qq6uZxSC5gVgevyq2ZvT000/DVVddBT169BCSExMTIZepKnH2htcEpgJVCv369wP8EwNuWvEloG7eOSBtRh3OPLi6V7LIkDL7IYj69n+yIhUNGkHO7BVQU8pUlvhnUFDi0aCmITE6lAm4s60h7WoszeTzLIfu73Tl8eoBENmoBaQ+PgbCFTyDh216G6p//5kdH1kOtfXPaE7cc3cmVVcevWlQRR4zeXyUnb9DzRAG1BJNfe9bxaMKZvKIvNnBm4CpKsrvvvtO2AW5Y8cO2LB+A2IGx48fF1SWAuHiHwrABQsWQLNmzWDQoEF1ubp06cI2mvxXoPHatWvXujQ9b8QdjGrt1SEvYh1GHwjFc09x2zbK4KitFw95894M6A0Esk6bOA5KvFgxDl3qnHz1YyhnamulIKzLjb0RIn7/SSnZpzgrrEv7xLCfMtNRBf2ANXUGh0cDpkyZIuyAxPU4DO3atYPff//dbQ8PHDgAX331FTRv3hzGjRsn5B02bJgwA5w9ezZs/3Q7pLJZ3PTp093WozZRXIjW8gUl1qGWBzXlopmtwSRmc5APaIgXv8LRnFOwBTPGwW4Y4+zs1AvvQPLiWRD//moZ++FZJwT3O7kzXoQydmZSbRDXpbF8sO5oxb7j0QTsvxjoqIKIhO9XUwVcdna2INyc2Ua9co3CeSznPB06dICMjAznqLr7uXPn1t0H8g1+7eILAb/28AeAM0J3O77CmQ+w1KcnKB7kzp3yHFRc3C2Q4aK+aUWAHd/JnzQLKs+/COrPf4x5mDi7bolVh5aXQeqM0XCa+QUsvOPMR6evTWqdufj6m/CVP6Pyi0cVnH/bRrUdaO2YKuBQxbhn9x7oeulZVeK+vftsods1+0Hw5Ws3hB3WbTDjXggtKZaxjbYJS/vcLIunCEJACYGSvgPZMYLzhOcpLD9XkiWEUUmvPgfhJ45C3kOzfT5ionXm4stvQsK4joQeQpa0CvoNiKkCDg9r40aRiy++GHDTyAsvvAB4TMBfqkW9YBMfYtxNKR4GdTd70qtd53p8+dqt/9wjEPHXEefiwj3alxSclcpSjImwAo569FTsh/MXt9HPgx798LaOyvad4OTSzdBg2giIZJtM+FDvo3chLPukYMfSEev97matMxdffhM8z3rRegjZYHue9MJeqR5TN5m0b99e2FCC597wuECjtEbC5hFch7NyEB9iLefgtPaP18vztFh/2JpXII45LuVDVfPzIOfxRaZ54kZ+rIAjj4saWuyHWWca1fCstUxNWhPIenkjoLUbpRCz5wtImzgIwhR2Xyrlxzhx5rJiUFtVh6z53wBPu2pXz3g9hGwwPk96joFzXabO4NDQ8rnnniszzcWrLZ0ZtsK9Hg+x1n5487UbeehbCJs7VdYU7pjMZscBHOxqZrACjnr0P1D64SsWjphYyEl/DZKWPgsJ7y6XFUfTXmnMWzg6xzViA5M3vwkZk1yEOHtSq51Boap1g4i/nyexj9gO8ou4BarGwdQZ3FNPPQVoicQ57Nq1C1586UXnKMvd81+GPG0Ew56+dtFSCfr1CmHnCvmQO/V5qG7eko82nOZx42nDGVLZIM83T6us1h7FQkKgYOw0yHv4WXCEyl8n4acyIe3+WyDqO/m5S7076Ok34U174uxJrXYGhUXrBtGg5fgP//zwtDf9cJdH7GMwaBzkT6Q7ZHROu/vuuwUDy3goG8OXX34JS5YsgSeeeELnlvStblinNIgKD4EwtqqO1+Gd03xuAL+i0KGiv2xR4robvlz4UHjrCCjr3pePNoUWXwZazhOawjjXqNgPLS81rkrbkcX9hkA2m83VRsfIeEcnqg0fGQ7oksnqQevsSQ8h6+/nibe3ytNWHyNf+DNVRYlWSEpLS2H6jBlw3XXXwfr16+GpmU9Bq9atfOmD4Xnf3J8FFdVnrK3UsOvqfVnQqYlv6j7xKwqZxy8pXx0rulMz1MtYC7E7t8lwqWjXEQrumyaLNytCfBloOU9oFu/O7Yr9cI4LxvvyblfDqRfXQYPH7oGwvDMfrSIOeKwg9cmxkDf5WYARE8Voy11xtqRVxai1U/5+nnh7qzytlX8rlTd8BldbWwvOf3379oWrr74a1q1bB7NmzYLzWp4npFsJJJ4XrV95WJ/WOkQByasZwv/6A5KYnUA+1MQnCrvamD8iPoloQkA3BCqZLdOTi9+HKgUVeAj77afMmwphrz6vW3t6VyTOnuyuVXCHS71I6Wufp92VtVua4W+7AQMGuMRo0qRJdWmbN2+uu7fajR5feVrrUBSQ7KhF6qwJEFpxxo6dM265jy0A3PlGgRDwNwI1jZtD1ivvs2MEIyHqwF5Zc+ELHoekv49CwbgZsjSzI8TZk921Cu5wTGZnm3JLz3rRQDpQg+ECDs1z2T3gV55Wp4hiHSioUNgh7UtQEpBJK56HyN8OyqopumkYlLvYzi3LTBGEgA4I1DKNwan5bwlqyZhvPpPVmLDuNebNIg/yHmWzOYXNKbICFKEbAriHYPanx6C6xgHhbCOBmj0EujHj54pCmOFiz6b7/cyE2dUfO3ZMFQtmfuUdL6oQhKwoIGc1zIb2jw6VmeKqbdkWji/dBI7IKFV9NKKQmTj60j878Gk5HpnZvZS5DzMD3+8rQl3a83rImcG8xltMdW45HBXQU8sjbm4TvRVgtbjrc2H/1gotuI+ygzcBw2dw6AXgoYceEpCbM2cOhLBtxkph6lT5+S2lfMEaJ6pSsP8hlRXQaOS9MuGGRpSrn3/D0sItWMcvaPrNPMLnTnsBauKTIEHBSWrsjo+hAVOpZz+1hPmVs+5HWCCNl+LyRiB10Kkvhgs4tFYihsaNG4u3dNWAQCKz/6dkiqtgxGSIbXcROqvTUDsVJQS0I1Aw4UmoTaoPSSvmyypDFWbDx0YIxwwcCscMZAUCPMLdDmk9uq60vKFHvVasw3ABN/SOoXU43HXXXXX3dKMOgUi2iB+/YaWscHmHLlA0eAzI/QDLslIEIWAIAoXDJkBtQjIkvzADeL1N9L5d0HDKMDg1ZxU44uoZwo9VGxF3SCN/ao4QeeqX1vV/T/VbKd1wAcd3fv/+/fD5559Dfn4+JCcnCzYpO3fuzGcjWgEBVE3i+gZbSJWk1jJVTx6zVsL0v5J4IggBsxEoHnAnxDMtTvi00YDHBpwD7rhs+BA7MP78GsBNKsEa/K1CdF7eCHSMTRVweLB748aN0KtXL2jVqhWgf7j58+fDTTfdBKIDVKUBmDdvHqBgTExMhKVLlwpZfvvtN1i8eDFUsgOlYUzvj45QrW60WalvvsQlvv4CRPz9p6wI+uSqbnqOLJ4i/IOAv1VK/uHavFprB9wBueWVkPLMRJkpuahfDjAhNxROLXjbFCEnjqVaW5R6oBpMKkQ98HJXh/TEn7ucfkj78MMPIT09HUaPHi0ItXvvvVegN23a5La13r17w8yZMyV5Vq5cKRhtfuWVV2Do0KGAdCCHCOamJH6d/MhFxb87QxEzx0XBOARElRJ/6N44DuzXkrB7ctYytgEqUsY8HnVp+OAQCC06LUvzd4Q4lmptUerBH6oQtdqz1IOPQKjD1BkcAti0aVMJjk2aeD6M3LFjR5mRZtyNWcbMfmEoKSkR1J2SigOJYCpJwasy5/kcjwLkTp1nmGpS/NoVjyrgDxPVH8EW/K1SClQ8y9jZTPRqkTptFIQydbtzQE8EKOSEmVxCknOSX++1jqUev4lgUiH6dTBZ5YbP4JzNdA2+fbDg/+2vv/6CiooKwPNo6PR0yJAhPvcbZ4Gvr1oFw4cPh1XsOmJE4M5i6m16C6KYKxw+FNzzoKFeAsSv3WCfuaBKyTnwtHMa3UsRKL+kO2TPWw21zPUOH0Qhh54xjAr82PG0Jz7oN+EJIWPTDT/o3b9/f6966MlUF7rZQXc74hrcy4tehgs7XAhowPnzzz6Drdu2AZ6zUwoZmzNY+hnL5osWLRKEq1I+T3ER7JwZeiI3NDBPyZE3XAQhRYWSZmv/1R6qNu4GtgApifcnjzcu2QWoyhED2u/bNPYKkfT66k8evWbCi4yu+DyWzwyGf3gQTpdVQmJMJKQPaA8tkuUvbC+a0JzFFY+aK9axAiUeQ/Z/AxH39oeQ0hJZS7XMvmXVqi0A7JiBv4M4loVl1ZAQE+7zWOr1m/Cmn0o4elNOrzxRUdbX1hiuovSXqa4vvtgB4+8fL4xdT7ZpBdfiXIV+/fsB/okhS+U5MbWWBMR21VxTnpoAUZxwwz2UpybOgsp/3A451+tPHnkTdkj7gqWozilh3whY1uoqTldY4s/8+etanIW9sojhUHSWNvDOFY8GsuCxKUUem7aEqLlvQAPmVie07MxSg1hRKNt4EnLnNcxTwbtQWy9BjJZd9x0vkpigmn5NC5+9fIhjWcejj2Op9TeBnfK2H3U8ypAwJsIOlkwMV1E2atQIvPkTh+i+++4Tb91ek+vXh++++07Igzss01g7gRaivv0G4j7PkHWrpP9QqGSbS4wOWhfDRXWOmQv6emCGgtqfvv304NEOdVRceIng/VtRXfn7T9Dg4TshpKTYZVfQviK6sWImFoVr+vZjLvP6K0EPX5FW6Ie/8DG6XsNncL52MCc7R1YEd14eOnQIiouLYdiwYcKa3f333w/Lly8XXO3g1H3ChAmycraOYGeGkl96UtaFmuQUKBjzqCzeiAiti+FaF/SN6KM3bYiCGvP642CuNzwESh5RyCnN5KJ+/gEaPno3nJr3JihZPEHjwc6Bp53T/HWvh69Inm+e9hfvgViv5QWc0lnl6dOnK47Fyy+/rBgfCJH1PlgNkUd+lXUlf+wMt2obWQELReACPgoEMfi6oC+WM/saKILabBzF9t0KOXYYHN3wnJrzusx2JVrGRwfEYkDa6KDHs2CFfhiNm7/aM1xF6a+OBHK9oafzIWnlAlkX0RxXaZ+bZfF2iRBVnHZ3LskLZp62y3hYiU9ByM1dBbVR0TK2ovd/DQ0eHwNsh5ckDdfcosJDAOUaXpE2OvBjz9Pe8DPmskZ1psxQRN/XLfCWW7zBQY88lp/B6dFJu9eR9No8CC2W7pp0sKlt/oSZtu6aqOI0e7FcK4goqNE/IH694wsNaQraEajo0BVy0l9jM7YRzGNGpaTCmN07IPWp8ZCDXgj+2TncqUk8vDe8vSSf0YQez8LHv+SDOA/F60c/50PvNilGdyUg2rP8DI4zsxgQoPvSiQhm1SEuY62sSEm/IVDVxtwfs4ypII0QBfWKQW0Fv1pIU9AHgfJLroTsmUvAoeAvLnbnNkh59iF9GtKpFofUvKaqWvVQc6pqOAALWULA4eHvXIUt7oj3fWO920Vpt7Hxdudd8tLZcmPKbKt0wagpdusy8UsIqEIAvdGjU1SHgufvuO0fQvKLT6iq1x+FxA1HWowf8GpNntbKt7fvHq3tWKG8qQKuqIidW5k9G26++WYYPYbp1FnYtWuXxI7kNddcYwWcdOfBmx9C9P8+B3QjwofTzGJJbWIyH000IRCwCJQxz9+5jz4PqJrnQ/z7qyFRwc8cn88IWo/Zl7g2nRYfIdik1Fvl7c27xwisjGjDVAH30ksvQWxsLKxcsULwAIAdbt++vSDkjOi8mW14/CGwWW0Sm73xoeqc1lB003A+mmhCIOARKO1zC+RPlv8msOOJby6C+PUrTMeAn23xtDcM+lvlXVBaLWGDpyWJNidMFXAHDhyA8ePHQ0pqah2MSUlJUHjaeCvidQwYdMM/+Dwd9/E6iPzzNxk3+WMeq1tUlyVSBCEQ4AgUs7Xn/LHKx4SSFj8NcVs2mIqAv2dfenSukG2Gcg487Zxm93tTd1Hi7A2FmbOAQ1NPSczxaaAHd7utQsrLIOl1hWMBF10KuB5BgRAIZgSKbr8XQgvzIfEtqTk+VF7Wf+4R4Vxo2ZV9TIFInH2Z0riXjdZwB+J52stqbJHNVAHX+5re8AyzSoLWSBxsu+TBgwfhjTfegL59+9oCPC1MuvshxL+7HMJyT0mqx+3CBS6+XCUZiSAEggABdOqLXgbiN78t6S16CU+ddb9g7aSCfRBSkCOAe3VqnHZ7KuzdkReyaYypKsrBQwbDFVdcIXgEwJ2UL774InTt2hUGDhxoUzi1s40/2gQFR6alV/eHyvM7aG+AaiAEAgSB/IfSoaTnDbLe4Jm5BtNHQcQfv8jSKAKgcUKkBAaeliTanDB1BodOSm+99Vbhz+Y46sZ+wttLIJQzKOtgtjUL2BcrBUKAEHBCgL0/cme8yH4vRRDzf186JTBHl8VF0GDKMMh65QOoSfPsRFlSOMCJ6b3PCRrDBIYLOLT0703o1KmTN9kCKk8oU0vW27hK1qeiAcOgprH+1jFEdzXOFjjokLIMfoqwMgLsAHjOM8uh4UNDIeqg9N0Szn5PDZn7nayX34Pa+EQr98JQ3twtjxjKiAGNGS7g8GiAp4Azu9dff91TtoBLT1zzMoRWVkj6VRsdA4V3jJfE6UWI52GwPrKCrxeqVI/RCDiYvcpsZnw57f6BEHH0sKR5pBs8NgJOLXgbHJFkYUYCThAQhgu4VatW2R5WceZTUnVYN0edYSf/hnoKJrmKbh0Btcwljj+Cx7N4/miU6iQE/IAAztBOPbca0sbfDOE5WZIWon7cBykzmd1KNtMDDzsq/PHbljBDhKEImLrJBHuKm0t++OEH2PH5DvjxwI8CbSgCKhoTZz56OupMXLUQQjjr6LX14qFo8BkLLyrY9FiEP3vH0x4roAyEgIUQwLW2bCbk8HfDh9ivt0P9F2bw0TLaH79tWSMUYRgChs/gnHv2++Hf2TGBZ5jXiyqozzxy5+XlATornTZtGrRp08Y5q6Xu9Z75hJ84BnHbNsr6WHjbvX719ebuLJ6MGYogBGyAQFXLtpDNPBA0ZBtMeA8E9diRgupGzZjKf5zLnuj923bZECUYgoCpAm7hiwvhuuuug0GDBgGuu+FZuA0bNgjHBdw5L503bx7gZpXExEThiIGI1MaNG+Hjjz8WzH517twZRo8eLSbpesWZjp6OOhPY2hue33EONczWZNGtI52jdL8PpsVm3cGjCi2LAJ5/y522EFKYO50Qzh1J4qvPQTXbsIXHbpSC3r9tpTYozjgETFVRnszMFI4IoHDDgFc8A5d18qRbBHr37g0zZ86U5EGBt2fPHnjllVdg2bJlgtCUZNCREM3x6OGoE9feFGdvQ8eCIzZOR66pKkIgeBAoZcaZ8yc8KeswvmlS5kyGqB/+T5aGEXr+thUboEhDETB1Btfx4osFw8rdu3ev6zR6E8B4d6Fjx46QyYSjc/joo48EYRkZeeYQY7IfzX2JMx89HHWiuaGQaqnx05qk+lBMBpWdh5fuCQGfESi+5W4IP5UJCe8sk5RF1WUqOwiOZ+Sqm58nSdPzty2pmAhTEDBcwM2ZM0eYqWFvcYPJ/PnzAVWLKSkpkJubC0eOHAFUL/oaUOD9+OOP8OabbwrreCNHjoR27dopVpOxOQO2btsqpC1atAhQUKkJuF6otqzQXubfELllvaxpx4gHoWGLc2TxaiI086imUR/L2IFH7JK/+DyaVwozNh2E02WVkBgTCekD2kOL5FgfUTyT3V88qmLGRSFDeXxyIdTkn4Kwre9LuAkrOg2NmafwynfYAfH6Z429i5kM5VFs1MerHXj0sUu6ZzdcwDVu3FjSiRYtWtTRzZs3B5ydqQkoLIuLimHhwoXw888/w9y5c4WzdKL607nOfv37Af6JAQ08qwlaZ3DJi56GKG7nZA3b7px5zS3gUMkT3w+tPPL1+YO2A4/Yb3/x+ejmw3A4u1yAFnfmTn3vW8EzuBqs/cWjGl5clTGcx8lzIO3vo7KD4CF/HQG490Y4tfAd2Rk5w3l0BZabeLN5dH53u2HT1CTDBdxdd93llw7jLswrrrxCmB3izC2UnXc5zTwVoPsdKwbBaslH78hYw3NvtPYmgyWgI2jnnp+Hlx3wzp69AtLGDoAItmPZOUQd+hZSZj8IOTOlngmc89C9fREwdZMJwoZHBP744w9hVyRuFBH/fIX00ksvhe+++04oduzYMahm61q4y9KqIWHDStk2ZuHc28B7rMoy8eUnBPjzhzztp2aDqtpatis5e+4bUJMg/+CN3fEx8wj+fFDhESydNXwG5wwsHvBGVSIKufLycoiOjoaKigrhTJw7U13pzMXOoUOHoLi4WHC1M2TIELj++uvh+eefh/vuuw/CmX26Bx54oG6tz7lNK9yHMGPK9TatkbFSdPNd4KiXIIunCP8hIFquwFkUChbcRYcbDYwMdB7RGLRxQ0kOnpGbPFT2cZn45stQ1aI1lPa+yRhmqBVDEDBVwL322mtw0003CVv6b7vtNli3bh2sXr0aoqLcv2CmT1f26PvYY48ZAprWRuKZcOM9BtQye3pFg0ZprZrK+4iAaLkCi5llj1Pcuecj65RdBQIVF14