My model Final - Raw temperature.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## My model Final\n",
    "### Vladimir Fyodorov\n",
    "29.05.2022"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "from datetime import datetime, timedelta\n",
    "import datetime as dt\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "import matplotlib.pyplot as plt\n",
    "from keras.models import Sequential, load_model\n",
    "from keras.layers import Dense, LSTM, SimpleRNN, Dropout\n",
    "#import tensorflow as tf\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### About CDD/HDD\n",
    "https://www.cmegroup.com/education/articles-and-reports/weather-options-overview.html"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Loading Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "https://www.youtube.com/watch?v=JWqLS51a2MM\n",
    "\n",
    "https://www.ncdc.noaa.gov/cdo-web/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Date</th>\n",
       "      <th>Raw Temperature</th>\n",
       "      <th>Rolling 5</th>\n",
       "      <th>IsSacramento</th>\n",
       "      <th>IsMinneapolis</th>\n",
       "      <th>IsLas Vegas</th>\n",
       "      <th>IsNew York</th>\n",
       "      <th>IsAmsterdam</th>\n",
       "      <th>IsCincinnati</th>\n",
       "      <th>IsDallas</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2015-01-01</td>\n",
       "      <td>4.233333</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2015-01-02</td>\n",
       "      <td>4.433333</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2015-01-03</td>\n",
       "      <td>5.600000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2015-01-04</td>\n",
       "      <td>7.133333</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2015-01-05</td>\n",
       "      <td>7.316667</td>\n",
       "      <td>5.743333</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7667</th>\n",
       "      <td>2017-12-27</td>\n",
       "      <td>2.775000</td>\n",
       "      <td>4.400000</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7668</th>\n",
       "      <td>2017-12-28</td>\n",
       "      <td>3.750000</td>\n",
       "      <td>3.980000</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7669</th>\n",
       "      <td>2017-12-29</td>\n",
       "      <td>5.825000</td>\n",
       "      <td>4.200000</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7670</th>\n",
       "      <td>2017-12-30</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>4.445000</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7671</th>\n",
       "      <td>2017-12-31</td>\n",
       "      <td>-1.625000</td>\n",
       "      <td>3.145000</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>7672 rows × 10 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            Date  Raw Temperature  Rolling 5  IsSacramento  IsMinneapolis  \\\n",
       "0     2015-01-01         4.233333        NaN             1              0   \n",
       "1     2015-01-02         4.433333        NaN             1              0   \n",
       "2     2015-01-03         5.600000        NaN             1              0   \n",
       "3     2015-01-04         7.133333        NaN             1              0   \n",
       "4     2015-01-05         7.316667   5.743333             1              0   \n",
       "...          ...              ...        ...           ...            ...   \n",
       "7667  2017-12-27         2.775000   4.400000             0              0   \n",
       "7668  2017-12-28         3.750000   3.980000             0              0   \n",
       "7669  2017-12-29         5.825000   4.200000             0              0   \n",
       "7670  2017-12-30         5.000000   4.445000             0              0   \n",
       "7671  2017-12-31        -1.625000   3.145000             0              0   \n",
       "\n",
       "      IsLas Vegas  IsNew York  IsAmsterdam  IsCincinnati  IsDallas  \n",
       "0               0           0            0             0         0  \n",
       "1               0           0            0             0         0  \n",
       "2               0           0            0             0         0  \n",
       "3               0           0            0             0         0  \n",
       "4               0           0            0             0         0  \n",
       "...           ...         ...          ...           ...       ...  \n",
       "7667            0           0            0             0         1  \n",
       "7668            0           0            0             0         1  \n",
       "7669            0           0            0             0         1  \n",
       "7670            0           0            0             0         1  \n",
       "7671            0           0            0             0         1  \n",
       "\n",
       "[7672 rows x 10 columns]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "weather_data_path = 'Weather Data/'\n",
    "\n",
    "# making list of cities\n",
    "list_of_cities = ['Sacramento', 'Minneapolis', 'Las Vegas'\n",
    "                  , 'New York', 'Amsterdam', 'Cincinnati', 'Dallas']\n",
    "\n",
    "# making empty weather dataset\n",
    "df_weather = pd.read_csv(weather_data_path + 'Las Vegas 2015-2017.csv')\n",
    "df_weather['City'] = 'Las Vegas'\n",
    "df_weather = df_weather[0:0]\n",
    "\n",
    "# filling weather dataset\n",
    "for city in list_of_cities:\n",
    "    df_weather_of_city = pd.read_csv(weather_data_path + city + ' 2015-2017.csv')\n",
    "    df_weather_of_city['City'] = city\n",
    "    df_weather = df_weather.append(df_weather_of_city)\n",
    "\n",
    "# modifing dataset\n",
    "df_weather['Date'] = pd.DatetimeIndex(df_weather['DATE']).date\n",
    "df_weather = df_weather.reset_index()\n",
    "df_weather['Temperature'] = (df_weather['TMAX'] + df_weather['TMIN'])/2\n",
    "\n",
    "df_weather = df_weather[['Date', 'City', 'Temperature']]\n",
    "\n",
    "# Calculating final temperature\n",
    "df_weather_final = df_weather[0:0]\n",
    "for city in list_of_cities:\n",
    "    df_city = df_weather[df_weather['City'] == city]\n",
    "    df_city = df_city.groupby(by=['Date']).mean()\n",
    "    df_city['City'] = city\n",
    "    df_weather_final = df_weather_final.append(df_city)\n",
    "\n",
    "# making final version\n",
    "df_weather_final = df_weather_final[['City', 'Temperature']]\n",
    "df_weather_final.reset_index(level=0, inplace=True)\n",
    "df_weather_final = df_weather_final.rename(columns={'index':'Date'})\n",
    "df_weather_final['Temperature'] = df_weather_final['Temperature'].ffill()\n",
    "df_weather_final = df_weather_final[['City', 'Date', 'Temperature']]\n",
    "\n",
    "# making sliding windows\n",
    "df_weather_final_2 = pd.DataFrame({'City': [], 'Date': [], 'Temperature': [], 'Rolling 5': []})\n",
    "\n",
    "for city in list_of_cities:\n",
    "    df_city = df_weather_final[df_weather_final['City'] == city]\n",
    "    df_city['Rolling 5'] = df_city['Temperature'].rolling(5).mean()\n",
    "    df_weather_final_2 = df_weather_final_2.append(df_city)\n",
    "\n",
    "df_weather_final = df_weather_final_2[['City', 'Date', 'Temperature', 'Rolling 5']]\n",
    "df_weather_final.rename(columns={'Temperature': 'Raw Temperature'}, inplace = True)\n",
    "\n",
    "for i in range(len(list_of_cities)):\n",
    "    df_weather_final['Is{}'.format(list_of_cities[i])] = np.where(df_weather_final['City'] == list_of_cities[i], 1, 0)\n",
    "\n",
    "df_weather_final = df_weather_final[['Date', 'Raw Temperature', 'Rolling 5', 'IsSacramento', 'IsMinneapolis'\n",
    "                                     , 'IsLas Vegas','IsNew York', 'IsAmsterdam', 'IsCincinnati', 'IsDallas']]\n",
    "df_weather_final"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_plot = df_weather_final[df_weather_final['IsNew York'] == 1]\n",
    "#--------------------- Visualizing the RNN model results--------------------#\n",
    "plt.figure(figsize=(15, 10))\n",
    "plt.plot(df_plot['Date']\n",
    "         , df_plot['Raw Temperature']\n",
    "         , color = '#4782B4'\n",
    "         , label = 'Temperature in {}'.format(list_of_cities[i]))\n",
    "plt.xlabel(\"Date\")\n",