CuY/Oh5RZE+DM4aSzlaTMe0Q4I1f5b983uZ2the6shICpAi7zxAmZ7zecjd1zzz1w++23Wwkn/XhhW5TjmXqSDyU3DIZaBfUJn0+krTDzEHmx85XWv6wzekY903jIO/yvPyDp9RcknRf8yM24F04u3YQ7iiRp3hJG9cFbfoI9n6lrcDGxsVBSUiKMAW4G+fPPP6GosAgqmLoyUAMe6pZ56w4Lg8LbfJu9iTMPnHXgDrx5O/4KVMj82i9+vYun/do4VS5BwMhnuvCuSVDynxsl7SMRVpAneB9gW7Jlad5EGNkHb/gJ9jymCrjLLrsMdu/eLYzBNddcI9ignPTAJLisW7fAHBdmNijh3WWyvpX26gc1zEaeL4FmHr6g5TqvaLkiLT4CWjeIFtbgXOemFH8iYPQznTt1HlRcIDcqEXnkVwh/6E48qOtzd43ug88MBlkBU1WU48adNXqKa3Bt27aFsrIy6Nq1a0AOQ8zObRCBZ2+4UKjCYwDONPS0h8mxFDQkrX9ZZ6gNf6bx+MAzr0KjsTdCeNYJCRBhX22DpKWzoWCcZw8EzgUN74Nz43QvQ8DUGRzPzUUXXQQ4q8MzbIEY4te/JutW2SXdoar1BbJ4TxE08/CEEKXbDQEznulaZsUke/ZKqI2RW45JWPcaVH74rk8wmtEHnxgMssyGz+AmT57s1fZ93PIfSCHy1x8hWsHAayEzqqwm0MxDDWpUxsoImPVMV7U6H3IfXyTYp+S9D5yzaAbk/qstVLbr6BV0ZvXBK+aCMJPhAg49AQRjiN+wQtbtSjZzq+h0uSyeIggBQsBYBMou/w8U3PcYJC+ZLWk4oroKGswYDSeXZ0BNSkNJmr8I2ompH7KGC7hrr71WP+5tUhOa5Yr9bLOM26JB/vX3JmuQIggBQsAlAkW3j4bvd+yBnj9tl+TBXc+p7PhA1ovrANi6nb+DuBMT2zHrbKa/+2hU/YYLuG3btkGfPn2E/qFzUlcBLZMESoj/YLXcJQ7T/Zf0Una6aFS/6UvRKKSpHbsgEJ3+Avw2fhC0yfxZwnLUT99DyvOPMkeq0rNzkkw6EbQTUycgWTWGC7gvv/yyTsDt2LHDZU8CRcCFVFYws1xvyfpZdCPbhvyP7zpZokERWr8USUAaNFDUjGEINKkfD2EvrwbHfTdCSLbU8XLctvehslU7wJmePwPtxNQPXcMF3DPPPFPH/XPPPVd3H6g3sXiw+3S+pHsOJtiKBwyTxJlBaP1S1CogzegztUkIeEIA19qqXl4HEcOukdmsTFr6LFQxIVfOdj/7K+BOTDTcgL9PFHZIU1CHgOECzhWbDnYIGv/EEChHBeLfXy12qe5acvWNUJucUkebdaP1S1GrgDSr39QuIeAJAUeHLpA3+VlIeXayJCvuskx56v4zm04a+0fw0E5MCeSaCFMFXHZ2NixevBh++uknKC0tlXRk82b5pgxJBhsQkQf2QuTvP8k4tcrmEq1filoFpAwYiiAELIRASd+BEHH4ECSsXyHhCr2BN3h8DGS98r7MUaokIxGmI2DqieqXXnpJcG2Dasto5kFgwYIF0LlzZxgzeozpwOjBAG4u4UM5+zJEFYcVgviluGJQW8GDNNK+BDrU6gtalNeOCBSMnQ5ll1wpYz2SCb7686bK4inCWgiYOoP79ddfYdWqVRATEwPs9De0adMGHnroIcDD4P3697MWUj5yE5qfC7FffiIrVXzTcFmcXSNEAWlX/olvQsAjAsyqUu6Ti6HRvTdA+Mm/Jdnj/vsBVLbtAEW3jpDEE2EdBEydweE6WxizpI8hLi4OCgoKBKen+XnSTRk8XPPmzQN0q4POTfnw7rvvQr9+/YS6+DQj6XoZayGEOXJ1DjXsaEBpj+uco+ieECAELI5AbXwis1m5HGoVzsAlLUmHqO/+Z/EeBC97pgq41q1b13kTQDuUs2fPhlmzZkHLVi3djghaQ5k5c6YsT1ZWFnz/PTuvkmLyBg5mhbzeZvnRgOIbhrCDGaZOmmWYUQQhQAh4RgDtxeZNmSvLGFJTA6kzx0HYqUxZGkWYj4CpAu6RRx4BFGwY0LPAhRdeCC1atICpLN5d6NixIyQkJMiyLFu2THCWGsLUnWaGmK+3Qzj3wDvYbLX4xjvMZIvaJgQIAQ0IlPa+CQoH3iOrAX3IpT7B9g0wZ8Z2CHh+9YHNh2Hk+l+E6/GiCjuwrYpHUwVcDfv6EQVVFNtkMmzYMBg9ejSkpKb63Jldu3ZBSv0UYR3P58I6F6j34RpZjWWXXwM1DRrJ4imCECAE7IMAus8pv+hSGcNRP/8AyYtmyuKtGCGeXw0GZ8mm6stGjBgBF1xwAfTo0QO6d+9+ZrOJiieinHkAX7duHcxOlxpKdVVVxuYM2Lptq5C8aNEi5p0+zVVWt/ERERHysieOQeTer2Tlwu+ZJM8ry6V/hCKP+jejqUY78IgdtAOfxKOmR7GusFscF68Hx63dIOTk8br8eBO/+W2IvuJqqB1gjKbGLY8SzqRESdVhSUQJ2yqg9h0oqciChKkCbuXKlYDmurZs2QLLly2HiztdDD179oTLLr0MwsLPbD7xBrMTJ07AqVOnYPz944XseXl5MHHiRFj4wkKon1JfVgXu0HTepYlrd2oCPhR82cRVL0OU04F1rLeq2blwslV7ZjlVXTtqeBPLKPEoplnlagceESs78Ek86vNUe8Ix8slXIG3SIJmlk7AnxkNOg2bsKND5+jDiphZPPLoqGhchTUGaf49JcyhTuJxk9WCqijIpKQluuukm4fzb4lcWQ8vzWsKaNWvgjjt9+wJq2bIlrF27VjhygMcO6tevD3jGTkm4+XVA2OaSuE+YxXEuFN8wmIshkhAgBOyMQGW7iyB/wkxZF0KZ7VnHwyNhwlvfWnZ9K5jOr5oq4JyfDpx15RfkQ1FRkXBkwDmNv09PT4eHH34YMjMzhXU7d14J+LL+pGP+9zmE50hnaQ62a7Kk763+bJbqJgQIARMQKO4/FEr63CJruUn+cZi49Tk4nF0u2JSUZTA5Qjy/qtbAg8ns+9S8qSrKP//8Ez777DPYuXNcmT4sAAAZRklEQVQn24BUCZdffjlMnz4d2rVr57YTmMddwFmcGSEu421Zs7i5pJadf6NACBACgYdA3uTZzJzXQYj84xdJ5/5z7Bu4++B78MllpL2RAGMwYaqAw2MCl156KYwfPx46derEjJmYu71fC/ZhbOaGMzg+FPdjZ98oEAKEQEAi4IiKhpxZy6DRmH4QWlIs6eOk/W9AXusLWVxbSTwRxiFgqoB76623oKiwCH76+Sdho4mzNwG7+YPDtbcQtgbnHKrTmkJ5lx7OUXRPCBACAYZANdtElvvofMEAs3PXwhy18MTW2XDq9h6kxXEGxsB7UwXc7t27hZ2OaY3S4O+//4ZmzZoJ17Zt24L9BNx62bAVX3+bYGNTlkARhIDOCIjOZ3ELOO6Kw40EvhrP1pmloKqurHtfKGSOUBPeXS7pd2RBDqQ+MxFOzX+L3gUSZIwhTN1kgjsm77//fsFlDh70Rtc5aF8Sd0XaKUT9sAci2Pk35+Bg6taS6293jqJ7QsBvCIiHdzMLyi27ucFvnbdIxQWjp0J5h64ybqL3fw2JqxbK4inC/wiYKuByc3KgZ6+ekl727dsXvvjiC0mc1Ym4T+SzN/T4S5ZLrD5ygcMfOZ+1wFgyw/E5MxcDGlXnQ8LqlyB63y4+mmg/I2CqgItn9iTzcvOELjZo0AAOHjwIx48fh1puLcvPGGiqPqS8DGJ3fCSro+S6QbI4iiAE/IUAOp91DjztnOav+2CycegKw9r6DSBnxkuAGhznIHgCZ6rK0NxTztF072cETBVwva/pDQd+PCB08cYbb4QZM2bApEmTAGdxdgno8y20TOqNvLZeApRe2ccuXSA+AwAB8fBu46RoaN0gWliDM7pbopo0GGwcusO2otPlcPruB2RZwpiPyNRZE4B9wcvSKMI/CJi6yWTI0LNb6FGooZeAsrIyOPfcc/3TWz/UqqSeLOnFnLUq+I7yQ/NUJSEgICAe3lVrvkkPGElNehbFwuETIfrA/0H03p1nI9ld9Pe7IfH1BXB65MOSeCL8g4CpMzi+S/jjtJNwA7axJOrbb/huAKknZZBQRBAgwKtFeToIIDjbxZAQyJn+ItSkNDwb989dwpsvM8EnN8guy0gRmhGwlIDT3BuDKwj7YA1INe3MsHLzllB5wcUGc0LNEQLmIyCqSdPiI0xTk5qPwlkOapNTIOdxth7HfEE6B3xnpDwzidbjnEHx072pKko/9cmwakM3yU1z0ezNMPipIYshIKpJLcaWqexUdLwMTt/zICStmC/hQ3CSmv4AnY+ToKI/If200L/+gK0x8qfvIfTo75L+CWffFIyvSjIRQQgQAkGFQOGd90MZOzbEBzwfl7BmMR9NtI4IkIBTCWbc9g9kJSsu7gY1qeqcp8oqowhCgBAIDATYelzuDOX1ONxwEnlgb2D004K9IAGnZlDYNt/YzzfLSpZcc5MsjiIIAUKAEKhNqs82nSyUn49j75LUpydAaHEhgeQHBEjAqQA1et9OCMvLkZR0sGMBpT2uk8QRQQgQAoSAiACej0N1JR/CT2VC/Tl0bIDHRQ/alptM5s2bB/v374fExERYunSpgMOyZctg7969EM4cjOJxg8mTJ0N8fLweGMnqiPvv+7K4ssv/A456/mlP1hhFEAKEgC0RwA0n0d99A1GcWjJ25zao9/5qKL55uC37ZVWmbTmD6927N8ycOVOCaaeLO8HSJUthyZIl0KRJE1i7dq0kXS8ipKIcYr7aKquO1JMySCiCECAEeATYkYGcxxdBbXwinwLJS56BiN9/ksVThHoEbCng0OJJArNj6Ry6dO0CYeFn7PGhR/Dc3FznZN3uY9iXlsw0F3tYyy7tpVsbVBEhQAgELgI1DRtD7tR5sg6GVFZC6szxgB/RFPRBwJYCzlPXt23bBp07d/aUTVV67KcfysqVXnU9QARzwkWBECAECAEvEChjtmqLFNSREX/9AcmLnvKiBsriDQK2XINz1zH0MRfG3FagGtNVyNicAVu3nVEzLlq0SFizc5VXEl9UCJH/96UkConI2+7xvg5Zaf9GRDDBi2uSVg524BHxswOfank8mlcKMzYdhNNllZAYEwnpA9pDi+RYvzw2ann0CzMuKjWEx5lMVXnoWwj95YzBeZGVehlrIarPAKjtPUCMUrwawqNiy/aJDCgBt2XLFmGjyZw5cyCEnT1xFfr17wf4J4asrCzx1u01bssGSKmqkuSpZrbmTjZvA+BlHZLCBhBmGt/1tnt24BH7Ygc+1fL46ObDgqNU7Cc6TZ363rewsH9rJHUPannUnRE3FRrFYzg7OtDo3hsglFNLhk4fA6can+vWp6RRPLqCqUWLFq6SLBMfMCrKPbv3wMaNG4XNJ9HR0X4BOPbzDFm9pT1vIFf0MlQowm4IkCcAc0asukUryJ/wpKzxsKLTkMJMeYHDIUujCO8RsOUMLj09HQ4dOgTFxcUwbNgwGDJkCGzYsAGqq6th2rRpQu/btGkDDz74oPdIeMgZyh44PP/Gh9Kr+/NRXtHoHBL9Z+GLBa2uo6FatOVHgRAwAwF8BtGPmxiC2hOACIJB15J+QyBmzxcQ++UWSYvR3/0PEtYuhcKhYyXxWolgevfYUsBNnz5dNsbXX882evgxxHzxCYQwAeocHE2aQ2X7Ts5RXt+LziGxAL5Y5u34y28qIa+ZooxBiwB+YOEz6PzBFbRgmNDxvClzIZKtx4XnSJdLElfOh/JOV0Dl+R104yqY3j22FHC6jbQPFcXtkKsna/re4kMN0qykEpLiQZS5CJAnAHPxx3NxuWw9ruFDQyHESS2JH9UpzJTXydc+AUeMPpt+gundEzBrcP58PEML8iCKWf7mQ+11g/gor2leBcTTXldEGQkBQiAgEEBj7YVD7pP1JeL4UUh+8QlZvNoI/l3D02rrtWI5EnBejErsl0w9yYyiOofqxs3BcaH6s3bkHNIZTbonBAgBROD0iMlQoaCOrMd2cMfu+FgXkILp3UMqSi8emdgdH8lylfTqB1r2apJKSAYpRRAChACzpXvwgfnQbkJ/iKmSWjSpP/8xKGezvNrEZE04BdO7h2ZwHh6VkNISiPpxnyyXcDxAFksRhAAhQAhoQ+CZ30JgdlepqrI2Ng7yxz+uWbhp48x+pWkG52HMHOzBOr7x/wBtUOJMDo8KVKQ0gvG/REPJj7sgjlnooi3+HkCkZEKAEPAaAdwE8mGra6D733uhz9GdcCjtfEh6YTlUN7H+wWqvO2lQRhJwXgBdWy8BSq69VfgLYY4JX3xnV53VByxOW/y9AJGyEAKEgFcI4KYPPDo0q9v9cDC1DezqMQQWkHDzCjs+E6koeUQ80A4m7L5PbCnJxW+7lSQSQQgQAoSADwiIm0BiUpLhq553wOSrz/GhNGV1RoBmcM5oeHkvfmGJ2QN5m63YR7oSAoSAMQgE0yYQfyNKMzgVCItfWI2ToqF1g2hhDU5FNVSEECAECAFCwI8I0AxOBbjiF5bZ1rxVsE5FCAFCgBAIGgRoBhc0Q00dJQQIAUIguBAgARdc4029JQQIAUIgaBAgFWXQDHVgdjSYXH8E5ghSrwgB/yFAMzj/YUs1G4CA6PoDzw0dzi4XziQa0Cw1QQgQAjZAwJYzuHnz5sH+/fshMTERli5dKsBcWFgIs2fPhuzsbGjQoIHg+DQhIcEGQ0AsakGAP4PI01rqprKEACFgbwRsOYPr3bs3zJw5U4L8O++8Axd1uAhWrFghXJGmEPgI8GcQeTrwEaAeEgKEgCsEbCngOnbsCPzsbM+ePdCnT2+hn3hFmkLgIyCeSUyLj6AziYE/3NRDQsAnBGypolTq4enTpyElNVVIwiuqLCkEPgLimcTA7yn1kBAgBHxFIGAEnC8dz9icAVu3bRWKLFq0CPDAtpoQERGhuqya9tSUIR7VoKZchrBUxsXXWMLRV8SU89sBR2XOjYsNGAGHG05yc3KEWRxeeRWmM6T9+vcD/BNDVlaWeOvT1Q6WTIhHn4bUbWbC0i08XicSjl5D5Taj2Ti2aGF99z22XINTGvUuXbrAtm3/FZLw2rVrV6VsFEcIEAKEACEQJAjYcgaXnp4Ohw4dguLiYhg2bBgMGTIEBg8eLBwT2P7pdkhla3DTp08PkiGkbhIChAAhQAgoIWBLAedKeM2dO1epjxRHCBAChAAhEIQIBIyKMgjHjrpMCBAChAAh4AYBEnBuwKEkQoAQIAQIAfsiQALOvmNHnBMChAAhQAi4QYAEnBtwKIkQIAQIAULAvgiQgLPv2BHnhAAhQAgQAm4QIAHnBhxKIgQIAUKAELAvAiTg7Dt2xDkhQAgQAoSAGwRIwLkBh5IIAUKAECAE7IsACTj7jh1xTggQAoQAIeAGAVtaMnHTH0qyEQJ/n66A57/8C0qqDkNcBAD6dkP3NxQIAUKAENADAZrB6YEi1aEKARRuh7PLIbOgXLjO2/GXqnqoECFACBACSgiQgFNCheIMQaCovEbSDk9LEokgBAgBQsBHBEjA+QgYZdcPgfjoMEllPC1JJIIQIAQIAR8RIAHnI2CUXT8EcM2tdYNoaJwULVyRpkAIEAKEgF4I0CYTvZCkenxGADeULOzfGsz2TOwz41SAECAEbIEAzeBsMUzEJCFACBAChICvCATcDG79+vWwfft2CAkJgebNm8PDDz8MUVG09dzXB4PyEwKEACFgdwQCagaXnZ0NH330Ebz00kuwdOlSqK2thc8//9zuY0T8EwKEACFACKhAIKAEHPYfhVpFRQXUVNdAZWUlpNRPUQELFSEECAFCgBCwOwIhDhbs3gln/t977z14++23ISIiAi666CJ47LHHnJOF+4zNGbB121bhftGiRYJAlGXyIgLbqKqq8iKneVmIR/2wJyz1wZJwDAwc7bD0E1ACrrCwEJ555hmYNm0axMXFCfdXXHEF9OnTx+0TdezYMbfprhLtsPuPeHQ1er7HE5a+Y6ZUgnBUQsX3OLNxbNGihe9MG1wioATcjs93wN59e4WNJYjj1q1b4eeff4ZJkyYZDCs1RwgQAoQAIWA2AgG1BtcwrSH89ttvUF5eDqh5/e6774SdlP4CecKECf6qWrd6iUfdoATCUh8sCcfgwVGfnqqvJaCOCVxwwQXQrVs34UUUFhYG5557LvTr1089OlSSECAECAFCwLYIBJSAw1G4++67hT/bjggxTggQAoQAIaALAmEzWdClpiCspLamFv7V9l+W7jnxqN/wEJb6YEk4Bg+O+vRUfS0BtclEPQxUkhAgBAgBQiDQEAioTSaBNjjUH0KAECAECAH1CATcGpx6KFyXxHW96OhoCA0NBdy8gofDnQPu2Fz88mLY/+1+we7lAw88AG3btnXO4td7PMc3Z86cujayT52C2wcPhltvvbUu7ttvv4XZ6enQoGFDIe6yyy6D4cOH16X762bevHmwf/9+SExMFMynYTt4XnH27NmAptUaNGggnFtMSEiQsbBlyxZA26IYBg0aBNdee60sjx4RSjwuW7YM9u7dC+Hh4YK3g8mTJ0N8fLysOU/PhqyAygglHletWiXYXRWxGzZsmLDJim9iz+49sPzV5YKVn97X9IYhQ4fwWXShlXjEc6knTpwQ6i8tLYXY2Fh45ZVXZO0ZhWNWVhbMnz8fCgoKBHu1eEZ24MCBlnomXfFotWdSNohWjEBLJhTcI3DXXXc58vPzXWb6+uuvHexwuYOZCXP8+OOPjokTJ7rM6++Empoax5AhQxyZmZmSppiQccyYMUMSZwTBBKuDnUV0jBkzpq459kN1vP3W2wKNV6T5cPr0aQd76Tnw6nzP59ODVuKRCQVHdVW1UD3yp8QjJnp6NvTgD+tQ4vH11193vPPOO26bwOfhnnvucRz/+7iDma5zjB071nHkyBG3ZdQmKvHoXNeSJUscb7zxhnNU3b1ROOZkZwvPIzZcUlziGDlypICHlZ5JVzxa7ZmsGzwL35CKUoevjv/973/Qq1cv4Yuwffv2UFJSArk5OTrU7HsV+/bug0aNGgl/vpfWv0THjh1BnGGIte/Zs4dZl+ktkHhFmg8Y16FDB6Eslsd7pXx8OTW0Eo9dunaBsPAzHsfbtWsHubm5aqrWrYwSj95U/tNPPwnPQpOmTQTzdVdecSV8vetrb4r6nMcdj+wdCOxDUPid+FyxjgVSUlPrtCuxcbHQtGlTyGG/VSs9k654tNozqeOw+K0qUlF6CS2b/Qg5r+17LfTrLz1bl5eXBw3/Uf1hppSUFMhmPxp8UI0OO77YAd27d1ds9tdff4Vx48ZBcnIyjBo1Cs477zzFfP6OZDOyOmwQI1RZ8gEFSqoTfoipWUJm27ZtLjFFvt09G3y/9KY/+eQT2LFjB7Rs2RLYLFn2MYEvb8RODKkNUgXrPiJt1PX7778X1NTowspVMBpHpuUANpuFdue3A6s+k848OuNm5WfSmU+z70nAeTECuLaAa0VMTQnTp0+HZs2bAX6tigG/TvmA/uiMDmj4ed++fTByxEhZ0/9q8y/ANZuYmBj45ptvBDudK1askOWzcoQZmK5Zs0ZYd+3d+8yMk8fH07PB59eTvrH/jXXrqExdCUzNBlOmTJE04aiVP5uhIcYrblAIu/rwQoaNxrGsrAzS2Zo0fujF1YuTYOYL4c9n0hWPVn4mfcHOiLzGP+lG9ErnNlC4YcCZT9euXeGXX36RtIBfyKfYxg4xCLMPp69mMd7f1927dwuzsvop9WVN4Y8YhRsGtPbC1maEhXZZRgMicMOJqMLFK6/CRBYQU5x9iAExrV9f3i8x3R9X3OSCG00effRRQf2s1IanZ0OpjF5xOM648Qn/rr/+evj9999lVTdo2EAy883JzjEcR3RdhSrAnj17yvgTI4zEET8En376abjqqqugR48eAgtWeyaVeERGrf5MiuNplSsJOA8jgV9RpSWlQi68R1XLeedKVXu4IxEdq+JM7uDBg8JOMTPUk1988UXdD5bvVl5unsAfxuO6DPrNwx+1GaFLly6wbdt/habxih8NfMA4xBrVl/iH90r5+HJ60bjzcOPGjYB2EHAHrVLw5tlQKqdXnPiRgPXt3LlT0e7q+W3PB1Rz4R++NHfu2gndLu+mFwte1YNahSZNmkjU+M4FjcQRf6MLFiyAZs2aCTtzRT6s9Ey64tEOz6SIp1WudNDbw0icOH4Cnkl/RsiFsx5Us9x5553w4YcfCnEDBgwQBAd6EceXcGRkJOAxgfPPP99Dzfomo4FpthMNVq5YWadycebx/fffF77+8JgD8jhq5Cj494X/1pcJhdpQDXTo0CEoLi4WZmpshyegCyM8JoAzNFxnQ7UvzuLQ8wN6ZMct+Rg+/vhj2LBhg3CPxwSuu+464V7vf0o8YrvV1dVQr149obk2bdrAgw8+KBxtePHFFwUVr6tnQ2/+sD4lHn/44Qf4888/hdklzoAmTZworG3i8QuRRyyLKmlUR+NHzdVXXy08vxivd1DiEWeWzz33nLCxA38rYnDm0UgcETN0p4VrgaJ6EY9XoB1bqzyTrnh89dVXLfVMimNp5SsJOCuPDvFGCBAChAAhoBoBUlGqho4KEgKEACFACFgZARJwVh4d4o0QIAQIAUJANQIk4FRDRwUJAUKAECAErIwACTgrjw7xRggQAoQAIaAaARJwqqGjgoQAIUAIEAJWRoAEnJVHh3gLKATQYj4eHKdACBACxiBAAs4YnKkVQoAQIAQIAYMRIAFnMODUHCHgTwTQLBYFQoAQOIMACTh6EggBAxFAe5HMJxsMYs5o0RloRUWF0HrG5gwYMWIE3HbbbfDEE08IFlMwAU1s9evXD5wFFxpURosvGNA24UMPPSQ4EcWyb6x+Q4inf4QAIQBAAo6eAkLAQAR27doFs2bNghUrVwJ6YkcBhR7P17y1BqZOnQpoKb5hg4Ywd+5cr7k6fPiw4PNt7dtr4Y477vC6HGUkBAIdAXKXE+gjTP2zFAI4GxMt519yySXwxx9/wG+//SbYiGzbtq3A68hRI2Hw4MHC7M0b5pOSkuCWW24RsopOWr0pR3kIgUBHgGZwgT7C1D9LIeDs8geNXqORbPQzKAo9ZBbdGsXFxdWpKT11wNkxrKe8lE4IBBMCJOCCabSpr5ZEAP0MonV9MaD7mJKSEkHoiT78UBCKoaCgQLylKyFACLhBgAScG3AoiRAwAoFevXrBZ599JqgqKysrBZdHrVq1gsaNGwOqH1EA/nf7fwV3N+hCyNm5rhH8URuEgF0RIAFn15EjvgMGgc6dOwP6yXv22WcFX20ns04KXsTFDo4fPx4++OADuP3224WNKeibjgIhQAh4RoD8wXnGiHIQAoQAIUAI2BABmsHZcNCIZUKAECAECAHPCJCA84wR5SAECAFCgBCwIQIk4Gw4aMQyIUAIEAKEgGcESMB5xohyEAKEACFACNgQARJwNhw0YpkQIAQIAULAMwIk4DxjRDkIAUKAECAEbIgACTgbDhqxTAgQAoQAIeAZARJwnjGiHIQAIUAIEAI2RIAEnA0HjVgmBAgBQoAQ8IwACTjPGFEOQoAQIAQIARsiQALOhoNGLBMChAAhQAh4RoAEnGeMKAchQAgQAoSADREgAWfDQSOWCQFCgBAgBDwjQALOM0aUgxAgBAgBQsCGCPw/JLIk4hPEddQAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 88,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Display image with trend line\n",
    "Image(\"image_2.png\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Again, the graph above for bike availability on a weekend is non-linear. We can see mostly on weekends there is high bike availability, with slight drop offs early in the morning and late in the evening/night. I would hazard a guess that the largest drop off in availability in the evening is due to people leaving bars and clubs and looking for a quick and cheap means to get back home."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Predictor Model Design"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It is now time to design and develop the predictor model to be used for the creation of our pickle model files. As the data is of a non-linear type, a decision tree regressor was chosen due to its ability to fit polynomial shapes well. Within this regression model, parameter tuning would be controlled by using grid search and time series cross validation. Grid search is an approach to hyperparameter tuning. It methodically builds and evaluates a model for each combination of algorithm parameters specified in a grid. For our model, we are going to specify a max depth range of 1 -> 11, our performance metric will be the r2 score, and our validation will be assessed used time series split with cross validation.\n",