    "plt.ylabel('Temperature')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### HDD/CDD Options data\n",
    "HDD = MAX(0, 18°C - Daily Average Temperature), where DAT = (MIN(T_t) + MAX(T_t))/2\n",
    "\n",
    "For example, if the average of a day's maximum and minimum temperature on a midnight-to-midnight basis is 11°C, that day's HDD is 7"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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>City</th>\n",
       "      <th>Date</th>\n",
       "      <th>Return</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Sacramento</td>\n",
       "      <td>2015-11-02</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Sacramento</td>\n",
       "      <td>2015-11-03</td>\n",
       "      <td>-0.004268</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Sacramento</td>\n",
       "      <td>2015-11-04</td>\n",
       "      <td>-0.009484</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Sacramento</td>\n",
       "      <td>2015-11-05</td>\n",
       "      <td>0.001749</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Sacramento</td>\n",
       "      <td>2015-11-06</td>\n",
       "      <td>-0.013142</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         City        Date    Return\n",
       "0  Sacramento  2015-11-02  0.000000\n",
       "1  Sacramento  2015-11-03 -0.004268\n",
       "2  Sacramento  2015-11-04 -0.009484\n",
       "3  Sacramento  2015-11-05  0.001749\n",
       "4  Sacramento  2015-11-06 -0.013142"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "financial_data_path = 'Financial Data/'\n",
    "\n",
    "df_price_final = pd.read_excel(financial_data_path + 'HDD_Returns.xlsx', sheet_name = 'Sheet1')\n",
    "df_price_final['Date'] = pd.DatetimeIndex(df_price_final['Date']).date\n",
    "df_price_final.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Date</th>\n",
       "      <th>Return</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>City</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Amsterdam</th>\n",
       "      <td>375</td>\n",
       "      <td>375</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Cincinnati</th>\n",
       "      <td>374</td>\n",
       "      <td>374</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Dallas</th>\n",
       "      <td>384</td>\n",
       "      <td>384</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Las Vegas</th>\n",
       "      <td>414</td>\n",
       "      <td>414</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Minneapolis</th>\n",
       "      <td>449</td>\n",
       "      <td>449</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>New York</th>\n",
       "      <td>384</td>\n",
       "      <td>384</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Sacramento</th>\n",
       "      <td>325</td>\n",
       "      <td>325</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             Date  Return\n",
       "City                     \n",
       "Amsterdam     375     375\n",
       "Cincinnati    374     374\n",
       "Dallas        384     384\n",
       "Las Vegas     414     414\n",
       "Minneapolis   449     449\n",
       "New York      384     384\n",
       "Sacramento    325     325"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_price_final.groupby(by='City').count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA5sAAAGpCAYAAAADRiKfAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/MnkTPAAAACXBIWXMAAAsTAAALEwEAmpwYAADzS0lEQVR4nOz9eZQk11nmjz+x5F5774ukbkmtfWlrsSUvyAs2NrbwgFnMwIC+wBiDzcB8hx3Od2bOADPzG8+weDwYMMZgZGywYWy84n2RZFlqbdbWq7rVS+1VmZWVayz390fEe+NGZERkRFZWVWbV/ZzjY3V3ZWVkZmTEfe/zvM+rMMYgkUgkEolEIpFIJBJJP1E3+wAkEolEIpFIJBKJRLL1kMWmRCKRSCQSiUQikUj6jiw2JRKJRCKRSCQSiUTSd2SxKZFIJBKJRCKRSCSSviOLTYlEIpFIJBKJRCKR9B19sw9gI9m5cyc7dOjQZh+GRCKRSCQSiUQikWwKx44dW2CM7dqI59pWxeahQ4fw6KOPbvZhSCQSiUQikUgkEsmmoCjKuY16LmmjlUgkEolEIpFIJBJJ35HFpkQikUgkEolEIpFI+o4sNiUSiUQikUgkEolE0ne2Vc+mRCKRSCQSiUQybBiGgQsXLqDZbG72oUiGiHw+j4MHDyKTyWzaMchiUyKRSCQSiUQiGWAuXLiA0dFRHDp0CIqibPbhSIYAxhgWFxdx4cIFHD58eNOOQ9poJRKJRCKRSCSSAabZbGLHjh2y0JQkRlEU7NixY9PVcFlsSiQSiUQikUgkA44sNCVpGYRzZlOLTUVR3qgoynFFUU4pivKbIf+uKIryJ+6/P6Uoym3u31+mKMpXFUV5TlGUZxRF+eWNP3qJRCKRSCQSiUQikUSxacWmoigagPcBeBOAGwD8uKIoNwR+7E0Ajrj/eweAP3X/3gTwHxhj1wO4C8C7Qh4rkUgkEolEIpFI+oCmaTh69Chuuukm3HvvvSiXy7E//3//7//Fs88+u27H83M/93M9/f63vvWtuPvuu/t2HB/60Idw6dKlNf2Os2fP4qabburTEQ0Wm6lsvhTAKcbYGcZYG8BHAbw18DNvBfA3zOHbACYURdnHGJtmjD0GAIyxKoDnABzYyIOXSCQSiUQikUi2C4VCAU888QSefvppTE1N4X3ve1/sz/dSbJqmmfhnP/CBD+CGG9JpTeVyGY899hjK5TJeeOGFVI+NopdiM83rHHY2s9g8AOC88OcL6CwYu/6MoiiHALwEwMP9P0SJRCKRSCQSiUQicvfdd+PixYsAgNOnT+ONb3wjbr/9drzqVa/C888/jwcffBCf+tSn8Gu/9ms4evQoTp8+jVe/+tV49NFHAQALCws4dOgQAKdY+5Ef+RHce++9eMMb3oAPfehD+KEf+iG88Y1vxJEjR/Drv/7roccg/r6RkRH8zu/8Dm699VbcddddmJ2dDX3MJz7xCdx77714+9vfjo9+9KP87++77z78wi/8Al7zmtfgyiuvxNe//nX8zM/8DK6//nrcd999AADLsnDffffhpptuws0334w//MM/xMc//nE8+uij+Imf+AkcPXoUjUYDx44dwz333IPbb78d3/d934fp6Wl+vL/927+Ne+65B3/8x3+MY8eO4dZbb8Xdd9/tK9zPnj2LV73qVbjttttw22234cEHHwQAfO1rX8M999yDH/3RH8U111yD3/zN38T999+Pl770pbj55ptx+vTpHj/N9WUzR5+EdayyND+jKMoIgE8A+BXG2ErokyjKO+BYcHH55Zf3dqQSiUQikUgkEskA8P4vPIPTs6HL3p65as8Y3vl9Nyb6Wcuy8OUvfxk/+7M/CwB4xzvegfe///04cuQIHn74YfziL/4ivvKVr+AHfuAH8Ja3vAU//MM/3PV3PvTQQ3jqqacwNTWFD33oQ3jiiSfw+OOPI5fL4dprr8Uv/dIv4bLLLot8fK1Ww1133YXf//3fx6//+q/jL/7iL/C7v/u7HT/3d3/3d/iP//E/Ys+ePfjhH/5h/NZv/Rb/t+XlZXzlK1/Bpz71Kdx777144IEH8IEPfAB33nknnnjiCViWhYsXL+Lpp58G4KikExMT+N//+3/jPe95D+644w4YhoFf+qVfwic/+Uns2rULH/vYx/A7v/M7+OAHP8gf8/Wvfx0AcMstt+C9730v7rnnHvzar/0aP47du3fji1/8IvL5PE6ePIkf//Ef50X1k08+ieeeew5TU1O48sor8XM/93P4zne+gz/+4z/Ge9/7XvzRH/1R1/d6o9nMYvMCAPGsOQggqEFH/oyiKBk4heb9jLF/jHoSxtifA/hzALjjjjuCxaxEIpFIJBKJRCLpQqPRwNGjR3H27FncfvvteP3rX4/V1VU8+OCD+JEf+RH+c61WK/Xvfv3rX4+pqSn+59e97nUYHx8HANxwww04d+5cbLGZzWbxlre8BQBw++2344tf/GLHz8zOzuLUqVN45StfCUVRoOs6nn76ad4ree+990JRFNx8883Ys2cPbr75ZgDAjTfeiLNnz+Kee+7BmTNn8Eu/9Et485vfjDe84Q0dz3H8+HE8/fTTeP3rXw/AKcz37dvH//3HfuzHAACVSgXlchn33HMPAODf/Jt/g8997nMAAMMw8O53vxtPPPEENE3DiRMn+OPvvPNO/vuuuuoqfgw333wzvvrVr0a/wZvIZhabjwA4oijKYQAXAbwdwL8O/MynALxbUZSPAngZgApjbFpxcnz/EsBzjLH/tZEHLZFIJBKJRCKRbBZJFch+Qz2blUoFb3nLW/C+970P9913HyYmJvDEE090fbyu67BtGwA6Zj+WSiXfn3O5HP9vTdO69jhmMhk+5iPq5z/2sY9heXkZhw8fBgCsrKzgox/9KH7v937P95yqqvqeX1VVmKaJyclJPPnkk/jCF76A973vffj7v/97rlgSjDHceOONeOihh0KPk14nYyxyLMkf/uEfYs+ePXjyySdh2zby+Xzo+yIeJx3jILJpPZuMMRPAuwF8AU7Az98zxp5RFOWdiqK80/2xzwI4A+AUgL8A8Ivu378CwL8B8FpFUZ5w//f9G/sKJBKJRLKRtE0LTcPa7MOQSCSSbc34+Dj+5E/+BO95z3tQKBRw+PBh/MM//AMAp4h68sknAQCjo6OoVqv8cYcOHcKxY8cAAB//+Mc3/Lj/7u/+Dp///Odx9uxZnD17FseOHfP1bXZjYWEBtm3jbW97G/7Lf/kveOyxxwD4X+e1116L+fl5XmwahoFnnnmm43dNTExgfHwc3/rWtwAA999/P/+3SqWCffv2QVVVfPjDH4ZlDfd9b1PnbDLGPssYu4YxdhVj7Pfdv3s/Y+z97n8zxti73H+/mTH2qPv332KMKYyxWxhjR93/fXYzX4tEIpFI1pc/+5dn8Z8/9uhmH4ZEIpFse17ykpfg1ltvxUc/+lHcf//9+Mu//EvceuutuPHGG/HJT34SAPD2t78d/+N//A+85CUvwenTp/Grv/qr+NM//VO8/OUvx8LCwoYe79mzZ/Hiiy/irrvu4n93+PBhjI2N4eGHk2WMXrx4Ea9+9atx9OhR3Hffffiv//W/AnDChd75znfi6NGjsCwLH//4x/Ebv/EbuPXWW3H06FEe8BPkr/7qr/Cud70Ld999NwqFAv/7X/zFX8Rf//Vf46677sKJEyc6VN9hQ2Fs+7Qx3nHHHYwabCUSiUQyXPzHjz6CC4s1/OW7Xr3ZhyKRSCQbynPPPYfrr79+sw9DMoSEnTuKohxjjN2xEc+/qcqmRCKRSCRJsRlD0xjMnhSJRCKRSCSdyGJTIpFIJEOBzYCW7NmUSCQSiWRokMWmRCKRSIYC22ZoGfZmH4ZEIpFsCtup9U3SHwbhnJHFpkQikUiGApsxGJYNy5YFp0Qi2V7k83ksLi4ORPEgGQ4YY1hcXPSNTtkMNnPOpkQikUgkibFtZ5HVMmwUc3KvVCKRbB8OHjyICxcuYH5+frMPRTJE5PN5HDx4cFOPQRabEolEIhkKLLfYbBomijl5+5JIJNuHTCaDw4cPb/ZhSCSpkVvDEolEIhkKbOYpmxKJRCKRSAYfWWxKJBKJZCiwuI1WJtJKJBKJRDIMyGJTIpFIJEOBp2zKYlMikUgkkmFAFpsSiUQiGQps3rMpi02JRCKRSIYBWWxKJBKJZCggZbNpmJt8JBKJRCKRSJIgi02JRCKRDAXi6BOJRCKRSCSDjyw2JRKJRDIUuLWm7NmUSCQSiWRIkMWmRCKRSIYCGRAkkUgkEslwIYtNiUQikQwFli17NiUSiUQiGSZksSkZCNqmhbYp1QqJRBKN7NmUSCQSiWS4kMWmZCD4g088jj/5zHc3+zAkEskAY0kbrUQikUgkQ4W+2QcgkQDAwkoDtZY8HSUSSTRyzqZEIpFIJMOFVDYlA4FpM7SlNU4ikcQgA4IkEolEIhkupJQkGQgsy0Zrsw9CIpEMNFLZlEgkEolkuJDFpmQgMG0Gi8kFpEQiiUYqmxKJRCKRDBey2JQMBJbNuGohkUgkYXhptLLYlEgkEolkGJA9m5KBwLRstOTok56oNgzMlOubfRgSybpD+1Gy2JRIJBKJZDiQxaZkILBshrZcQPbE/d84gd+5/zubfRgSybpjSWVTIpFIJJKhQhabkoHAUTZtMCattGlZbRpYrsl4JcnWhjHGezZlQJBEIpFIJMOBLDYlAwEpFoYlx5+kxbIZmm1TFuqSLY3Y0i2VTYlEIpFIhgNZbEoGAs8eJ4vNtFg2g82AlinfO8nWxRY2U6SyKZFIJBLJcCCLTclAYLqKZluGBKWGCvVm29zkI5FI1g8xrVoqmxKJRCKRDAey2JRsOpbNQMtIuYhMDxWb9ZYsNiVbF1I2i1kdhmXz814ikUgkEsngIotNyaZj2Z79sy2toKmh96/ZloW6ZOtCymYhpwGQG1MSiUQikQwDstiUbDqm5SkU