    "\n",
    "Each model trained, which is outputted as a pickle file, will be unique to a given bike station on a given day of the week. This is for the following reasons:\n",
    "\n",
    "* There will be a large variance of bike usage depending on the day of the week (working days versus weekend)\n",
    "* There will be a significant difference in bike turnover depending on if the station is located in the city centre, or on the fringes of the city.\n",
    "\n",
    "For model training, the key inputs for the model (once seperated out by day of the week and station number) are as follows:\n",
    "\n",
    "* Hour of the day \n",
    "* Main weather conditions (broken down as binary categorical values)\n",
    "* Main temperature degrees C\n",
    "* Rainfall in mm\n",
    "\n",
    "The target value is:\n",
    "\n",
    "* Bikes available\n",
    "\n",
    "In total there will 791 different pickle models generated. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Scoring function for models\n",
    "def performance_metric(y_true, y_predict):\n",
    "    \"\"\" Calculates and returns the performance score between \n",
    "        true and predicted values based on the metric chosen. \"\"\"\n",
    "       \n",
    "    score = r2_score(y_true, y_predict)\n",
    "    # Return the score\n",
    "    return score"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "*fit_model* is where the model fitment occurs. All the data is broken into splits of 20. As previously mentioned, each day will be 20 hours in length in our dataset, so this is how we decide our splits. The regressor will be a decision tree regression model, and it will be allowed to train on a max depth ranging from 1 -> 11. We will score the model using the r2 value, and gridsearch will be used in conjunction with cross validation to generate the model with the best hyperparameters. \n",
    "\n",
    "To automate the process in future, a *models_score_dict* text file has been created. This keeps the score of each model in a text file. Our deployment of the predictor_model.py file on EC2 works as follows:\n",
    "\n",
    "* Crontab is scheduled to run the predictor_model once a week at 3am on a Monday. This is so that we have new day of data for each day of the week.\n",
    "* The model will run and generate a best score for the fitted model. It will then look up the previous best score for this model in the dictionary text file. If the new best score is worse than the previous best score, it will move on to the next model. If the new best score is better, it will overwrite the previous model, and update the dictionary with the new best score.\n",
    "\n",
    "In this way we have automated the model training process and ensured the best model is the one that is stored on EC2.\n",
    "\n",
    "One important point to note regarding our models is there are some prediction discrepancies, as visible in some of the recorded best scores. This is happening for several reasons:\n",
    "\n",
    "* Decision tree regressors can be prone to overfitment. \n",
    "* Overfitment is happening mostly due to our small data set size. While we have a lot of rows of data, when you consider how many actual days this is (for our time series split samples) we in fact only have 8-9 days worth of data for each day of the week. This is small, and will cause overfitment.\n",
    "* We do not currently have data that covers a wide spectrum of the weather. This means that as it gets warmer, the model is seeing weather it has not got much learning on, and is failing to generalise well.\n",
    "\n",
    "All these issues should gradually reduce as the model continues to train on new data each week over the course of the year. This will help it generalise a lot better."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "def fit_model(X,y):\n",
    "    \n",
    "    n_splits = int(len(X)/20)\n",
    "    \n",
    "    cv_sets = TimeSeriesSplit(n_splits)\n",
    "\n",
    "    regressor = DecisionTreeRegressor()\n",
    "    \n",
    "    params = {'max_depth': range(1,11)}\n",
    "    \n",
    "    scoring_fnc = make_scorer(performance_metric)\n",
    "    \n",
    "    grid = GridSearchCV(regressor, params, scoring = scoring_fnc, cv = cv_sets)\n",
    "    \n",
    "    grid = grid.fit(X, y)\n",
    "    \n",
    "    if key_name not in models_score_dict or models_score_dict[key_name] < best_score:\n",
    "        models_score_dict[key_name] = best_score\n",
    "        f = open('/home/ubuntu/oui-team/pickle_files/models_score_dict.txt', \"w\")\n",
    "        f.write(str(models_score_dict))\n",
    "        f.close()\n",
    "        return grid.best_estimator_\n",
    "\n",
    "    return grid.best_estimator_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A key point to note in the below function, *station_day_model*, is that is only uses whole days for its training. It will remove every row of data from the dataframe that does not finish on midnight for that day."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def station_day_model(station_number, day_week, df_current, features):\n",
    "    \"\"\" Function that creates a data frame for the given week day and station number. Ensures Whole days\n",
    "    only are included in the data frame. Creates the file name for the pickle model. Checks the previous\n",
    "    model best score versus the current model best score. If higher, updates the pickel model.\n",
    "\n",
    "    :param station_number: current station number\n",
    "    :param day_week: current day of the week\n",
    "    :param df_current: data frame for model work\n",
    "    :param features: list of features to be used for model\n",
    "    :return: Void. Updates pickle models.\n",
    "    \"\"\"\n",
    "\n",
    "    df_current = df_current[df_current['weekday'] == day_week]\n",
    "    df_current = df_current[df_current['number'] == station_number]\n",
    "\n",
    "    break_case = datetime.time(23, 59, 0)\n",
    "    df_temp = df_current\n",
    "\n",
    "    for index, row in df_current[::-1].iterrows():\n",
    "        if index.time() == break_case:\n",
    "            break\n",
    "        else:\n",
    "            df_temp = df_temp.drop(index)\n",
    "\n",
    "    df_current = df_temp\n",
    "    X = df_current[features]\n",
    "    y = df_current.available_bikes\n",
    "\n",
    "    file_name = \"model_\" + str(station_number) + \"_\" + day_week + \".pkl\"\n",
    "    key_name = file_name[:-4]\n",
    "\n",
    "    f = open('/home/ubuntu/oui-team/pickle_files/models_score_dict.txt', \"r\")\n",
    "    models_score_dict = f.read()\n",
    "    models_score_dict = ast.literal_eval(models_score_dict)\n",
    "    f.close()\n",
    "\n",
    "    if key_name in models_score_dict:\n",
    "        previous_best_score = models_score_dict[key_name]\n",
    "    else:\n",
    "        previous_best_score = -9999999\n",
    "\n",
    "    reg = fit_model(X, y, models_score_dict, key_name)\n",
    "\n",
    "    f = open('/home/ubuntu/oui-team/pickle_files/models_score_dict.txt', \"r\")\n",
    "    models_score_dict = f.read()\n",
    "    models_score_dict = ast.literal_eval(models_score_dict)\n",
    "    f.close()\n",
    "\n",
    "    current_best_score = models_score_dict[key_name]\n",
    "\n",
    "    if previous_best_score < current_best_score:\n",
    "        with open('/home/ubuntu/oui-team/pickle_files/' + file_name, 'wb') as handle:\n",
    "            pickle.dump(reg, handle, pickle.HIGHEST_PROTOCOL)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_models(station_array,df_db, features):\n",
    "    \"\"\"\n",
    "    Loops through each station number and day of the week for model creation.\n",
    "    :param station_array: list of all valid station numbers\n",
    "    :param df_db: dataframe of all database data to be manipulated\n",