0kabHlJ4GoZUNiVbF1HZBGSxKZFIJBLJMCCLTcmmI5XNtUGKT0PaaCVbGIsrm7LYlEgkEolkWJDFpmTTEZVNuYBMD1c2pY1WsoUJKpsykVYikUgkksFHFpuSTcf0KZtyAZkWaaOVbAfoPC9KZVMikUgkkqFBFpuSTcfyjTSQNtq0WNJGK9kG8IAgqWxKJBKJRDI0yGJTsulYMiBoTVDPq7TRSrYytCclezYlEolEIhkeZLEp2XREG21LFpupIWWz2ZbKpmTrQsqmTKOVSCQSiWR4kMXmgNI0LPzHjz6CS0u1zT6UdcenbEobbWqo2KxLZVOyhaGAoEJWztmUSCQSiWRYkMXmgDKzXMfDJ+fw/MXyZh/KuiOVzbUhlU3JdiAYECR7NiUSiUQiGXxksTmgGJZTgInhOVsV0xaVTbmATAspPrJnU7KV4cqm7NmUSCQSiWRokMXmgGK6xaao+m1VLEtUNrf+6+033pxNqWxKti7Bnk2pbEokEolEMvjIYnNA4cqmtc2UTWmjTY2XRiuLTcnWhZTNjKYio6lS2ZRIJBKJZAiQxeaAYnIb7dZX+uSczbXh9WzKxbdk60LnuaoqyGVU2d8tkUgkEskQIIvNAcXgNtqtr2yKNlqpbKbHtCiNViqbkq0LKZuqoiCf0eXmikQikUgkQ4AsNgcUKiC2k41WAdCWPZupsaWyKdkG2D5lU5M2WolEIpFIhgBZbA4o2zEgqJDTpbLZA2S1lsqmZCsjKpuy2JRIJBKJZDiQxeaAsh1HnxRzulxA9gCdIy3D2hbni2R7Qm57TfZsSiQSiUQyNMhic0DhAUHW1lc2Sb0tZnVpo+0By2bI6c5XWRbrkq2Kp2wCuYwmbeMSiUQikQwBstgcUKhnc3sEBElls1dsxsAAlPIZAHL8iWTrIvZs5jPyWiGRSCQSyTAgi80BxdxWNlpX2ZQ9m6mh82NEFpuSLU6wZ7Mpi02JRCKRSAYeWWwOKMY2nLNZzOpoSRttKjqLTbkAl2xNSNmUPZsSiUQikQwPstgcUPiczW0w+kS00balWpEK2owo5XUAUtmUbF1oY8WZsynTaCUSiUQiGQZksTmgbEcbrTP6RCqbabCljVayTeA2WlVBTpcBQRKJRCKRDAOy2BxQSNHcDjZaeq3FrA7DsrdFgd0vgjZauQCXbFVsUdmU1wqJRCKRSIYCWWwOKNvKRisEBAGAIXuxEkOLbUqjrUtlU7JFsZi/ZxOADBSTSCQSiWTAkcXmgEI2WnObKJuaqiBLsyKllTYxVGyOSmVTssUJptECcq6sRCKRSCSDjiw2BxQqNu1tYBOzbBu6qiDrLiBlSFByeJKvDAiSbHHEOZtUbMrxJxKJRCKRDDay2BxQDNc+a26DYtO0GTRNRU53i02pbCaGLMh5XUNGU9FoyWJTsjWhS6GqKMi71wqp5EskEolEMtjIYnNA4TZaa+sXXpbNHGWT22jlAjIpljB7sJDV0JBKj2SL4imbQC7r2mjltUIikUgkkoFGFpsDyray0Vo2NE1FliubcgGZFEuwFuazOprSRivZotg8IEhFXvZsSiQSiUQyFMhic0DhabTboNg0XWWTEiZbxtZXc/uFp2yqKGQ11KWNVrJFEUefyIAgiUQikUiGA1lsDijGNrLRmpYNXSqbPWH7bLS6DEyRbFks0UYrezYlEolEIhkKZLE5oFCRuR2Glls2c9QKXaoVaaGAICo2ZRqtZKvCbbSKYxkH5LVCIpFIJJJBRxabA4rpptFuh2LTUTYVZPmg9q2v5vYLOj90zQ0IasnFt2RrYvlGn8gwMYlEIpFIhgFZbA4o28lG66TReqNP5AIyOf6eTR0NQyqbkq0JKZtiz6a0jUskEolEMtjIYnNA2U42WmfOpoKsu4BsywVkYkS1p5DVZA+bZMtii8qmtNxLJBKJRDIU6Jt9AJJwvGJzGyiblu0qm9JGmxZxzmY+q8s0WsmWRVQ2VU2Bripyc0UikUgkkgFHKpsDCo082Q6jT6yAsinViuTwYlNxAoIMy94W1mvJ9sN2g8SIfFaTlnuJRCKRSAYcWWwOKNSzaVlbv9g0bUfZVBUFGU2VymYK/Gm0so9NsnWxmXOeE7mMJjemJBKJRCIZcGSxOaBsJxutaTHomrOIzOqqVCtSEJyzCQANaaWVbEEsxqDKYlMikUgkkqFCFpsDChWb28ZGqzqnYlbXZEBQCoJptADkrE3JlsS2GTTBRpvTZbEpkUgkEsmgs6nFpqIob1QU5biiKKcURfnNkH9XFEX5E/ffn1IU5Tbh3z6oKMqcoihPb+xRbwyGSTba7aBs2twel81IG20aTJ+y6dhoGwlCU46dnsd97/2KLEwlQ4NlM6jCHSuX0aRlXCKRSCSSAWfTik1FUTQA7wPwJgA3APhxRVFuCPzYmwAccf/3DgB/KvzbhwC8cf2PdHMw3F7N7TD6xJmz6RSbOV2GfqTB17OZS65snpuvYqbcwLn56roen0TSL2zWGRAki02JRCKRSAabzVQ2XwrgFGPsDGOsDeCjAN4a+Jm3Avgb5vBtABOKouwDAMbYNwAsbegRbxCWzXjMv2kzMLa1C07TtqFpZKOVymYa/HM2qdjsvgCnAKoLC7X1OziJpI/YdqBnU9poJRKJRCIZeDaz2DwA4Lzw5wvu36X9mVgURXmHoiiPKory6Pz8fE8HutGQWpVzR4HYW7zYtCxP2cxmZM9mGnxzNjNko+2ubJJN+8Li6vodnETSRzqUTRkQJJFIJBLJwLOZxaYS8nfBqirJz8TCGPtzxtgdjLE7du3aleahmwapTlQ8mFt8/IkzZ9M5FaWNNh22EBBUTGGj5crmolQ2JcNBh7IpezYlEolEIhl4NrPYvADgMuHPBwFc6uFnthxUXJKyudX7Np05m97ok7YhbbRJ8SmbNGczjY1WFpuSISGsZ1MqmxKJRCKRDDabWWw+AuCIoiiHFUXJAng7gE8FfuZTAH7KTaW9C0CFMTa90Qe60ZgBZXOrz9q0LAadlM2MVDbT4LfROspmPYWN9uJSbctvZki2Bs6IJNmzKZFIJBLJMLFpxSZjzATwbgBfAPAcgL9njD2jKMo7FUV5p/tjnwVwBsApAH8B4Bfp8Yqi/B2AhwBcqyjKBUVRfnZDX8A6wm202e1hozVtYfSJrsmAoBSIabSaqjjWwhQ2WsOyMV9prOsxSiT9wGbwKZu5jHOt2Oo97RKJRCKRDDP6Zj45Y+yzcApK8e/eL/w3A/CuiMf++Poe3eZByua2sdEKAUG5jCrVihSIabQAUMhqydJohYL+wuIq9k4W1+cAJZI+YQmbUoB3fWwbFvLZTb2VSSQSiUQiiWAzbbSSCEjJ5AFBW9hGy5gz5kVzp7VrqrrlbcP9xLIZVMVTfApZPXFA0Eg+A0D2bW4WL85X8dS5xc0+jKHBtlmHsglAhgRJJBKJRDLAyGJzAPHSaJ3demsL22hN6jnUnEWkrikwtvDr7TdOH5v3NXaKzWTK5u7xAko5XY4/2STu/+Yp/KePPYq27FFOhM3gS6PlgViy2JRIJBKJZGCRxeYAEuzZ3MpKn+W+Vt0tmHRV5X8n6Y7FWMcCPGnPZkZTcXDHiFQ2N4l6y0C9ZeKJF6S6mQQnjdb7c153ro/Sdi+RJOfDXz+Bj37r1GYfhkQi2UbIYnMACabRmlu4Z9MKUTZNm4HJ0I9E2IGEzkJWT5RG2zYtZHQVB3eUZLG5SdCImm89v+UDtvtC2JxNQBabW5W2aWF5tbXZh7HleOD5GTx6en6zD0MikWwjZLE5gHSOPtm6hRcV0hQQRJbQJAmTM+U6njy7sH4HNwQEQ1MKWS3xnE1H2SxhodpM1Oe5Vp49v4Szc9V1f55hgd7zh47P8u+8JJrgnE1ZbG5t/u/DZ/HOP/uG3HjsM5VaW15vJJIAH/zy8/jc4y9u9mFsWWSxOYB4Nlq3Z3ML22jppkdzNnVX4Uwy7uXjD53Bf/vHJ9bt2IaB4OzBxAFBpu0qmyMANiYk6E8+8zT+5mvH1/15hoWmYaGY01FtGHjq3NJmH87AE6Vsyp7NrclCtYlKvZ3IqSFJhs0YKvWWL41cIpEA//LkeTx6Sir+64UsNgcQKrRyGdX3560It9G6i0jq3UySwFttGBuiyA0yncWmljiNNqOp2DWeBwAsrTbX7RiJettEtWms+/MMC03Dwp1X70Yuo+EBaaXtisUYNEHZzEtlc0tDn2ul1t7kIwnnxKUyyrXhsvlWGwZs5m1oSyQS5/tQlor/uiKLzQHEDKbRbgdlk0afuMpmkgTeestAe4vs0D5/cRmrPRRiUWm03axnVGzS+74Ricctw0Ktub03B0SabQvjxQzuvHoXHnh+Zkvb5fuBVDa3F1RsDmpB97sf+Q4+8dCZzT6MVFAPrCw2hx/LZlisrv8m8XZgyX0ft3I+ymYji80BxAgGBEllM5R6y4TN2NAX45bN8Gt//W388yNne3ps0EZr2azrYoJstGRf3ogdvWbb7Kmg3qq0DAv5jI5XXb8P5Vobz56XVto4rIg5m1LZ3Jp4xebgKZs2Y6g2DKy2hmvzrFJ3i80tskm7nXnguWnc996votoYvO/HsLFAxabchFk3ZLE5gNAJn9sOAUHua9W0XpRN50Y/7Opm27RgWDaWe9jBDy7AaVxOt1mbXNmkHtl1PsdsxtAybdRksQnAOe8Ny0Y+q+HOq3cjo6n41vMzm31YA43N/Bsrw2CjfercIh4+MbvZhzGUtNz5s+X64C2m24YF5v7/MEGFu1Q2h5/F1RYMy8ZsubHZhzL0LK7IYnO9kcXmAEJK5naYsxlMoyVlM8nNkHoTh73YpMVytdGLjdafRlt0Q6W69W0apo3sBiqb9BmRGr3dIetnPqOhmNNx+1WOlVa+N9EMo432I988iQ9//cRmH8ZQ0uQ9m4Nno6XNvEHe6AiDLMlS2Rx+2u5mTC+b1BI/81zZlPff9UIWmwPIdrLR2sE5m+5iMoma6ymbw3XDD0KFWC8W0+CcTUow7jb+JNizud7KZtMtfhmAxpBZz9YD+nzo83rldXuxsNLE8YvlTTyqwcZm8Kn4uqZCV5WBXvCXV9vYwsaUdYXbaAdQ2aRCuDVk9x6pbG4daMNgSc6iXTOLvGdTfi/WC1lsDiDbykZr+wOCkiptjDFebA77Li0Vy70pm4y/Z4CTRgsgdlwAY0zo2aRRM+v7HooFgezbBJqG8/nQhtLLrtkDXVWklTYG22YQ9lUAONfIQS42l2stqVb3yCD3bJJzZJDPvTCocDcsW84vHXJok3pJhgStGWmjXX9ksTmA0K5jwVU9tvJuC6m2VPQk7SE0LJv/zLDbaNuGc/y9FpvBgCDAUxKjHsMAt2dzY2y04qJMJtJ6NjzaHBgtZHD08E488Ny0XARGYDO/jRYY7GLTsm2s1Nvy8+yR1gDbaEnZHLZ7j/heyuTN4YbWidJGu3akjXb9kcXmAGJaNhQAWd0dS7GFbwrUj0rjO+j/u/Wp1gUr5rDbaMmK1evoEzUwZxPwvz9B6CaV0dVUtuW1IPbVrbakskkL6Zw73ggAXnH9XsyUGzg1s7JZhzXQWLZ/zibgFJuD2rNZqbfBAMhaszeaA6xs0mbeoJ57UYjv5bA7grY7hrtuWKrKYnOtkI12K+ejbDay2BxATMuxRpLqtJWLzUhls8sOk7/YHO4LhGijTauCBAOCCgl6Nun9EpXN9e7h8SubstikxSqFgAHAy6/dC1VR8MBz05t1WANNmLKZH2Bls7zqLOyljbY3uLI5iD2bbVI2B/Pci6Jc9woT2bc53LQt2bPZD2zGeMEulc31QxabAwiFt1ARsZV95P1RNof7/aHjtxmL7bUMI8pG2zBilE0qNnWNPzbJqJm14O/ZlDZaMY2WGC9mccsVU/jW8zPSehmCzfxjfoDBVjbJ3iaLzfTYjKFtOg6fSn3w+l6Htmez1kbOdUxJZXO4oc9P2mi7c26+isdfWAj9t5V6G4brJuy1ZW16uY6/++ZJed+OQRabA4hp2dA1RSi8tu4JTDtJVPQkVjaFoswYst3lIL5CLGXfplNsdgYENVrR7wm30WoqFEWBrirrvsvdlMqmD+rZFItNwLHSXlis4dz86mYc1kATtIwDA65suotAuf5ID82v3DGWh81662dfT3garTE8BVvbtFBvmdg5VgAglc1hxxACgmSRE88/PHgaf/Tpp0L/bcENB9oxlu9Z2Pn6M5fw1187IcMPY5DF5gDiFJsblxS6mXhptP45m912mLaisgmkX1QFlc1cRoOC+DmbVJxnXAutpqnr3qvQlGm0PiiNlpRo4uXX7oUC4FvSStuBk0Y7PAFB1B83aKrcMEDXiz3jTmFUHjD1ZhhttBX3fNw1lgewtdcV2wGy0bZMOzajQeJsCkW9Rwtuv+beiWLPNtqqu6YZps2njUYWmwOI4fZsahsU3rKZ0Guj3sGkts6tGBAEpC/EgqEpiqKgkNVjrYViQBAAZDQFxgbaaGvyxijM2fQrmztG87jhskk5AiUEm/k3VoDhsNHKWjM9dL3YM1EEMHghQbSZ1zbtodlMoLEnXNkc8k3a7Y74+UX1bT56ag4fe+DURh3SwGLZdmSOxSIvNgs9b8CQI23Y16LriSw2BxCT92xufRstvTbPRptsFIev2BzyHVpjTcqm3bEAz2e1ZGm0mtcnu1HKZjGnSxstvPcjF7DRAsArr9+Hs3NVXFysbfRhDTT2kNpo7S18/V4vWgFlc9BCgsQNjmFx1tD5SMqmtNEON23T4nOHo4rNrz59CX//wOkNPKrBxLBsGJYdus5ZWGlCVRTsHCvAtFlPluRVrmwO5r1oEJDF5gBi8J5NxWla3sI3BXptVGQmnbPZ8PVsDvf7I16gqr0om4EFeCGrx87Z9AKCvPd8vVPY6DXuGMlJGy2cxWpOVztsoQBw97V7AADHzsxv9GENNFEBQYN6g1+WNtqeabnXqN0Tg2qj9a6vg3r+BSF1eNe47NncChiWzVXqpdVm6M+0DAu1lrntFTcSNcLUzYVqE1MjuTWNGqQ1zbBsPG0GstgcQKhnE3AUv22lbCbs2axtIRuteIHqJSAoqPYUshoaSWy0vMBX131Do2VY0FUFY8WstNHCWazmA/2axNRIDkD8rNTtiG0j1EY7qIt9GRDUO/SZ7horOIm0A2aj9SmbA3r+BaHzcberbMqF8XDTNm3scTdjliOUTWrRifr37QKtb8JaLharTewYzQtrzx6KTXfd1hryteh6IovNAYRGnwAbE96ymQSVTU1L3rNJyavDftOkQiyjqag20i2qgmm0gKNsNuJstEFlU1V7usCmoWlYyGU0jOQzUtmEs8Ma7NckMhs0+3TYsBhDoNZ0is0B7Zsry9EnqfiVDz6Af3r4BQB+2/1YMeubDzkINASFZFB7hoOU623kMhpGC1kAgGENx3FLwjFMG5OlHDKaGmmjpU2b7T4ehZxbYcGJCytN7BzLrymQkxxpw7LxtBnIYnMAMYViU1eVdS8ENhMeENRDGu1oIQtVGf5i07BsZDMaRgrpCzE7zEab0+PTaN2LaVYTbbTrr2zmsxpK+Yzs2YSzQA2OPSEUxdl4GHZ7eL8JS6Ol93DQbvI2Y9y2KMcSJOPM7ArOzK4A8BbJeV3DeDE72MrmkKgZ5dUWJkpZvqiW15fhxrBsZHUVU6O5aGXTTUfd9sqmu56MstHuGM0nzgsJg/dsyu9UJLLYHEBMN40WcBS/rWyjpS+2FujZ7K5sGihmdWR0bUvM2czqKkbzmTWPPgGAQkbz7bwHCVU2N6DYdJRNHatNaQ91is1wGy3gqJtS2fRjs07LOAUsDZq6tNowYNkMCqSymQSbMbRNm29EUU9kLqNhopTlSapBLJttyv2xMYw9m/U2Joo56ZzYIhimjYyuYbKUS6BsDtZmzUZDymbwPlFvmc7sWV+xme56YtmMt7wM2qbnICGLzQGEAoIAQFUVWFv4pmAGeja1hMpmo22hmNOR1dWhVzbbpoWs7iib6QOCwtJokymbvp7N9bbRti3kdA2lXAb1lrHtF+CNthlpowWcjYBhUUw2CitG2Ry0BT/Z1iZKOWzhvcK+QZ8f9XOTQpDLaBgv5iIDgv7wn5/Cf/+nxzfmIAVabQsjeWeziI71/MIqvvLdixt+LEkp1xxlM6M735n1DoWTrC9t00JGUzE1kosOCJI9mwCEns3AuojGnuwcy3N3Xbe1ZxDRqSV7NqORxeYAsq1stJYNVQFfRHq++e49m1ul2GwZNnKustlLQFCw2Czm9MiZUoBnOyZlU9PWf0OjZTo22pF8BjYL753YTrTa0TZawPlspM3NgzYnopTNQSs2qTiaGslJG20CeLEZGCHAlc0IZebSUg0z5frGHKRAwzAxXnSCvOhYP3PsHP74009t+LEkpVJrY7yUlcrmFsGz0eYji0lS8pYjitHtArfRBu4TC26xuRYbrdj6NOxr0fVEFpsDiJhGuxFJoZtJMOAm6WzRestwi01t6BWgtmkhm9Ew2pOy2Vls5jMaGm0zcpHrKZua+//rb9mkgKCSqwbUtrmVtmlYkWm0AJDVpY1WhGZVho0+AQbPRkvF0dSoVDaTQJtjXNmkns2shomSMy4p7D7YMqyuLRfrQaNtYbzoBO20BVW2ZdoDubnAGHOUzWKObzIOe/vJdoa5tvOMpmKylMNKwwi9X0gbrYMXEOQ/5xdXnGJzl6hspryeiGu2Qdv0HCRksTmAmLbXs6kqypYeCu68Vm8BSbNFuy20621H2dwKQSpt09mhHC1ke1Q2O9NoGaIvfMGezY0Yr9NybbQj+QwAbPuQoKZhxiub2vAr9v2ElM2wjRVg8G7yZKOdHMlt6et3v2gGlM2mYUFVFOiqgomSU9RVQvo2m4aV2vbWD1pCsUk2WurbGsRNolrLhGkzx0Yrlc2hh9xuFBAEhFtlebEpbbQAnPuuSF+UzYZUNpMgi80BROzZ1LUtbqO17Y5iSdfUrrZOZ/SJa6Md8ptmy/QKsXrbTHWxC52zmXMW4FEhQZsxZ5OUvBIVm9t8hmTc6BNABgQFse3wYnNgbbSrLaiKgomitNEmQezZZIw56dUZDYqi8KIuLJG2ZVob3nvIGEOjbWLcLYLp2HlIyAAuOMtCD7GnbA7ecUqSQW6ujK7yuczBgtK0bL6JvN1Hn9D7EGwvWqg2MZLPIJfRvLF7Kdfbq1LZTIQsNgcQwxR7Nre2jdZJ3vUvILsV2DZjXkBQZgvYaA3PRgsg8fgTxhhsFpZG69gzo/oiDdNRDTQ+bkZZ9wVby7SQy6hc2dzOszaZe/7GKZtZXZOLQQGLDZeNdtntj1NVJbWN9vilMj701ePrc2ADCikOlu0UmpReDTgFEoDQWZutTVA226YNBgjKpr/YHMQFJ9m6J0pZbheUm1nDC90bsrqGqZE8AHSEBNF5qSqKVDZ5QFCnjXbnmPP+9ar4+3s2B++7PyjIYnMAEXs2ncXK1t0ZN22bz9YkNFWFFbOAoCJqqwQEcRutW4glHX8SaS3MkrIZUWxaNt/dBiiNdp2VTbe4KuWcQng7F5uGZcNmDIWYnk2ZRuuHTs+gij+oNtpyrYXJUg6qkn7O5mcePYePfuvUlt5kDCJuFtRapltsOtcoKurCQoKabWvN71PLsPBfP/EYLi3VEv08XVd5sclVWeeaNoibRFzZLObkHN8tAG+F0TxlMzj+hM7L3eN5NA1rW4fyxdlod446xaaXF5Lue0HrtZG8LudsxiCLzQHEHxC0/qrTZmJZjNsXiG6vmXaQKSBo2G+a4ugTAIlDgqwIa2ExR8pmtI2WdvGAjbHRtty5kiXZs+mFn3Tp2eyX8rCw0sT/+1cPDt3udtOwcN97v4JHT897abT+U32AlU1nzISiKGBIV3CemK4A2F4x+i3hWlVrGuHKZsAKaNkMhmWv+f54br6Krz87jafPLyX6eTrXRgoZqIrCA4K4sjmAnxv1u5L1N7MB464k6wdvhdFV5zqDsGLT+Zm9E0UA27tvk8714JpoodrEjlFSNknxT2+jzWiOa0vO2YxGFpsDiGExv412EwIQuvFPD7+AX/7LB3p+PClbps06lM1ur5kXm1kdWW34FaCW6Y4+IRttQmUzqtiklNPgTCnCGQYtFJvrPF6HFoU5n7K5fXdZycqTi+nZ7Kdif3qmgmcvLOPcfLUvv2+jWFxpYqbcwLn5qpdGOzTKZhsTpRy3/Sb9ejXbJl50P6dBe03rSVO4hq+2TNd273y2I3kdmqp0BATRdX+tG2WLVWcRnnTTkr6/+YyOfEbrCAgaRKcNqcKkxjqjlbbP+bXVoHMsq6nQVBXjpWyksrl30i02t2nfJmPM69kUrqmmZaO82uI2Ws1dc6cdA7faNDBSyCCrawO50TQoyGJzwLBspw+Pis2NSArthVPTFZy4VO7pRj9XaeBH3/NFPHl2AZZldxRLztzHZMpmZivYaN2ezRFuo00WU+4Vm4E0WneRVh8QZdObmadC11TkMxq3nG1HyM7UVdns03lNN9hh69Facb8HhmkLyuZwBAS1DAvFrAY63KTK5unZFV6YkjKxHRB7qepNA822V2xSSFDQRkuf+Vo3yqjXLek5RFa8QlZDNqM641dsxr9ng7j5Wa61MFrIcMdURhv+YL3tjKdsOt+RqZE8lqqBnk0qNre5sileH8QN+KXVFhjAbbQZtceezYaBUTdkaNjXouuJLDYHDPKLU2iOpqlrmiN2dq6Kr3z3Yl+OTWS1aYAhPI6+G3OVBmzGcGp6JXROpKNsbicbrY2crmG04Ow6J+1njFI2C7kEyqav2Fxfq7ZnG3WOaySf2dY9m83A+xFGRtf6Vhx6i+Dh+p5QsUmLeaBT2dQ1FZqqrEuxGTxH26aVuH/ecQ9ogrKZ7HEnL1X4fw9aAb2eiK91tWk66dXCZsxEKddho6Xzeu3KprNIT/r9aHBlU0Muo6EV6IcbxE2Ccq2NCVfVBEjZHLzjHBZEtWwzENNoAWfEUlC5bAaKzaDyuV0Qrw+iskljTzxls/c02pF8xnEjbaNrdlpksTlg0AKT92yqyppstJ8+dg7v+eSTfV+4UFN0LxcwUrWml2tOQJAWDAhSYq0MPCCIRp8M4E5yUmzGeGDPSN4pPpIGBNHGREex6dpo62kCgjZE2XQWj6W8jtp2ttG670chbvRJHxeD9H0ZtsVlte58D9qW7Y0+CSibgHNe9TuY4filMn70Pf+C6eU6/7t//1cP4n996slEjzcsG7qq8GIzacvm8Utl/t/bqdj0BwS5PZu6WGxmOzY26f2xbLam8TLpbbSuMyGrI+da5+rCKKdBtKeW6y3e+wrI0Uoixy+V8YUnzqd6zLHT8/iR9/yL73PfSLw0Wuc+PjWSiw0IUpXOnuftgriRLjooFla8GZuAl0abdi1UdW20zn1o8L77g4IsNgcM+mL4bLRrUJ3qLRM2Yzgzu9KX4yNo1z8Yt53omNxCY7rcCFc2u9wI/crmcNto6dhzugZNVVHM6amVzY45m24RE4z5JgzL5jcpwFGS13OXttlRbGa2tY2W93x1nbPZnxsXPV+7T79voyBls21YkTZawDmvolT8XllYacJmwMJKg//dbLmBLz11EY+/sBD7WMYYD3lLa6M9OV3hm06DFnq0nrQMi4c/1QM9mwBcG224cgOkVyNE6B6WdNOyGaJsikXHICZSVtxRPIRMo/X4/OPn8f4vPJPqMZeW66i3TH6N2miCs7KnRnJYXm35HBR0PheyOsaLuW1roxXTZUUHAjkavDRa5wKU1pbvKZsa2gPoahgUZLE5YJgBZbPbGJBuNNyb4MnpSpefTAclpoZdwL7xzCXMCIpAkJp7TDPLdV/yLtGtT9XfszncPnmyXWTdmP/RQiaFshluo8241sLoOZt+G62mKjAte92GzwfTV0fyGalsIt5G289NFN6zOWTfkxX3e9A27chzHXDOq34XZnQdFgsHWry973NPxxYmNmNgcNRpKo6tBN+tWsvAhcUabrxsynnubVRsNg0LI3kn3XU1kEYLODbaKGUTWJuVlpTNpD2MDcNbxNP3tN4a7Fl75VpA2dSlskkYpoVG20r1udHPrkUIWAvtgLI5OZKDZTPf2kHc5J0MUT63C0aMjTYrBDPqPSqbqw2n2MzpqlQ2Y5DF5oDh2WgV/v9rCUCgguPEpf4Vm4wxnpgavIC1TQv/7Z8ex+cffzHy8VQszpbr3G4mkuky95GKVbrZ24ytqSDfTOjiRJaxyVIOc5VG3EM4UdZCRVFQyGqRxWY7mEarqWBI3leWlqCyOZrPYK7SGMhF2UbQTBEQ1I8NAHq+YVtcUlCW2CsZVPEBcHWpn9B7RZtBNmNomzauPziBC4s1/OO3X4h+rDADj443ycd4atpxn9x8xTYsNtsW8lkdI3ldmLPpVzYbbcu3WPQVm2u4R6bt2fRstJ6yWROUzUHb/LRsGysNA5NCsdnNPbSdoM+rEjLHtdtjNus9FK8xgBMQBMAXEtQKFJthc2q3A5bgFgzaaHeM5qEotNZOX2xaNkOtZWI0n0FWBgTFIovNAYNfRFRR2VyDjda9MZ7qo7LZMm1+kQ0qm/MVx34Wt0tMFkrTZpgpN3jkNNE1jbZtIqc7yaa0szesX3L+ebuv44bLJvH8xXKihSadF0FlGHAK8eRzNl37yDrt0lJ6IxVXr7vlACr1Nj7x0Jl1eb5BhyubXUafMKzNHhh8vqFTNuuesslHn4TYaPPr0LPpKZv+cKW7r9mLV16/Fx/55slI9wbNaUtroz3h9mvefPkO57m3UbHZcgOBSvkMn7PpDwhyLKAVwUrbD2XTsGyumCbttfQFBOmdNtpB20Sj1zcRsNGu92zlYYG+2+UUYYd0Ld2s95BaIrI8jdbZSBA3/33FZik7FKNPLiyu4smzi339nbRWHcln+FoEcDaZyEILCOugFPdccjSMFFxlcxtds9Mii80Bo9NGq6xJtaOdnBcXqn3raxJHcwSLzdmKswCLK1zEXeDl1VaHstltzmajZaKYc6wPdLEd1mKT3xDc13HroR0wLBvPXVju+tg4a6FTbCZNo+3NPpIUSmckpeK2K3fhldftxUe/dQqz5Wi79aDwmWPn8OdffLZvv8/r2YxLo+3fJgotjoftO7KZyiZ9t+j3inb3n3/DDVAVBX8a0edlCv1UaeZsnppZwe7xAnaNOwug7WTJahomchkNxZyOcq0Nm6HDRgv4CwL/zLzeNmWWfYvzZN+PlmEho7mbna6a4Ss2B6xvq7zqn7EJyDRaESpG0iTr02b6ZiXSBjepqdhcjiw2nZ7N9WqV6Rd/+/WTeM8nn+jr76TPaCSvo9m2+HuwsNLgSbQA+Lz3NOsgsi2PcGVz+1yz0yKLzQHDDKhVax1LUW+bGC9mYTNnhls/WBX6ApYCu2WzZccCGlcg15pGR8+gSLfXXG+ZKOSchYi3KB/OLznvvXB7Nm+6fAqqoiTa3YtKowUc1SxW2RSSHvUeG+OTEuzZBIB3vOEGQFHw5198bl2es5988jtn8a3nZvr2+5qGCQVATo++/NL3ox82LW6jHbLviNizSafmRgUEcRut+/0UN4V2jRXwk/ccwcMn5/DQ8dnIx+qawpVNO8F3q9poY2okx4usqICvrQgpmSP5DA/sCVM2y31WNhcF22HSe0ijbXJXQs6dsznIyma57rxnMo02HPq8KimUP7qWbraNNhsoNoPKpqoo0FUFkyM5GJbt2+gfROZWGvy63y+4slnIgMFx5jHGsFht8SRaQNx0T74OojBHCggaxLFHg4IsNgcMb1hv/wKCqAfoZJ/6NukLRrtlIqRUdSsWD+4o8SJJU4MBQfHW4XrbRNFVhbK0KB/SXdpgz2Ypl8E1+8cTFpvRak8xTtnssNGur7JJ1hVRqdg9XsAPvuwwHnh+ZqAj2RerTby4sNrXRUXDHVivhBROhKfYr33hSovyYVtcinM2uY025I6V74Oy+Y/fPoMPfvl5/mduo3V/L31PaXH3r156GFfsGsGffuGZjnAi0Z2SZs6mZTPomsqLrO2URtt0ezRLOZ0H9viUzaKzmK5EKZs93iOp2MylCMxpGhYfLyWOPlHgbNxthoPgxfkqzi+shv4b9eoFbbTDdj1YL2jtkErZ3GQbbTCNNp/VUczqvukALdNG3r3PUL/uoCfSLlabaBlWX89NWj+P5B03XLNtYqVhwLBsn7KpqQoUpPtMaS086tpobTeJXNKJLDYHDM+CRYVYfDJrHIwxNNomDu4YwdRIrm+JtGQduHzXCJYC1ozZSndls94yMVrIYPd4AYDnlSd0Nx017vHFnFtsDrmNNpgqBzhW2uOXyl1neHk22s6vcT6rR48+CQkIAtbTRhveo3jlnjEAgz1s+im36O+nKtg0rNh+TUBQNreIjbbaMPDez343lQJZ5b10tqfir9OczYeOz+I7J+f4n2mzjN6zdsAKrmsq3v2mmzBXaeCj3zzp+120UMoKxWYS9xqFpWmqioy2vfp/xJ5NWsD5AoJ4z2bb9xiiV/cPXXv2ThYTfz8abZNvCPDRJ20ThZyO3CZZ6f7o09+NtHXTZl5HGu2Q3jP7zZqKzU2y0bYDNlrASaQN2mjpOzQZonwOGowxLLkbTbWE49+SQNcGr9i0+EgrsWcTcK7radbbqwEbLbC92h/SIIvNASPYs+kUXr1d0Fqu/ayY1XFk33jfik1aDFy2cwQtw/LZNanY7NazWcxlsG+yCCBE2ezyhV+ptzFacBYf2WG30VIvmGBrvfXQTlg2wzPnl2IfG9+zGZ1Ga1iWX9l0H79e/SdkLRFfIwBMhljjBo3HzzozFfu5MGu2Ta6MREGLiL7YaI3NT6N9/Mw8PnPsRZyaSWblbxkWLyBbpuXZaNepZ7Nca/neHyNC2RQLoJuv2IHvveUAPv7QGZ+qROdK2oAgcQxULjP4MfrPnl/Ck2fjZ44mRVQ2CfG9dsJ41HWx0Wqqgp1jhVRzNrmN1h29tdo03LnP/Q+rSsL0cj1ynFS51oauKr73ViqbHtxGm6rYdB6zacqm6VhkxbXTVGC8iVNseqNRgMG+11bqbX5OJh3/lgT6jHixaVhYcB0NOzqKTSXV96Iq2GjJnTZoPduDgiw2BwwjUGw6hVdvJy/N2CzkNBzZN47zC6td1bIkcGVz5wgA+Kwbc2UqNuPTaEs5nRebocpmzGteqDb5RSIz5Mpm2CL2hssmkdHUrlbauJ7NrgFBIcrmei0+mm2Tz/4UmQwJNRg06DNo92kMCeDfcY6inynLZDfczO/InLuTnPQYxGHp4pzNND2baT6v5Vrbd/4H02hbIZtCAPCzr7se+ayG933+af58dO1Ka6M1LSYUm/0PPeo3f/21E/jgl4/35Xe1XLW/5Aa/Af5roqIoGC/5xzeI70+vG2WL1ZbTJ5tirm1TSMqlYyzXWihmdWQz6ob3RrcMC8u1VuT5Uq47MzZF276uSWWT4AFBKUaDbH4are1zQwHomKXpUzaHwEa7sOKtI2uttRWbojPMtAPKpmFyq75oowX8bWuMsa7Xbd6zWcjw3I1B3yTcLGSxOWCQIkijT3RVgc16m4FIxUYhq+PI/nEwAGf6EBJUbRpQFQUHd5QAeBewtmnxHpg4e0mtaaKU17F3wi02A0WIpkXbaBttE/WWye0Pwz76JMxGm89ouO7gRIJis5uymWz0CT1+vQZUt8xw2yi/AQ7obuv0ch2z5QamRnI9jyG5tFTDR791yrcQFBerUfQ3IGjzR5/MVdzrQsLXU3XHnkyUsl3TaGn0SfAa+X8+/wz+08ce7fpchuUoU+Kx0fWrHVA2g5/b5EgOP/E91+CJFxZxdq7qvkb3Gq6rfIGf5NRxlE3n54eh2KzU2r5RAmuh6fYxl/Ke+hZ8rydKWV8arfj+9Po9WVp1Ni6zupb4+9EQnAm0wFxebTnKpqZueEgIzWWOWuRWam1uQyYyuhx9QrTXYqNdp3tmN4IbxoCjbPpstKbFN8dGChnoqjLQNloxrGs1QqVPwrMXlvHLf/kAn/XOlc2CU2w2XButqnjBSoSj+Duf6QPPz+Dt//OLsY6H1YYTdpnTVUHZHOzr9mYhi80Bw7PRej2bQG8LXdrZIRst4M1yWwurTQMjeZ0PEqYL3MJKE3SUUWosY8ztuRRstIE5kbqqRl7E6YI0NepcJKhIG7akTaJthismt16xA6emK7F2krhiM5/V0TKsjvPGsh1rtVhs8sJmDUFUcTQjlDzHdqZieUCHTT/hWgTvvHo3gHQbGqZl4yPfPImff/838KGvHvelloppllGQYt+PApFUv820ms+7C+KkRQEpmztG8745m6Hneibc3XByuoJLS7Xuz+UuMs0QZZOrwkbnphBxxS7H4UFJj7yfKq2N1rZ5/D7NbxxkVhrtvmzyMcZ8PZtE8JoxXsz6EkObPmWzdxvt1EgOWT25bbnZ7lQ2l2tusbkJPZtUbEY9b7nW9vVrAptvoz2/sJrou7kR9FZsbq6Ntm12KptTI3nU2ya/3oubmqqiYGIkN9A22gWx2OzRRmvZDP/nc08D8K7H9BmNij2b1SYmSrmQFi4FlvvzFxZrWGkYkZv2AK2FM1AUha/hNsNGPwzIYnPAEPt9xP+3eriocWUz5xSGO0fzfenbrDYMjBayHU3nM66F1kn0Cl9ctQxHpfDZaDtGn0Rbh8n+sIMrm/6F5mNnFjZt9lUv8EVsxv9VPHp4JxiA756LVjfjAoIorTeoPAQj0wGv2O/lHEtCq23xXT8RRVEwEZJoPCg88cIipkZyuGqvE2SUZhH5hSfO42++dgJ3XbMHOV3F8xe9uanOYjVhz+YaF66WzfjNbzMXl2SjTVo8U/z9ztE82oagbEbYaAF0FGdLq61Er5nOvzAbLR99EmJ3J+gaRI8XNwy1NdhoNyON1rTsRCFOjDGs1Nt9KYjbpg0GdPZsBhbTE6VcjLLZu412x2ge2UxyZbNpmHxGLl3XyrU2SrkMsinsuP2CchJibbTFgLKpqbBZ70U6YwyffvRczyOH/ujTT+G//9MTPT2233hzNpPfh7yAoM1Lo81onTZawFuPBds1Jks57iJijOF9n3s60TzvjUJUNnu10X7hifM8F8AIqM9iGu3CSrPDQgv4N2Ho+xT3Ham6wgsA3h87rPkh640sNgcMUpe80Se9z0Cstz1lE0DfQoJoN2c0YM2YrThjT3ZPFCIvwrTbVMx5NtpgsRQ3Z3PJvSB12mgtnJldwW/f/zCOnZ5fy8vbUIKjT4hrD0wgp6t4IsZKy4vNkAU4KWfBRNpgZLrz34r7b+tTpMelr06WBnO3tVJv4zsnZ3H7Vbt6Cuu5sFhDPqPht9/2EhzZP4HnL5YBOEXH0moLo4VM7OP7ZaP1Lcg31UabTtmsusrmzrE8TJvx60FUQBDgf61OsmEz0Wsuc2XTO/+9OZvxPZuA91kF1Q5H2UyXRkvfRXGcS71l4sf/8Et47Mz6X9f+1z8/hR95zxfxex8/hsdfiA7/abQtmDbry8KKiupuyuZEMYtKrc1V4qY7RxDobaOsZVhYbRqujVbtKSCI7j+WzVDM6XwUykYy544bi7LvOsqmv9j0HEG9XRPOzK7gf3/uaTx0onPObBIq9TZOXConUhM//LUT+NQjZ3t6niQYpgUFjnUzqVIZLGQ2mrZgkSWCszY7ik2hp7Nca+OfHz0XOid4s1hYafKCsFtAkGUzLKw0cfxiGd96bhqf/M4L+OCXn8dffeV53Hz5FIpZ3bseu2vRkhAQtFhtdiTRAv7pD/T4OPFitWlwey5XNmVAUCiy2BwwDNNfDKzFRttoOV+WgntjvHrfOC4s1tbcfF1ttDFSyHBrBu2WzZYbUBUFe8YLkf1/FGldymVQymfw9ldchbuv3eP7GT1mzuZCwEZLhUDbsrlVb62v79R0BV/97sU1/Y6ktGnwcmCXMqOpuPHyKT56I4y4OZvUUxQMhAqLTKdify3zXONomdGBOMG49kHh4w+eRrNt4YfvvrKnMSQLK04vmKIouP7ABE7PrDgbIjMrqNTbuOWKHbGP71cvsqhsb5ay2WibfPGQdDFH1lZyMNDrCFU2ddpY8V7rSsOAabNEr7ksKJs85Me9flHBR59D2HmcCSzcxZA3+mommrNp2dxlIPZsLqw0sLzawjMvrr8KcXauitFCBk+dXcT/93ePRP4c2Zz7oeLR64xLowUcZdOwbH5NaxkW7/Hs5dymhffUaA5ZLZki6YwTs1Bwj03sKyUb7UZv6swKGznB+2azbaJlWKE2WnpML9D3OY31VKTeMsEAPH4mPs34mfNLuP+bJ/HN56Z7ep5uUFsJrSeSvp621d1GayW8/vSCYXYqm92KzYlSlm/sTrsbFNU+jhhZKwvVJvZPFZHR1NjRJ+/55JO49w8+i5/84y/jlz/4AH7v44/hT7/wLP7x22cwWcrh3d9/E7IZ7/tM34nRYLEZomzqmtfLzJXNmA2F1YbBC+Qcb+eQymYY8V4uyYYTVJ7WMgNRDAgCgGv2O32bp6ZXcOuh+MVuHKtNA/snnXCgScEGOVuuY+dY3rWAhV+0SdmkRcJ9r72u42e0mDmbS6stFITUwqzQ20ZF71oXQJ965CwePD6D19x8YE2/JwlhqXLErYd24K++chzLqy1ukRHplkYLdA6GN0KKzX6G0YTRbFt8LmqQyVLWZzEdBJZWm/jUI2fxmpsP4Ipdozg374y1SHNeLa56N7PrDk7CeOgMTs+s8NCn26/aGfv4fs3ZFJXtzVI2aRMISNOzabjfc/c8dl9HnIov9sqQJSuZsultdpg2Q0YIKKNdaq8g6vyuBr8/RoiyaSfYLDQtxn+XWGzSNfPS8vr3uC2tNvHya/dix2geH/76CVi2HWrTp80ASmlWQj6XpNAmQT6j84Wb8+dAzyYfldRGKZ9By7Awks+g2jB62oylc2THSB7zlSYsm0W+XsKwnCCqPA8IChSbmzAfdVb8fpkWNGGsEqX3hgUEAb3fKymFc6XHYpPO6WNn5vHqm/aH/ozNGN7/hWcBdDp0+gW9/p2jBSxWW6jU2x3jMOIeF2ej/cg3T+LbJ2bxvn/7qv4crIBhdQYEeTZa57xumbav2Jwq5bC82obNGGaX3WKzMTh5CYvVJg5MlTBfafLzK4ynzi3iyj1jeNNtl2PnaB47RvPYOZbHWDHLNyOzutbR1kAK5PJqC6tNM/RzdiYhuBuNgWI1jNWmgcvcqQx8g3jAe+03C6lsDhjBOZtrCggSejYB8JCgU2u00q42POuAONtpttLAnomCEx8dpWxSsRlRfADOa49K/1xYafJgIsBvoxVTcddCrWWi1jQ3pPezHTMG4+ghpyB5KqJvs1saLYCO8SdhNtp1T6ONSV+dKOWwUm8PVJ/txx44DcNi+MnvOQKgt2KcgkcA4LoDEwCA5y+Wcez0PK7aO+Y7h8Po15xNXqSpyqbtuM4JkfZJC95qw5mlS5tJdB4ntdEuhfRhRiGO0zADCxSyRLZdq11QTQA6LYmijZaON6mNlvrXcxlvXiPt8s+4asR6YVo2yjVnse2d8+EHviLY3Na6uUevM5/VfJtS2RAbLeD11rUMi99Hevme0KgFstEC3c9P+j7xOZthyuYGOwjmhGIzuLlIGykTxT4rm2soNi3b5t/Vx87MR4ZnfempCzg5XUEpp69bAU/n7i53YzCpspnERnturooLwvzdfhKWRjtWzEJTFb4OCt53J0dysN1ea65s9nGe5VpZdEfajeT12DTaesvADZdN4c23X4GXXbMHV+8bx0Qp53O9ZASnAl2P8xkNuqrgohtMFVpshiibcRsKq02Dt8TkZEBQLLLYHDCCypNncezFRkvKpvMlmCjlsGssjxNrKDZtxpwvWN4rNuniNldpYM94EboaPRiXJ+TmonvWvAK783eIihHgtxtSqulah+rWWgYYnIvaetMKSZUjrt43hmJOj+zbjEvoJGWz0QoPCBIXzVw9X6eCr5uN1ma975D3m/lKA5899iLecPQg9k856n3axGOnX7DFe0J2jOaxe7yAY6fn8cz5Zdxx1a6uv8MLvlrbIovsp2OF7KbZaHtRNiv1NsYKGf7eUyJgyKnOzy1xoU293aI1NgqxZ5gXm4G+nZZhISuMMhEJbgz0bKO1bX9AkFtg08JrZrkR+dh+QNfxqZFchzU4SLUuzkFd4znqvs6c0LOZ09UOy/R4iQbTO8/dMm3+873cH6l43ztR6GpbNy0bJ6cr+MIT5wF411ex157StTdS2TQsG0uCJTB4/PReBXs21+KYArwNkJUeihVaAxzaPYrFaos7R3y/v2Xgr758HNcdmMDd1+5Zt7AsOr93jRcAJJ+1Sed83PWs0mijZdrrssnnpNH676mqovj6MluG5QvZmhS+PzPLg1VsNtsmVxtL+Uyksmkz5ozOixErAOeabPAeenedpCnIZzVcWHTOt11JbbQR1xY6FnJjZGVAUCzSRjtgmJYNTVX4jZZ2unuy0boXG9EW5IQElXs+vkbLhM08S8LkSA6VujNQenGliT0TBVxYrEV+QamfMl7ZpNfMkA382GK1iRsOTvI/a6qzKGmbNl80rlnZdBd3lLq7nrSNzkZ/QlNV3BzTtxmXRuspmxEBQbpYbPZ+jiWBZuiFIQ6bDrMKJ+Gh4zMYLWRx0+VTPR8j8ZFvnQIA/OtXHeF/l9ZyttIwYFg2dgg3s+sOTOAbzzp9RzRKJY5+WZvp8x8tRN/A15u5ijPTzGZpAoIMjBWzQuiCW2yGjT7RO5VNMdnQsDoXZiKishncDaeNq6AlTSSjaYHHOt9LXQwIinx2B8tmsBl8xaZno3U+t+VayzfjMYxG28Q7/vTr+NW3Hk3dKrG46vXDz1eoWLcAdG4MrjTEVFgbo4VUT+WDXmc+o3kz60LeayqYKNCp2fYWnb1cuy4t1zA5kkM+q3MVlZTsaqONY6cXcPxSGccvlnFqpsI/38lSDle7CdXicZayOrK6tqFptAsrTdgMuGzHCBZWmh2FLt0TI3s2e7XRNnpXNsnd9Krr9+HsXBXHTs/j0O5R38987FunsVxr4T/92B344lMXek697QatFWhjMGkiLe8HjDnv6L2pNU1kR+JHXaUlLI0W8NqaTLd/NxtQNgHH9eEpm4OxyUtZHDvH8hgtZHzXZJFG2+n1Fe32YYip0KRMZjQV+ayOS0vOaw9XNhXhmh9fbNaa/mPJhdyHJB5S2RwwDMv2hcXwsRQ9KpuFQFF3zf4JXFqq97zwpJ0wT9nMw2bAT/3JV8AA7BkvuL73CGXTLeSK+ZhiUw3fdQ0qRkTW3cXqV88m7bxuxOI8TtkEnL7Ni0s1nzpExNtoXWUzOPqEW/y8m5CnbK5TQFCMjZZugMtrSKT9iy89h9//+GMdYUhpubRUw788cR5vuu0y7B73Vs9pw3oWV7xeMIKstMWczv87jv4FBHnF5qYpmysN7BjNO46HxAW7a6PNkLIZExBEyqawIBWHl3d73cshyiY9xpuzaXXYOolIG63uzdns1rMZnK+c1x0bLe2eE6RIRL6W1RbmV5p4YW4l9ufC4GOlRvL8OKI+r5W6aKNdq/ruHytTzGdCi83xIvVsCjZa9z7US7E5vVTHfnf8VjZQfL3nk0/iv/3T4/jMsXOAArz59ivwWz/0Enzo3a/BR/7963DVXqclRbx2F/M0+mTjFptzbgI89Y11FpvhyuZabfp0b+wlIIjWAFfsGsHlO0dwLJCyfGmphn96+AV87y0Hce2BCeTXcQwQvf6p0RwUJHs9lu0FMRkx32v6jvRrHUEFJEBptJ3rBmpragW+U4Bwr11tDpyyuShMGSjlojdGawnWjwDcTZ+AsqmqyAs297A0Wl1V+TqIis6o4EQ6Rp5GGzHvWeIgi80BI7hjtSZlM2QXfK19m9SrQT71l1+3Bz/4ssN4+XV7ce8dV+ClR3Y7czIjezYNKEDs7rxGSlvgS84Vo5Bis2XafevZpKJlIy7E7RiLKeD1bT4Zom7GBQTlI2y0dCEUb1Recb9+PZtdlc01FJsrdQPLtRY+8e0zPf8OwAl00FQFP/aKq31/n9X8cxS7QQqRqGxe76rxLzm8syN5OAxNVaBg7aE+VKSNFbKbdhOcqzSwa7zgWJQSbmis1A3XRuu3yIYqmzEBQUD397BSa/NrLn0HvDmbno02bFYsICrf/uTajOY5VLrZaIO9+vSa2qbtS9fu1rdJ71MvgSo8MMfXsxlRbPqUzf4Um7QhNZLTI+eZlnI6KnUn5KRl2lxV6KUFYHq5jn2TZJX3LxSXVlu45Yop/OOvfx/+130vx8+/4Qbcc+N+7J0s+qzUwTTanKtsdrNu9wsKB7rcLTaD975Kve3ae/3vp/j5lmstfPjrJxJZvYlqs/d7JB/JlsvgpUd246mzi77v6we+9Bw0VcH/89prATjvcdvsTNrtB/R55zM6RguZRMWmeD2JWpcxxvjv6lex+f999BG897PfdY4hQtmcGs1jabXZ8Z0CvHvt/EoTCytNZHUVTcMaCNun2D89UshEptHS34/EtGEBrgAh9N8rcO6ptO4cyet8jSTi2Gj9SeRR5x19riS86KoCVZHKZhSy2BwwDCGREBD6FxPeCP7+gdM47s70q7dMPmOToGKz13mbdHMZEZTNn3/DDfjlN9+Md73pJkyUcs6czJg5m8WcHqpQELTgChasi3zsSbDYdHaxyn1SNmlxtxHKZruLsnlozyjGCpnQvs1kAUFBG63zZ3/P5vrZaOstE6bNogOCRpwd917HnzjKjwFVAT7+0BnfoiUNL85X8ZXvXsS9dx7q2MwIFhPdWOApl5517aq9Y7h2/wTecPRgot+hKIrTd9KnOZujxUzf0mgfOj6Dd/3FNxPfVOdXmtgzXnD7aLofg2U7n+lYIct7jqhoDjvX4wKCgPhNAsYYyrUW73nzFiikIDiFQ5yNVlUUX586LW5UxSs2uxUfVCyJAUH0mmpNg39fL3VRNukcDQaDJWGp2oSqKBgvZZERUr7D6GtAUECFKUUom4CTqlqutflz9mqjbRkWFqpN7CNlM/Adb7ZNjBdzXTeGfGm0Wb1vwV5JmS03oAA4sMMpmpuBvILl1VaHqgn4x/U8eHwW93/jJM6nCLOpCcpmkp7oH/9fX+JrDm/8mY7vv+1yWDbDZ4+9CAB4/IUFPHh8Fm9/5dX8Osw3k9ZhEU99fVldxbg7x7Ub4vkeZaOttUxevMeN8Uh8nJaN755bwtm5Kj+GUGWzlEOl1uYb5mE9xc9fLIMBuHKPYwXfrPYKEXJV7BzLYyTnBASFnVfBaQZRiAFBls28TTz3+xqVOKwLkxBaAWU0SHAtrCiKT1GV+JHF5oARTBnjNtoEN6+2aeGDX3ke//KkE2LQaHfaaMeKWeyZKODEpd6KTb6bE9PLqKlq5Be07habcXClLbCjtMStFoH+E11FvWnyII21LH4My+aP3xAbbYzqBzgL1lsO7cCTZxc6Lr5xxaauqchoakevS2xA0DoskL72zCUAiOwfK2adG2C5R2WT5rW9+fYrYFk2/vbrJ3r6PX/7jZPIZTT8yN1Xdvxb2gUk2WjFTZGsruGPf/YVeNmRPVEP60DsO+kVUrjGClnYjPVlluoz55dxemYFD5+c6/qzNmOYrzSwa6yAjJaseF5tOgFdo8UML3p4z2aMjTZYbNL3Iq7AXW06myG7AsWmeJxtN+QjblMoo/vTDzNumBAffdJlrzCobPqKzZbpLMLyelcbLY1qCW4yJWFxtYWpUSfVkSu9EQcu9uq11hwQ5E94vXLPGK7YNRr6sxPFHCo1zybIlc2Uroxp933cP+UUm5mAstloWyjkuvfZ6UK+QklQZHv53n7j2enUBdVcxbGoU9EdXOiW662OJFqgU9kEvIV8EujeaFh2V4vr+YVVLNdaOD3jrDm8kEAd+6dKeOmR3fj0sXNotE382Reexd6JAn7orsP88fkMjfHqf9+mmGEwXsolUjbFzzYqrVksWuOSVZNydq4Kw7L58YWl0QKOHZjBSygW1xaKGyD03AVn1BiNwhsEK+1CtYlSTkchq2OkkIHtzrMNwjcquvRsihubTmua257gXmN2joU3meuaN+M9qbIp9o+KKeISP7LYHDBM4YsBCDbaBBYSusCQJaHRtrjCJXJk3zhOzayt2Ixr0NY1JXJRW2safEZm5OMjrMMLkcqmirkVr6dxLTtLYt/fRtlow+wwIrce2oH5lSZfIBF0EQyzFgLOzTxy9InPRtv7eJ04GGP4zLFzOLx7lNtIg9ANcDlhCmAQ+oyO7B/Hm++4Al944jzOzVdT/Y4zMyv4xrPT+FcvPdwRpAGkD9NYdNWEbp9rNzLa2sco0OdPtvd+qJukQn/1uxcT/axpM+xOoWxSIeNXNr0RLkFowU8bKzZjWKo2OwrIMGihTYsPuuaIm3stw+q6KSTOdRP77ulwu6k/wZFEYtjEatNAKadj70Sxq422ZXjKXFqWVlt8XI+nfIVfS52eWuec6mWu3LMXlvGD//3zWFjptPz9uzffjF//V0dDHzfhKpv0Ogs5HQrSb5TxYpPbaP090o12pysoDEVR+OxVUo6A9CrchcVV/MEnHsPDJ2ZTPW6u0sDu8ULohgvg9GyGKpvCNY36OtMocOK9sVuBRr+fHlMTik0A+FcvO4xKvY3/76OP4Ox8Ff/29Tf4bL90XkRZwxlj+MN/fgqPnVlIfPyE2FYyXswmLDa944g67yoNsdhc+zqC2p7o2muYFt8gESGrLJ3fwWvWpFBQH9k3AWBAik23rx/wCsmw921VUMXjEBVGMeWbNi52jIaHEeqa51DxlNFkPZvO86pyzmYEstgcMIJe/DRzNmfLTsFFVsIwGy3gFJvTy/WeksjoMaOFmGKzm7LZxQIRFYpEVoupQGppVtf4awfWpmyKN9zqBtlo4xaxQHTfZlwaLeDs4