    "    :param features: features to be used for model training\n",
    "    :return: Void. Updates pickle models.\n",
    "    \"\"\"\n",
    "    days =[\"Monday\", \"Tuesday\", \"Wednesday\", \"Thursday\", \"Friday\", \"Saturday\", \"Sunday\"]\n",
    "    for station_number in station_array:\n",
    "        for day_week in days:\n",
    "            station_day_model(station_number, day_week, df_db,features)\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Machine learning models only work well with numerical inputs. In the below function, *start_modelling*, all categorical features are converted into binary values for the model.\n",
    "We have also future proofed the model by including features that have not yet occurred in the database, so that when they do occur, the model will be able to continue to train without issue."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def start_modelling():\n",
    "    \"\"\"\n",
    "    Initial function. First calls SQL query to generate the dataframe from the databases. Creates a list\n",
    "    with each unique station number. Converts all the categorical weather features to binary features for\n",
    "    the purpose of modelling. Creates a df filtered by the desired features to be used for modelling.\n",
    "    Makes a directory for the pickle files if it does not exist. Creates the mode best score dictionary if it does\n",
    "    not exist. Calls the function to start the modelling process.\n",
    "    :return:\n",
    "    \"\"\"\n",
    "\n",
    "    # Download latest sql info\n",
    "    df_db = sql_query_model()\n",
    "\n",
    "    # Creates an array with every unique station number\n",
    "    station_array = sorted(df_db.number.unique())\n",
    "\n",
    "    # Make categorical info into binary\n",
    "    main_weather_dummies = pd.get_dummies(df_db['main_weather'], prefix=\"main_weather\")\n",
    "\n",
    "    additional_main_weathers = ['main_weather_Thunderstorm', 'main_weather_Haze', 'main_weather_Squall',\n",
    "                                'main_weather_Smoke', 'main_weather_Dust', 'main_weather_Tornado',\n",
    "                                'main_weather_Ash']\n",
    "\n",
    "    # This is ensuring that the order in which the model was originally trained is preserved\n",
    "    # This is to allow for the reordering of data frame columns in the for loop below as new weather descriptions are\n",
    "    # added to the database history\n",
    "    categ_features = ['main_weather_Clouds', 'main_weather_Clear', 'main_weather_Rain', 'main_weather_Fog',\n",
    "                      'main_weather_Drizzle', 'main_weather_Mist', 'main_weather_Snow'] + additional_main_weathers\n",
    "    cont_features = ['hour', 'rainfall_mm', 'main_temp']\n",
    "\n",
    "    features = cont_features + categ_features\n",
    "    df_db = pd.concat([df_db, main_weather_dummies], axis=1)\n",
    "    df_db = df_db.drop('main_weather', axis=1)\n",
    "\n",
    "    for weather_descriptor in additional_main_weathers:\n",
    "        if (weather_descriptor) not in df_db.columns:\n",
    "            df_db[weather_descriptor] = 0\n",
    "\n",
    "    # define the name of the directory to be created\n",
    "    path_pickle = \"pickle_files\"\n",
    "\n",
    "    if not os.path.exists(path_pickle):\n",
    "        try:\n",
    "            os.mkdir(path_pickle)\n",
    "        except OSError:\n",
    "            print (\"Creation of the directory %s failed\" % path_pickle)\n",
    "        else:\n",
    "            print (\"Successfully created the directory %s \" % path_pickle)\n",
    "\n",
    "    path_pickle_file = '/home/ubuntu/oui-team/pickle_files/models_score_dict.txt'\n",
    "\n",
    "    if not os.path.isfile(path_pickle_file):\n",
    "        dict = {}\n",
    "        f = open(path_pickle_file, \"w\")\n",
    "        f.write(str(dict))\n",
    "        f.close()\n",
    "\n",
    "    # Make all the models\n",
    "    create_models(station_array, df_db, features)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "start_modelling()"
   ]
  }
 ],
 "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": "3.7.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}