kWm0YYom/22fh2/VMbpmRW85Y4rYoe+U4JeUhaEuY1i38S/ftUR5LM6Pvjl51Md5998/QRG8jreFqJqAl4/V+Jis9r0hQP1StLiLI6mO7JjLYpLEOqvfeTUXNeFCm2A7RrPu9eF7s9P1xixZzMuIEhRnEh72lFeqbdh2gx7JxzVKu49pGIzWJiKm3st0+KjT6JwbFtku/Ku4UrCnk1LSLAF/ONcKF5/72SxY8MpiGej7a1nk87bbj2b1YbBAzZ62cl/7vwyGm0Lp2YqaBnOhlvUdUxkvJRDud7yxqXomk+NSMqlZWfWXpiNljGGRtsM7ekKg87RfFZPfa0g6HvUSLlQnS3XnWKTz/gL9GzW2nxkjIjo1qi434E0RVGtZfD7cLdrAM36pE2kOrdCOmuIo4d24NCuUXz33BKOHtqBl1/rd3+QGhWloJ6ZXcEXnjiPL7qOrjS0BadP8mJT6NlMoPz3o9gkC3LTsNBsmzE9m26xWY4uNgHnfL98l9PnOwiJtItVr5VhNKbYpI2KJGm03uYf4245rmxG2Wg1lReXzW7KZiAs03leqWxGIYvNAaOj2Exho6WwACrKwmy0gLej1Uvf5mrD6R+KK5A0wfcepNbqPiMpStlcrDYxLoxDIDK6ym8SqrI2ZVO0Eq1ukLIZN5YBAA7uKGFqJNfRt0kJlxHCJgqZEGUzMMcVEDY0+hwQ9JljL6KQ1fCamw7E/txkKZc4IOj0TAU/+cdfxvFLZQBCOnIhi/FiFm9/xdV4+ORcaKBSGMcvlfHtE7N4211XRt7A0o4+Waw2feFAvZLR1EhlKSlNw0Ihq/dtlArgKGC7xwswbYZvPTcd+7OUorx7rJC4B5VU7vFSjqfRkqoRpeLnhMRK6tfcPeGolfHKpvNcuwLKpngdbht2100hv23L67v35mxGPtR3jKE9my0DpXwG+ydLmC03Ygsrrmz2YDlcqjb5jn+maxptW5jtmP4cpcXwxaUamobF1cFuTBSzWKm3eVGWy2g+NSLx8y/XMZLX+aapGBDkpAAjkbJJx1DIatBUxVM2U74ndJ1O+14urbawYywfqmzajKFSb2GiGKNsWjYfJVNLaPdkjGG1YfA5xEmVzRWubBpQFYW7FhRFwY++4ipkdRU//4YbOjYmPWUz/PgeOu6owb2sZwyubGoYd8+tbhsX4rkWtc7pd7EpOtEWqy3YDOE9m+5m0fSSs5kSzEqgRNq9E0WMu61Qg6BsLlabiZRNEgO6tWL5ejYFt2Chm41WVWBazOnTN7zNwzBWmwZ0VfHdF3IZTSqbEchic8AwA6NPUtlo3Rv4cq0Fw7IjZ7KtJSSo2jR8toEwdE0FQ/iOUK3V3UZLabSdymYztLFbvOjuHCusSb3x22jXf8evZdhdF1qKouDWQzvw1NlFnx3Psp2ZrFGqYSEXomyG9GwqbsBJP0efVBttfP2ZS3jtzQe63hgmRnKJezZPzzojHS4tOjfTVfczokLxrS89hF1jeXzgS88lSld8+MQsVAV460sPR/4MLzoSLgSjztO09CMgqNk2kctoPSsuYSyvtnDHVbtwYKqErzwdb6Wdd1XoXePJezapL3HvRFFIo41WNgFnUUWLA+rt5spmXLFZ94IpxJ81LZufty3TQsuMTqMFaGOgs0eIiuOuAUHBNNqMZ8ekIeZ7J4swLJu/vjDEnsM0tE0LKw2DtyjwgKCQ965tWmgaFi/Qe7nekkI7vVRD0zC5va0b46UsbOa5G/JZzadGpHn+fZMlfu0UlU0qavIhLShh5HSV39OCqbZJoftOmseZbr5AKad7A+WFgKBqw4DNOseeAFE9m8mKjpZhwbQZ73ftNmuTlFNR2Szldd9967U3H8Df/4fX47AbWiNCCnOUsvmQaz2+uFhL/BqINgXm6SrGS1kwdL/vJ7LRuq+1kNXWHBBkWjZemK3yxGFqGQpTNumzpu9XsCAlZXPPRIFvtGx2sWnZtrNp4l576F4e9r7VWiayutp1g17MOxDbGrrbaB33jXjdixpbRWth8Tx2JiPIYjMMWWwOGB0BQWlstMIsxsVqM7Jnc7SQwb7JIk72EBK02jB8toEwvGPuvBDXmmkCgjqVzbCLhHjh2T1eWGPPpnOBy2W0DUqjDe+9CHL08E4s11o4N+8lBlo2C+1hIwrZZD2bgKOg9zMg6EtPXUTbtPHm267o+rOTJScFMMmCcdodyEy78cFRPLmMhp9+9bU4OV3B191wojguLNawd7IYe05qqj9tNA5n8db2JdH2Sr8CgvIZLXHBbNkM33h2OrJQt2wbK/U2JkdyeO3NB/D0uaXQGbAEt8zl9OTFZtlTnWix1OimbOpesbnoKpuJbLSrbSjw0gnFWZmUeNgyLLQNO3LOJtAZta932GjjXzNd68IDghxlk17PdEzfJr0HwZFH3SA1mN6HuD5lmh9IBXovKaG0oXBxqe5uuCUr7CjsZlawCca1bURxaanGLbSA3yrPA2xSKJt0/Qim2iaFzu80m0H0mEJW5xshYkFGRWRoH7qQRuv1bCY7Z+iaS/2uK12KM7pW08/VIkICo2zLcT2bc5UGTs+s4JYrpsAAnJ5JN1/WCNhoge5KbRIbbaXujFPaOZpfc0DQuXknHOiOq3cB8NwiYcpmVtcwWsjw3u4OG617X9rn3vNURdl0G+3yqjPKiFoZqNgMa2NKkvkBOJtlBu/Z9NZJ9H5E2WhpHSRe06I+49WG0eGGyrljeiSdyGJzwOjs2Uxhoy03+M7/RVf5iZpneWTfOE70YqNNoGwGZ9aJ1JPYaPkojoCyKex+idBA7lJORymnr7Fn07kx7BkvpLpJPHt+Cb/ywQdSLTIYY449L6YXjLj58innedwkOcAZhxNXbOYzWkex2Q5RNp0/Kz3NqguDgoGuPziBK/d27lQHmRxxEvSS9MvQTTTYZyRe9F9z8wFcuWcMH/rq8a6fx/mFVRzcMdL1eTMJCz/qPe2bjXatAUGujTYr9GjF8diZefzBJx6LDNso19pgcPqmX3PTfjB4icNhtNwUVz7KJcF7OFOuY49bWGU0FQoEG23E6e6z0brK3+7xBDbaegtjxSxf0NI1x7AYX9S03Z7NuO9pRxptyoAgk/ds+hdF9ZaJRttCKafz4iiub5N21RspbbRLgXE9cbZrKhpowRb8jtVbJn7lgw/ghGt1D2LZjBeLl5ZraLbNyNFIQUi5oY3VvK5B06LbNsKf38ZcpYH9vmLTKxLpmhnWghJGVig2e+2NpudMV2y6x+mqu5qq+D4LKiLjlM2WYXHFMenmKt0j900WoSCJsukPCHLWAN0LBiKuZ/Pbrqr5b+65BkB6t5bfRuuc+92LTW9cSpyNdryYxUghs+ZNa3pNd169G4DXBx+1ST01kuMbEVHF5t4JZ17saCGzIdkUcYjzfQFR2ey8htVcVbwbWV2FzZzvurj5N1HKQlMVfm8IklGddVBLcAjEpdEGi82srso5mxHIYnPACBabvPBKpGzWcZW7uD+/6ChgUYrNkX3jmKs0Ei3wRapJlE0tXNlsmxYMy+4aW00FtnghNy0b5dVWaPAK7dJOlHJOCtkavuxkw9kzUUi14/fA8Vk8f7GcKuiGFnLdLCGAc2MfK2Tw/EWh2OyibDpptJ0BQarSmeqpJQxvScJT5xZxYbGWSNUEPGtPkveOFtpc2QzpIdZUBT/3vddhttzAPz96LvJ32Yzh4lINl7kz6uIQ00bj8GZsDkhAUNtEPqsl7julYv5UxKKNPqPJUg77p0q4dv8Evvp0dLEp9iQnLZ6nl+vY5xabzuwylSutUed7PutXNkcLGb6pFfce0hzC4Pgfy6ds2o6Ntoc0Wq5sdrl+R40+IcVxJJ/B7vE8VEWJLTbbPY4+8cLXyEYbU2y63z1P2fT/zMnpCp6/WOZ91UHmVxowbYYdoznMVxpYbRrJlc0SKZveaIeMpqbaKJurNGHZDPumvO+9mEbrKYbJjun7b7scb77dudbx71nKe5Bno03+OHoMbSjndL+qQhty42GjT9zjXFptgd65pHZPcfzZaCHTtdgkhZU2Keoto6u7ScQbfRJebB6YKuHmK3Zg11g+tVuLb77qKu9t7TZrk64nxZwea6MdK2Yxkl97sXlquoJiVscNbqI72WizITZawCsogc6eTSqyLnMtuaOFzKbbaBcCxSadG2GZGasJlU3x+2zajLejvf7Wg3jvz70ycnSf5oaNiT3vcWm0waDMnJyzGYksNgcMw4xQNrvcTNumhaVqi4+YuOAOaI5TNoH0O4HVkC9YEG6DDSiT9UDkeeTjeYHtfcmXa85NMUwxosXs5EgOuYyK9hqKJjpGR9lMfhE+49p30uxo0yItyUJLURRce2ACxy+W+d85xWb0Vzif1TvnbLqbGcE+T11TUlvRovj0oy9itJDBq27Yl+jnD+4YgQLgdz7yHXz4ayf4TmcYtNCmBUHYBR8AbrtyF26/ahf+7punIm+mc5UG2qadTNkU0kbjoBmbUTadNPRjQHTTcGy0SZXN+Ypz/JHFprtwpAXNa27ejzOzK3zYeBCxJ9nph4k/x2zGMFtuYK+gOok7+FE9m7lAz+bUSC7RfNRyvY2JUo5fcwzLBmMMpu0pm/WWCctmsZtC4vlh+JTN3uZs0iJxcdX5PIp5HZqqYs9EIXbWJimbaUef0PNMjSZRNp3v02TJmckZPEdp9FCjFX7u0vHfftUu2Mz5+aT9kWR1FG20cYF0YVxa8ifRAv4QME8xTFYQfe8tB/H6Ww86x8NTYXvs2UzxOmhBzItNQd0HvA25MGWTzjPRAp90zqY3zF7HWDHLz4coxI1BxliikEARrmwGzunVpoGnzi7ibje99si+8fTKppDOPl5KZ6MtZPXI69lKo42xYqYvxebJ6RVctW+M27WpXzlszibgbRgBnWuLI/vG8T/vuxt3XOVYckcLma426PVmseq/Z2qqglJOx2pI/209obLptY34lc2sruHKkL5ggopSMbsjlbIpbbSRyGJzwDDtqICg+BN4YaUJBuDqfWPIaCrOd7HRXk3FZsTucxRhPvUgGrfB+o+ZbBHd02jJOux9yZfcnfewXjhaSE+Wcsjomi8kIS21lomMpmJyJI9G20q0iGGM8YHVaWyPoh0nCdcdmMSL86tcfe3as5kJDwgKu0lRY/xaWVpt4sHjM3j9rQcTqxWHdo/iv/7ky3Dt/gl85Jsn8VN/8hX8t398HM9dWPbZD2stgy8EKNgl7nz8sVdchdWmgWfOL4X++wX3O0K7vHEkVRlp0T4oNtpm20JeSKPtdiOcd3fNKYgpCFc23e/hPTfsh6oAX4sICupQNrsUz0tVJ9xMLASoWFUVRIZhiT2bFDaRJIG3XHPSOsWfJZWMrlPVZtt3HGFkhfPDtBj/jiW20brPmQkom7QQG3EL370T8eNPvDmbVqKALGKp2oKuKhhzizk+DinkfOGjaYpZZ3Mv8DMvusVmPaLgpeO/40pnwbvaTG6jHStmoECw0boW0jTXruCMTcDZ1CUbqtgLmZbeezbTBwTV3WKe94tm/DP+yrUWVAWhKo7qhsJRgFdGU1PYaL3WhbFCNlbZNC0b1YajZFo2Q71lOuPPUimb4Tbaf37kLEyb4Z4b9wNw1jQXl9KFBLVNC7qqQBPO/aQ22jhlc6VuYLyQRSmfWVNAULNt4szsChcHJkpZbqONWjfQSBpNVXxrSeLGy6b4dXS0kN18ZXOlCV1VeLEPOIm0a1M2yc5udYRuxkE/V0tQbFbDejaljTYSWWwOGB0BQWRJ7aIIzLi2oj3jRewcy+P8Atlow2/iI/kMDkyVUu0EWraNetuMtCAQXsCP/5jpJlDscrHgljbh8WS1mIpJo50ccYbAr3X0SUmIxE9yA16oNvnubhrbY1voF0nCtQcmwACccK1ClEYbRSGnwwgkqzmqS+fz6Wo6K1oUX3j8PCyb4ftvuzzV444e3on//PY78ZfvejV+4M5D+M6pOfz7v3oQ/+4vH8BjZ+YBeIpIVle9QeExPcQHXJtclFJK6v/BBDZaMW00jsWVJjKairEu6n8S+hIQZDgL+YwQgBIHLT6nl+uh5/6SYKMFnKLz6OGd+Oozl0ILqrZp8+9nknRdsvFSzybgfT+iVE3Ar+osrDQxNZL3haBEUa61MTGSE/rMbb6ApAUxLcZiR58IGwOGGRYQlMxGS9/njKZCVbz3m1oP9k0W+XsUBimbDN2tnKZl41OPnMV8pYGl1SamRvP8PU5iox1156AGF1cUYhbsFyeml+vQVQW3HNrB/y7pxpSmqhgrZrny4AQEpes3v7hUQ05XuYrLj8G1oYq9kGnpNY12LT2bVIzlAp9FueZYOaPuERldxULVWTfsnyr2YKPNYKzLbEr6N0pSXWkYiXIbRHRNha4qvoCg1aaBT3z7DF52ZDcvxK5x///UdPKQIMPy1lsZTUUpp6NSj2/naIs22ggRoFJvYbyUxUhOx2rT7LrZFMUDz8/AsGzcfY2j3k6UclyNDkujBbyNwCTfqdH8YNhoxWsP4BxX1JzNkSTKpnDtN23GnSvd4MWm8NxhGwo2Y6hFKpuy2AxjU4tNRVHeqCjKcUVRTimK8psh/64oivIn7r8/pSjKbUkfO6yYFgsom+HJrEHmKrRIK2DnaJ4vUuJ2Z69OaTuhwJxuX/aoOZneMOf4x3tzH73HB60WIrSQ5j2baxx9UszpXiJaggvxGSEBL82FhhYGSQKCAODa/RMAwK20doI0WsCf4meYVoSymc6KFoZlM3zu8fM4enhHImtqGPunSvj5N9yA+3/ldXjXm25EtdHGH3ziMVg244rEkX3jfBETl4484dr8qIAKcmFxFSP5DLfmxZHVk9mznRtnLlKBS0NfAoLcRGovXTT+/JyrNHghGZbsuFxroZjTfQuZ1950ALPlBp4TwquItD2b9Bnv8xWbrkoYG4al8uTWhWoTB3eU+KZK1HO2TQv1lomJotezaQjFZilwDYgdfRIICKJrYFIbrWjnA5wiNZfReHAPLc73TRZRqbcj1RvR1dGtb/PP/uVZ/J/PP4P/8NcP4dTMCldExOMITaNtGChknXE6ucDiijEm2Giji809E0VMlHL8OptU2QQ8Ky2pc2mVzdMzFRzaPdaxeeG4Fyx+3GnUN6JXZZPujWnm6lKxKYYTiWMXHNU+OhVb11TuGDowVcJqjI324ZOz+IsvPgvASwkt5jIYK8bbMKnd4TJebLbde2y6zbh81m8R/qdvv4DVpomfevU1/O+u5sVm8jWNsxnmnXvjbjJ6HLxnM8JGa1o2VpsmxgpOQJDNWOoeauJLT13E3okCbnQDAieKWW7RjgsIAuKvV8RoIVxB3EgWq82OdV0pr0fO2Sx2cdYB/k0f53qcVNlMZqOtt0wwoGOj29nwkTbaMDat2FQURQPwPgBvAnADgB9XFOWGwI+9CcAR93/vAPCnKR47lBiWFejZTDb6hJJod43lfYmtccXmNfvGMb/STDzjkOxT3ZRNTQvvMyVrQvc02k5lc7HahBawWhCespnj6kkaC5lI3Z0DmkbZFC2HaQpd+t25hDvoowVHjX7eLTa792w6v1dUGKiHL4jeh4CgR0/NYa7SwFtuTxYMFEchq+PeOw7hp19zLVabJk7NVHghcv3BSdRbJtqmFdtDrKkKpkZzkcrm+cUaDu4oJSoMs7qWSHVYikhM7oW1BgQxxvjoE74IjvmMLZthsdrEy65xUg/FQeJEebXFi1Hi5dftRVZX8ZWQoKCW4QXrZBL0Bc+U61AA7J7w0gITKZtZHU3DwoUFxxp9xa6RrsomT+scyQkBQYwXftxG6173Yns2AwFB9NxKUhstT6P1vs+5jOYLCAK8cS6zy+HjZkRlK0pZBIDPHDuHf370HF5z037UWybOzlV9562mKlCV8A2olXqb3wOc9EXvZ8q1Nnd5RC2wZ8p13pNL7oOkyibg9SDmM5ozIzhFsckYw5nZFR6kJ0JOAq4Y9mKjpTTalAvOeg822mBvqdi3DFA/cvS9OuPOw1YVYO9kEfWYe93HHzqDT3z7BTTbJlabji1WUxXXRmtEnt/Ur3mZu/m4uNKEIcywTUo+o/Me1ZV6G//08At45XV7cdXecf4zE6WcExKUqtj0r7fGuyi19BjAKfLDNrJoc2qsmOUbVr1YaecrDTzxwgK+95aD/Nonhv9069nsNr8bcNZy9bbZ17FnaVlY6ZxLPZLPdKTRtk0LbdPGSIJzR2yLsKz4TXkRKkq72WipQA9udGfXuP7cymymsvlSAKcYY2cYY20AHwXw1sDPvBXA3zCHbwOYUBRlX8LHDiWdNtrO/sUwZisN7BzLQ1NVX79Y3EX9yP50IUFhYybCyET1bLq78d0892F9qovVFqZGcqELTrFnM5fQLhgFDVBPo2yKClAaJeo7J+egqQpPmUvCdW5IEGOsexqtuwgRF53i4l9Ed1PY1sJnHnsRUyM53OVafvrB0UM7AQBPvrCI6eU6xgoZ7HcXqJV621E2YzY/do7meaBCkAuLq4n6NQFSrrrvTi+sNPuSRAskL3CjoJtePqsnspQur7Zg2QxH9o1j52g+VCFYqrV8ChjgXGPuumYPvvnsdMd33m+j7f56Zpbr2DmW9y0AEymbroWQVLXLd4527dnkcwiLOV5c0eIEcGzoCrxAnLjFm9iPGh4Q1K1n0w0IEl6jOLON3CBUpF1aroX+HlHZiio2nzy7gP/z+Wfw0iO78atvPYr/9pMvw1gh0/FdiLI9rzTa3CYeTF+k9z/u+aeXvRmX+6ec/0+nbPptgnqKsU2zlQZWm2ZEsUk2WqcAibIpxtFzz6bbf5kmIIgeQ3bfoO2+UmtjPGTGJkHXhPFiDqP5DFqmHXrctaaBZ150XAsvLqz67IPjxSwMyw5NinWOwfmOXb7LObdoRmzaYjOX0dByNy++fWIW9baJH3vl1R0/1y0kiDGGjz90mgeaBddb48UcL5CjaJtOonsuo4UmlVKxOl7M8l7rXkKCvvzdi2BwAqgIUamOSqOdSmOjTbGpvh4w5mxw7hzrLDaDx+Q549Kk0VodEx7iCLPRhq2LqhFr4V5HH20HNrPYPADgvPDnC+7fJfmZJI8FACiK8g5FUR5VFOXR+fn5NR/0emNYDBlhwZE0IGi2XMceN9Z6V0Jlk264JxLGhfMUui79aFEJuvVmQhttSJ/qYrVz94vgabSlnC/yuheCNtokF+Ezsyu8JyXp8zLG8M3npnH08M6uSrHItQcmsFxrYa7SgGWz+AU4T/HzFgJRymbaWXVBZsp1PHJyDm98yWWJm/GTMDmSw6Fdo3ji7AKmlx1FhCLqF6tN1Ntm7ObHzrE8FlY6VaBay8BitZWoXxPwdiy7sbTaeePslTjbKWPerMIo6HPPu5ZHIH4zhMKBdo7lcdXesXAb7WrLt7tOvOamA6jU2x3zOZ05silstILqRSTt2QQcNTajqdg7WfAKyIiF/zIfep91j0/x9ThnNBVZXeXKZvzoEzEgyOvZpGPuttEdTKMF/DY4WpzTbMiovk3x+hOmLF5aquH3Pv4YDkyV8Bs/eBSaquDqfeP4m3/3Wp8lEYj+vFbqBg9TyQYCgqjYvGxHKTQgqNpoY7Vpcpv0/jUomzzlWFW7bsYSdE5fLShiBG0o1dtmT/2agPN5O8nEvfZsprPRUtEDOOdLMxAQ1E3ZBBzrKC3g6yFW2sfOLPDNkrPzVd/4s26hOuVAzyZds9L0bAJ+Gy19b8M2Co9QSFDEffu5C8v4wJeex5efugDAKTazvmIzgbJp2cjoFEzVed6tCAFaIz0Wc4wxfPHJC7j58inf9XBixPs8o5XN9MXmZvVt1lsmmobVYaMNm0+aNGAS8I8+sWzGRZtu0Hrbr2x2fpe58FLoVDaB9KOPtgObWWyGrRyC39yon0nyWOcvGftzxtgdjLE7du3alfIQNxbLZrAZ83nxk9po5yoN7HGtZ2JRFhcpX8plcHBH8pCgKOtAkLDRJYD3Be6W8hfWpxpXbE4Us46FeNwLBem1SbvWMlBMYaOtNQ1ML9dx3YGJVM97amYF08t1vOr6vamOj57n+MUyTKtLQJD7PtcTKJtpZ9UF+fxjL0JRgDe9JF0wUBJuPbwDz7y4hPMLq9g3WeJW6otummxcD/HO0TyfIShCj70sYW9pkgVkrWWg0bY6lL9eIWUpzKJ2crqCn37vV/HQ8ZnIx9PiLJ/xejbjzk8Kntg9VsDVe8dxYXG1Q7GIKjbvuHoXRvIZfDWQStsyLH4D1t1iLs5SOluu+/o1Ae8GHneuU+Fx4lIFB3eU+IZXXCgR9WaRLZjsmKZQbOYyGl+IdR994o5NEfruqT7uHhAUbqMFHIcCvZ5S3rk2RY0/aRkW33wJjoqotQz8p489CgD4Tz92h89hks/qET2MUcqmW/Dp/j7Bc/OrGMnrOLBjJLRnk/fkBmy0aSyrNGuT5i+msdGenqlAVYArdo92/BttKDmzadNbaImkLgiRXm20+azO2wDE/tm2aaHWMjss777jFAbdx22uPnJqDiP5DLK6inNzVScR1L3mjnUpVsq1FjRVwZ6JIhR4n3/ans1CVucF+UrDQE5XQzdNKSzoVMhGGQD808NnAXhrkXZA9RovOum6cdcow3SuaVGbMXRdGS+K72u6UUSnZ1ZwcamG193i11B8ymbE9aiY05HT1YTFpvM93qzxJ5SnEFzblXIZNA3/NAAahZJM2fQcbqZtc7ddN7w0WjEgqPNcqEUom2Sjb/W4/tzKbGaxeQHAZcKfDwIINv1E/UySxw4d3u6298VQFGd33oq5mRqWjcVqE3vGnRs42WgLWS1WDQDIdlJOdHxVIYUuDo3P2QwGBDnDu7spX2GjU5ZWm9gxGn7jfMX1e/Fn7/weTI3kuRrQq7JJabRJbbRn3H7N61wrbFLb47eem4aqKHj5temKzcN7xlDIanj8BWe3OVlAkL/YDFU2U86qEzEsG59/4jxeemQPdo0Xuj8gJUcP7UTLtLFQbWLfZJHfcGl0SayNdiyPetvsCFQ5nyKJFkhmAe3njE0gftYh9fL97ddPRi6MmkJPVzaBjZZu/LvGC7h63xhsBrwg9CM7ATzhC9iMpuJVN+zDg8dnffbJtmnxGzBPfI3Y1GgZFharrTUpm6dnVnD5Lq+QiFOkec8mKZs6FZvO8WmqgqxQbHZTNhmcTUGx7z6xjZYXuN5rpO9pMbCZsneiiEsxxSYF6IjKpmUz/Pd/egIXl2r43R++jSuKcUQtpqvuDEHATV80/Dbay3eNopDVQm20UcVmb8qma6NVla7OH+L0zAoO7hgJvQZSsm69ZfIWhF5wCvCUPZs8IChNsWn5FFhn9InzeF7wxCmb7jVhopjjalFw1qbNGB45PY/br9qFy3aO4Oz8qm++YFdls+b0jWqqgpFCBrNuan6SWYkieSFteqXexmhEqBsf6Rayppkt1/HA89O+12mEBARZNostDqk1QFfD12VUuPmLzXTK4XF3JN3Rwzt9fz8ZEuIVRFEUTI7kElnTN1vZ5MGPHTZa5/wQ37caD6jsXmyKm6vOpnzKgCD3uTRVgRVy7eZr4Y7RJ731bG8HNrPYfATAEUVRDiuKkgXwdgCfCvzMpwD8lJtKexeACmNsOuFjh45gIiGhawqMGNVpcaUJm3mhGrTYTbI7e83+CSxWW5EhKiLeMOeEymZgRyjpMGeubLqPbxoWVptmpLKpqSq31PTaMwM4N9aGe4y6pqKQ1breJCgc6Do3KTaJ1dKz0O7gN+ukZDQVLzuyBw88P4O22U3ZpIAgv402qmez12LzoednUK618ebb+69qAsAtV0zxeYX7J4t8sXmBK5vR5yOdM4uBvs0LizWoioJ9CRbdQDIL6KJbAPZjxiaA2D5LCgI5PbuCb5+YC318Q1A2NVWBgvjzc26lgWJWRymn88HXZ4Rik+xrYcomALz2pv1oGRa+fXyW/51jo/VUxqjXA3gWu70BZZPP2Yy5W5HKZVg2t+wB8SNrlmst5DMav07q7s8atqBs6poQEBTTsynYtgyL8aLRCwiKPnZAGH0SomyOBFSg/THjT9qmxb8fDcNbMH/oK8/jOyfn8AvfdyNuPbQz9LEdr0lTYQSu4ZbtJW0C/j5BJ4l2FVfsGkUxp4faeKnYpA2Fw3tG8dqbD+DWK3Z0/GwUVEzTeaFF2BnDOB0RDiS+lkbbQiFiZFgSKNU2KU6QFymb6Wy0oktILMjIvhqXRisqm1FBNqdnVrC82sKdV+/CoV2jODdXRa1pcvsgnQdRszbFRNyxQpaft2mL+XxG420BK3VPWQ8yUcph93gBJ0PGn3zq0XMAFOwczaPubj62A+ns47x4jg5NpARbLeT74TzWGw1ERRO9r5eWaonGoJy4VMFYIcNbo4LHB8Rfj+698xBe7c4fjWM0RTbFekDrzuDajm+UCxtZ9B4m6ff1jT4RrsfdEOds8r7ckHvmakRLGX0mUtnsZNOKTcaYCeDdAL4A4DkAf88Ye0ZRlHcqivJO98c+C+AMgFMA/gLAL8Y9doNfQt+hRVHQiz81kutYLIvQLgtdgKdGclCQ7IJ+JEVc+GrTibzvpkxGjWuh8J1u0O8n6zDF/ycJXsmsQdlstJ04a7qYdRt4vFht4tOPnsPO0Tz2TBbc5+1+kTkzu4JLS3W86vp9qY8RAL7nhn1YaRh47sJy7I5dIU1AkNo9KTSKTx87hz0TBdx+1frY1Ev5DI7smwDgLFKLOR26quDCoqNOxvUQ7xpzPpeFwGbKTLnu2K4T9nJkE1jj6Dvav4Cg6P5jKjZH8jo+8s1wdVPs2VQUxTeeI4x5N2RMURS+qBF7uChQJ8qad+PlUxgvZnFM6NtsGZ6y6SW+hh8DhYcElU36Tsed6+I5fcUuodiMVTb9PW20oSD2T2Z1lY8tiVc2vZ5YX8+mmlTZdG20gYAgIETZnCxittwIba1omTb/7Ojz/+rTF/EPD53BW26/HG+5I3lStBh6RIhJm4DfRrtca2G1aeCKnSMoZPXQ/r/p5TomSzl+bcrqGn79Xx3FgYQOA8Cz0fpSjhMom5V6GwsrzdB+TToWw02j7dbqEUcukNDbjZZh8XMsbjNopd7G333zJD+XgseZFcKayoF+5DC4simMoAkqeo+cmoMC4I6rduHQ7lEsVJtYrrU6lM0oG2al3ubq6mghIwRe9T76pNowYucYX713rEPZbLZNfP6xF/Gq6/fi4I4SV8kMy9+zOdFFqQW8BNuMpsJmrOO7vVJvo5jVkdU1bhdebRo4OV3Bz7zva3jq3GLX13viUhnX7J/oSEpPkkYLAG+760q8TggWioJcQdVNstEu8GLTf08hVVbMmyB3UhJlU0xfNy07cc+mxtNoDT7DNzSNtmlAU5UO9dgLCJLFZpDNVDbBGPssY+waxthVjLHfd//u/Yyx97v/zRhj73L//WbG2KNxjx12wkIiAGDfZCkyfRAQ7XLeom5iJJco5OCqvWNQAJxIUGxWG51DbMPw5mQGlU0j0U2GVBh6P+iCNJXAnki73b182XmAkXuDGMlnIi/Ci9UmfuPD38bCShO/8YNHU9l3nznvJPvdcXVvxdntV+1CIaslUDbTpdEmtaKJnF9YxVPnlvD9t13e1bK9Fo4edpSP/ZPOqJKJUg6XllwbbQJlM5hI2zbtVAmYYgAMgNBwHn7j7GNAEBC+AKWF14++/GqcnK7gkVOd4Wd0XcjzhX28Oju/0sRudyedCjxxlMLyaryyqSoKJkpZrhowxjrmbEa9HgDcYhfs2SRlNC69XjynkyqbjsXPb0szLebrnxR/b+ycTbJtGRYsm3XYaLuJGYY7m1NcXHJlM3B+750swrJZaPBVy7B4AikVe59//Dwu3zmCd37fjfEHEXxNIefLiqDaAO4mjFtYnZt3Nn8cG63uK9wJZ+zJ2qz2nrLpbUIkcWWcdkf5XBmrbFprLjbFoi8JdZ/tPPp1fPO5afz1107gnJukGrTRUnqxzZhgEe+ubDoBQX4FjvjOyTlcs38CE6UcT5S1bMbPyZF8BgoQOZuyXGvxz0t08aQffeKNdVlptGMdQdfsn8ClpbrvtZyYrqDWMvG6Ww6imPM2QtpmoGfTfb/iZm1SqJAekbpfqXs2c01VUMw5MyOfPOtsws2Uw8cWEU3Dwrn5VS4EiJTcjVYgXtlMSimvQ1U2T9lcWGlivJjt6D/Nc2XT+26kCwgSezaZbxMvDlJAa03TUa/V8CwLspIHNwOy/L4pbbRBNrXYlPiJstHunypy+1EYDUHBIHaN5RMVhoWsjst2juDZ850D2YOsNuPHTBBhczIBL+k1CeIoDu7rj+jZFMmuQdkMzgEdiRgsDAB/9OmnsLDSxO/965fi5it2uMmXyZ6XbnTjKS20RC6j8fEi8aEpfhstYyyyZ7NXG+1njp2Drir4vqOXdf/hNfC2u67Eb/zgUd7bMS4Mt47rIaYd06CyKRZBSRADgk7POOE8wWCtpdUmRvJ6qiK223MCzg3ziRcWeIIi4BWB33/75dg9XghVN8WAIOf3abHn51ylgV3u+6upCnRV8f089YnGBSA58/Cc53UCz5DYRju9XEcuo3WoMbxnMy55mSyVquLrR4yzPzvKpvda9ICymdEUX7GZjRt9ovvnswUDgrpZ5yzb7thkpOcOLq6oGA+7J7QNC6P5DFRF4ZsNy6stXL5zJHVKdNh7R2NguI024ymbpOzvGS/w63zQSju9XMe+yeQqZhjBgKAk81sBMYk2pti0XBvtWorNTLSD4LEzC/jGs9O+84HuByP5TOz3kzZjSEVstPxFsbjhmUjZJBttMcs3WMXe9kq9jeMXy7jT3RQ9JPRC09pCUxUUsnr06JO6t6EjqpGpR59kdX4+OzbaGGWThwR512famLx85whK+Qx/nc7oE6FnM5Gy6dhog+0+RKXe5uN5ABrjYfI1VpTlmDgzuwKbMVyzv7PYpI1WVYl3eiRFVRRnU32TRp9EBT+GJenXWgYUOCOpuiG2U4lOk26QAkrKpqYp4Wm0EcLLWsSOrY4sNgeIqGJz32QR1YYRufvEFYyM9yV81xtvws9+7/WJnveV1+/FEy8sxBa0gGO1iEv+JHQ+uiQYEJS82NSE0IcoX38YvHeqh+hpUmSK7kVktJCNLDbnK03cftUu3HT5FADnJpCkr895HtMdq9B7UfI9NzgW3LhiU1OdxTKP1bdsZ/EfZaPtIY324ZNzuPPq3bE76P1grJjFa27ykvnE4Iu4TZWsrmG8mO1QNtPM3qLfQ0mqc25qa1DdXFhp8oHa/cDbOLHw8YfO4G+/cZL/Gy3uijkdb3/l1Xj+YtlnXxV/hhakUemigFO8VuptX8CTWEgAnrIZt0ki2t3osUmVzZnlOvZOFEJ2i/0qYejzute+A1Ml38Ii1kYbGHqvu+N/gjZaIk7ZpJl3pFIFlc1uaeKGxXzBcM5rcovNwPlN4TrTgfPPshlMmyGX0VDMabxndykiQbgbYecLLZRFG21bUJycf8sI/eL+sKj5SqNDuU7LSF53LWxez2bYgjDI6ZkV7B4vRG6YZviczd5HnwDRyma51sLv/cMx/MEnHsPvfOQ7vH+RCvLxYja215OuOyt1555Ub/vvp7TQbRkWyvV2ZGIr4SmbOTdM0G+jPXZ6HgzAnUd2AwB2jxf4+yK6SXIZLbTYbBoWGm2vh3hU6PNNO8M0n3FCl5ye4fhNb1IExZFu08t16KqTWF/K6V5AkNU5+gTobqONUzZXBGUTcDaLqo02dzR1m+N50g0HotaRIBOlrK9AXiujhSyqXY5pvViohgc/5gMb5YCzkVfMdaZmhyH20Fs24xsDXR/H52yayOkqNFWBHTFnM2yTOxviCJI4yGJzgOA9myE2WsAZhh1Ggy8qvQvQtQcmQm0YYbz59iugqgr++dGzsT+XWNnkPZshNtqEkee6sFu9WHWCPJIUqmtJo6UbUJErm5nIAr9tWR02FiextPtFxhmv0vvOOeBYaYsh4wqCFLNeUEcroHSJ6FryWXWEYdmYLddxeE/nGIH1hhYFSXqInfEnnTbaNDYkrspZNrfzBDciFqutvs3YDD7n9HLdt8tLI0VURcHrbz2IXWN53P+NEz7FpME3oYQ5lxHnJxXj1OMKuIWE8D1adi1xce93Xhi+TvZKUgS7FZvT5XpHOBDgbY4kSaO9PDB7L8pGazOGSqiN1puzqbsBQYCr9Ma8blr80biPTKBA7qZshu2+kyI8ErhW7BxzZogGx5+0eXGvIu+OimibFlabRk+bQXqosukWlIKN1rQZLJuh2nCUh1I+w/MCxL7NuUoDDF6x3CuKouBHXn4VXn7dPuE4kyibFVy1J1zVpNdimFaHYpgWMTRJ5CPfPIWmYeHHX3k1nruwjN/624cBCE6XUjZ2NBAvNt3PoNm2fCGA1BvdNixUai2Ml3IdGzciXhptFoqiOIqfcE175NQcxotZvo5QFAVXuOqmuAHipOB2XlcqXF31K5u93PvoGrZUbcFmiLXRjhez2D1e8OVQXFqqY89EEZqqopTLoNEyYbs2f3G9lctoyGe02GLTsGxkdFVwcHWq/+KG3Eghg5PTFf47uymbJy5VMDWSi7yXTJRyqYv1OEYLm6tshqW3F0JttMnasADvXkP3wOBGXhS0eW9YtqNsRmRZiInMImudhrCVSXTGKopyjaIoX1YU5Wn3z7coivK763to2w9xkSNCg7yjlEdxxEEv7BjN45XX7cUXHj8fGldPRH3BgoSNLgHcgKCEkeea6u1Wk9Ui7sZJ9JIG9uWnLmBmud7RExA2WJhoG3ZHg37UIiNIGoU3iqyu4d/fewt+6K4rY38un9X4+UG7z7mQXXuagZiGmeU6bIZEYxT6DS1gkpyPO8fyncqm2fn5xeErNl0FvKPYXO23sukpFXOVhu/GK/beZjQVP/aKq/HchTKeOOuFT5ydq7oqV/eezXm3/2/3uHf8Gd2/iKzU2l2t33nBUkffwVwCZZMx5szYDClEeOGWwDJ++a5AselaI4NUGwZsxjoCgtrC6BNd9Wy0caqmeIz1Dhttsp5NM8RGS4VEcIHlzC0sdNwPaDMpl9FQcNM7qfesl9mv1MMq0hEQJARiVJsGRgqOhbfAbbTeORtMol0L973mWtx2pZOqGzWCQqTZNnFhsRbZrwk46nS9ZcG0Wd+VzYuLNXzm2Dm88SWX4adfcy3e/sqr3Q0kk79HE8UsbBatgs+tkLLpfKb1tomi2LPpnqNNw+IjR+Lw0mi9ayld0yyb4dHT87jz6l2+TR4qNsVQtpyuhSo5XiKuv2cz6YazCNkqZ92CO85GC7gj3UQb7XIN+6ec866Y08HgbAwZIZuO46UstyGHwW20PPCs00YrFsMjuQxvQShm9a7F5snpSqxQsHMsv+b1g8hoIXpTfS20DCvWTmpYNsq1driNNiwgKGHAJOA5zbxiM52yCTibN7qqRgYEha09aGNVptF2knS19RcAfguAAQCMsafgjBuR9BHeKxS4+O3rUmyG9Wym5Qdeegi1lomvfPdi5M9UG0Zs8icRlkZr2TaahtWjshk9YzNIt76wIDZjeM8nn8SHv36CW+DoGEfdHpqwG2lYz18wRCaKpCNguvGqG/bh1kPxIwPEVEieThqycNYTWtFELrp9MElnVfaToDUrjh2j+Y6eTcNMZ6MV+yfJarYq3KAtm2Gp2krUV5z2OafLdXfYvMVVj2Dv7RuOHsTO0Tzu/4bTu9lom/j6M5dwzw37+I02Lo2WFrJ+ZVP13TSbhtX1GpMXbNtUqHIbbcx3s1Jvo9G2wpVNUhdjNpsmR3L4/tsu74j7j1I2yRIsjoagvmWfjdZ9j+P6NQFvY6Ae2Emn+jhJGm0wxCIXYaMFHLdLXLFJyuZSl3E1cYQp4Sv1NjKaZ88UE5OrwkiKsHCy4IzNfuGEm7FY9fjMXBUM0f2agLO4pM8pSV9Y5O8J+Z791VeeR0ZT8ZP3HAEgjGSqtrgaTgVK2HfUtGyeyr7SMGDZbv+92LMpFP7BfuTQ48xoyOkqL6zF4Jzjl8qoNgzcefVu32MO7XaKzaCNNmxxXQ4om3St7k3ZdB4zT8Vml02vI/vGcWmpjtWmAcYYppfqfFOU0p1rLZMXjiLjxWy8suneO+j7Km6qNw0LLcPyjWahNdNoIYNrDozH/u56y8T5hVVc445SC+MnvucIfudtt0X+e1qighCbhoU//vRTPSfV/sePPYL/8/noIRFRMzYBMSDIO69WEwZMElnd2TwCkhebYltSPqNBVWN6NkPWwt6cTVlsBkn6rS8yxr4TUJaiJTBJT0T1bOazOiZHvPTNIM22CdXdyemVGw5O4qq9Y7j/GyfRaJu454b9vv4tZ5fKjk3+JLyezc4QhOQ9m15gzdJqC9cdmEj0uJzQ55YEw7TBAHzn1BxXRbzRJ15sebDPMcyGKYbIxNEPZTMpBbGHTliMBtHV5LPqiIvunMsDm6BsksKWpId451gelXrbt0HgWKHSpdECzudeD1E2K/UWbMb6lkQLeMXZeTflk8EZbUHz9HzhNbqGH3n5lfjTLzyLp84tYXq5hkbbwhtfcpnvZyKVzUrnjT+b0XyFWssMTzIWEXs26btAvWRxo09mIsaeAMmUTU1V8O/efHPnYyP6qGmO3uSIX9k0LRaYs6m6r6GLskk9m2SjDSib3YvN5AFBALBvooATbm8XQe931i0gmm0L5dX4cTVxhPW7VhpOPxq9LtG6KfYxUQEj2minl+vI6WpPKmsctOi37M6+V+KMGw50VcTYE8Cf7pl2DqT/9/iVvmcvLONbz8/gJ7/nCHc+kHVwodrkGxTjvNi0Ou4PC+4sbcDZmCGXQ9EXEESqio1yvc1n5UbxltuvwE2XTfHPUlQ2Hzk5B1UBbrvSn5j+vbccgG0z3wajON9ThFR16q8nNbKXjVba3KBNsag5m4Q40u3Q7lHU2ybf5CgJ40jCevfHi1muRIbh9Wx2Xs9ItQzLFLjh4CQKWR3Hy+XI3316pgImHH8Yu8YKvk3BtRI14u3UdAWfe/w8XnbNHh5ImIb5SpP3F4fhBT9GBwSJm1X1ppmqTSWjq/xeHZdt4XuMqGzq4aNPGGNYbZoRyqa00UaRtDpZUBTlKjjrHSiK8sMAptftqLYpUcUm4FhpI5VNw4lAT2IzjUJRFLzrTTdhx2geH/jS8/iZ933NtwNHN6G45E9CC1E2g0mv3dA1he9WL6yEJ5aFETebMAx6z6sNA4+emoeqeAslupiEXYjbps1DQbznThZ5X2sm711dK6KySbvPYeqU05sQ3S8UxsWlGsYKmUTqYr9JZaMVVASCFgxJEaPUubIphGnQ7+7XjE0AyGrOc764sMr/jizRYSNs3nTb5ZgayeEj3zzJx11cf3CS/7uzGRJ+fs6vNDBR8kfQizMU+XN2KdBp+DqNPQG8fsY4G+3MsrOIjFM2exmtk9G1UGWTRkOMB5RNw7T4JpkuhHh1e91ZPbzYpDVOVxttTM9m2G7+Xjc0Ttzw4DZ5XUOhX8pm4LOq1v19+/S+tEwLK/W2UGy6yoRgg5tZrmPvZHFN96kwus1vBZxF/Gghw9OWwxDP/fwais2ckEbLGMMHvvgcJkdyeNvdXssD3c+Wqk1+ztA1Lez7QfZRwAnqa7iKTSHXqWy2DKujHzmMy3aO4FU3eLOeSzmdt5I8cmoO1x+c7LjfjxayeNvdV3aM6Am10dYCNto1KJt0T6a+1W7rECrWTk5X+Cb9fjf7gtYgVBgGnWTjxVyCNFo11EZLjxsXviP0fDdeNoWxYjbWRkuhRmFJtOvFWCGDWsvsUPDoM03qEgtiWDYuLdUi1xR8LnXI2o6UY/H6kV7Z1LjrL6kQo2n+8zps9Em97fT7hgkvYuuLxE/S1da7APwZgOsURbkI4FcA/MJ6HdR2JSogCAi3TRFrnQtG3HBwEu/9uVfi3997CwzL5lYzwCu40iib4kWYZlgmHn2iqrAsm+8+Ji02M3q6nSVx8f3dF5dQzOn8RioqmyKW7cwyCypjccmXIsEUwfWkIETG04U7as4mQ3cFRuTiUm1T+jUBTwVIUujSbqgYEpQ2jZbPUTSt0J5N6gldF2VzUSw2PZU6+Dk66uZVePLsIp6/WMYbX3KZb1EYZ/OeX2l27JaLMxQBp0+5m8JXyOqwGYNh2Xw0DR99IliRg8QpmzyNtgfjRpSyuRxShNEGl2ijzQXsonHPA4g2Wr8a21XZFGZzElTwhKn3Xmicd0/gAUHcRmt5duEu/XthhKbRNvxjJ/jmnmGj2jQ6Cop6wEa71rEnYYijtl6YXcE7/+wbHQv6UzMrzkzpmELXr2z23pKSETYdHzw+i2cvLOOn7rnGd4+ecu32i6tNNNomFHj31rDvBxVZ+yaLruXczWnwjeZx/nt5tQXDsn3qWhJG8hmstgwsVps4NbPSYaGNIqrYrNRb3NINeNbXYk89m87vmEtoox1zQ4KcYtP5jlDPJuVG0DWgo9gsZVGptSOLJBqXwtc5dqeyOVbstNHecNkkxotZ1Fpm5MbIiekKdo8X1j3dXcRb5/jNii3uUOmtcGqbFpqGxd/nIAsxNlqAXDL+OZtJnEyE2LOpJQwI8odFqaGjT6h9JsxGSw5DqWx2kuj2zRg7wxj7XgC7AFzHGHslY+zsuh7ZNsQLCOr8YuyfKmKh2gy9qLfa3Xup0rDDXYCJFgZaWCfp2SRlU7Qf0AI96c4U9eHwRXzCnXlvPmCyC2TwoiDeCKOUTdGuJhI3WkIkTVDSWilkvREILUH5CBK2QdCNi0u1TenXBARlM8H5SMomheAA6dNo6WcNy+ZKhFhsLq1GW4J6hW58YuooqVfNiHmpb7rtckyUstBVBa+75WDH74u6Cc4LMzaJ4OiTVgI1WAx24D2bFGSkRytQ08t1TJZyoa+JHt/LXLmoDaByrQ1V8Ssk9LOew0ThFuBsNxut7reNBm20SZTNoNXr9qt24d9+7/WhIxBofIh4btAg8XxGQzHr9M6Way2M5PWexiyFztkMhJ9wNc20XNXTr2zSPYQxhumIAKi1ogm9cydnKjg7V8Xp2RX+76Zl4+xcNTaJFvAXHWvp2cy5PZuGZeODX34el+8cwRuO+r+LpZwzHmax2uJzPeMseFRkXbVnDNWG4RWborLpfsY0kmmiS0EWpJTPoN40cez0PADgpWssNoMhRTQOpJeNVjrP5ioNqIqSyCF1zb5xp9hcrkFVgD0Tfhst2XyD17SJopMKHJwRS3Abrdp5PeM2WuG9v/Oq3XjLHVfgugMT/O9XIvogT06XE08R6Be0YRvcoKFrvxiw9mf/8iz+4cHTiX4vrYWiRJLFahM5XY10J+WzOt9cZYyh3jJSbVQ4PZvu5l/Ce4d4j8npThpt0EbL18IRx53LqDIgKISkabR/oCjKBGOsxhirKooyqSjK7633wW03aKEf1ksWFxLUiFh49ko+JNyBK5sJlCRNVaAq/otwWhut5jZmL7o782kUo6weP7xehC6IN1zm2A3FGyFdTILKZlSxmeR5nYumuaaeoDTkszoPoOCjT0LTaLtb0USabRMLK81NUzYn3YIqSf+XZ1nzdliNkGCIOMS5XXQ+iJsQCytNqEpvClK357QZQGVIQ7DRhn3n8xkN/++9t+Jd339TR3JsVOHFGMP8SgO7xzuVTV/PZoLrDCUdNw1LUDb9NtqwdNiZch17J8P7kLw5m7FPHUpUQFC51sJ4Meez5tLP+udsOsfe7XUHbbR6wEbbTdkMU9rzGQ1vu/vK0H6jPe57NVMOUTbd0SfNtuXM2OxRJQlLo11pGL5+uaxQZNfbpm+Woqoo/NqzXGuhZVjrUmxmBDsjPZ84A/fCYg2GZcf2awKebR0ACpner8/0nnzqkbO4uFTDz77uutCNkqkRZyRTrWWgkNN8g+iDzFUamHTHYfiUzZCAICpM06pjpZyOetvEt0/MYsdoLvFIKydILOQ7Vm/7AriyuobbrtyJmy6b7PjZbtD3b77S9PUMx3H1vnFML9dx0lUL6TyhezxXNkN6NoHwWZs0y1ZMoxXH7lQancrmgR0lvPtNN0HXVP73VOiKVBsGLi3VN6HYdDfVA+ucMBvto6fm8I1nk3XQ0VqIlOUgCytN7BiLnjLghM05x9BoW7BZsowGwrHRpkuj1TtstErHSLhuxWZWmD0s8Ui6VfwmxliZ/sAYWwbw/etyRNuY+J7NTtsU0S8bLVHkxabgl+/yBQsiBvwA4PO7Etto3TRasj6m6YVzUjeTBwQBwKuud3pXxGKYFk6dxaY/ZZPIxvTEES3Dgs1YT1aiXnAUDqeHrmnG2GhJHegyfJ6gG8jBTSo281kdf/gzr8D333Z5op8FPFWQuTbPNDZaMVgobM7mYrWJyZFcT+pb9HN6v4uKenGsSJSl9aVHduNNL+l8X6KKzdWmiUbbCrHRdvZsdlP4PGXTFNJoAzbaqGIzpF+TjgOIDwiKIk7ZDG4MeGm0zndAE0afdNuY8AKCDPfPzrHygKAu3yvLYokXRICjzowXs7jkUza961LBVTaXVluY6DGQh5RNshMy5szSFG20pPwuuso+LVwVRUExp6Hu3kP42JOIz3gtaIKdkYr9mbLnYjjljsC4KiaJFvB/39aaRgsA93/jJG6+YgovPRKuEO4cc1KyGy1X2dSjvx9zlQb2jBcwWsii3jL5Rpc4ooU+i1lebKa30QLAI6fmcefVuxP31kYrm62OY/iDn3gZvieQGJ0E2iCtt82u4UAEFW2PnVnwbYrSPb7Mlc3ONFrACxEToc8mq6v8O24FbLSqEr1OGnfV3TBl8+T0xvdrAkKxGTgmKhbbPneLjfMLq12zHegeCyAy2HKhGp/FUcjq3EZL68c0WRcZn7KZ7FwWr8E5PXz0SbeWMiedWdpogyS9u2mKovA7lqIoBQAbZyrfJng76uE2WsCZFxWk2WcbLU8Ca4nKpnMhSlpsUv8TERwr0vXxqgrTtnmxOZVipEQuk1zZbFvOhfTAVAk3HJz0LYZKeR0KOi/CvLe2hzmbXOHdIBttXuyho9EnET2bQLyyeXauit//+GOYKdf52JPNUjYBZzGRZJOFrNV8pzZixFAcYs+maKOlm+7iaquvMzad4/M+Jxo5QLaiZrt7MmyQbERYDtmLdwWUzVywZ9O0uwblFITCnhTMXBcbrWnZmK80IlUvT9nsodiMUTaDyk/GLTYNy4auKlAUhT93rsu5Qj/HAync94mOuJuN1rDsxIPHiX2TRb+N1vS+34WsM0twrtzAVK/KZqD4qbWcYAxRtaGFOrU7iIUAhRQB6zf2BAAygp2R3n9R2Tw9s4KcruLgjpHQxxPiRspa5mzS+V5vmfi333t9ZNE2NZLDUtXp2SzmdOEaE15s7h4v8EKICkqfsslttD0qm9QzatmJ+zWBmJ7NWjt132gU+Yy4CZxsDUHFpmUz33mXy2hQFYWPZgmbswmEK5sGDz1ToYVsnlXqbYwWspHpp/T9qISktJ6cLrvHPdH1tfWT0bxzTMF2oTBl0zCdEXbBUWJBxPfkUoyNNq7thMLmAHHdlM5Gm1bZVBWFu1FI2TSDPZtdWsqygfnUEoekK96/BfBlRVH+Ck4i7c8A+Ot1O6ptSlQRAzgq20hej1Q2+3kTJ/WxYfh7NhUkL5I01T+3kdSgxI/XFLQNp9gcK2TSWR615F920RL7Bz/xUp96oioKSnk9xkabPiCIj4DZIButOIKAp1X2UGyenK7gt+9/GNWGgYyu4vKdzsLtwCb1bKYlm/FCO+K+Z1GIMyJXmwZURfHNultcafZ9IS0qr4d2j+KB52di02iT/L7QYrNCMzY7ezbpPbNspwjrOvokQ5H1VoeyyW1ngWOYqzRgs2jVi8/Z7EXZ1FTYjMGyme/x5Vqr4/PSXSXPsr1kWHo9XUefBG20qqdsqkrC0ScpVfG9E0UcF8af0MZANqPyDcMFV3HvBVGJzuqaF37is9G6ymbVr2wCbrHZ8opNBcCeif6NbCA0wUZLm5qzgrJ5eqaCQ7vHup4/PmVzDddnOhdefeP+2HmJO0bzWKy2MDWaRzHr9dUG3TE2Y5irNHDXtXu4qkyvT3QKZQM22qCNvhtkUdRVBUcPx89vFsllNMdeKiQqM8acDZ1ifzQJcTM96esaK2axZ6KA2XKDO8MA5ztZyutc2ey00TrHHFZsivd+b5PDn0Y7FlMMUyEblkh78lIF+yaLiYvpfkG9tKsRxaa4+UHn5vmFWuz4FXEdNB0ikDDGsFhtxSqb+azGP6NVrmymCwgivSPNRp7uZhtkM2pPPZtBR5DEIWlA0P8PwO8DuB7AjQD+i/t3kj4SZ6MFnATCMEtCkmHraeA9my3vC1NtOLHTSdUFssES9ZaJjNAD1fXxXNmMvyCFkURhJMTCI5/tDNIIm0El9kaJRI1ZEOHF5kYFBOW8EQQtw4KqhJ9f4qy6IGdmVvAbH/42Clkdr7lpP7729EU8fHIWUyO5vtq31xNnjAfZgsJ7buMQ5ygalo1d4845SX0ui6vNvibRAtT77Hwuh0nZFMKeuqmMQaJGn8y7qlSwZzMn9CBT+AxZ9aLI855Nk99w6TsVZaMly2NYEi0gzNnsQdmMsiaWQ0ZD0OKkZXqLZjr2bvZhTXV6FMlGK+6kK4rS1XYWNvqkG3sni5gtN/gGkRgAJvYc9ppsKW6wAJ71b7Qo2GiDymbRr2xS8TezXMfOsXxPQUVdj5OHm3k22tmKsynLGMOZ2ZWuFlrAr6CnuTYEObx7DJfvHMF9r7k29ud2juZhWDbmyg0UsnrH+02Ua0667J7xAn9/SbkV7/u6e70wLBujhUzq84laO266fCqVXVEcuULUWiZMm/Wthz2jqXyzIM2orSNuny45w4hSTke5Tmm0ETbakL5Kfu/QVGGeuN9GG5eU6ymbnb/7xHQF12xwvybgfO6Ogyui2LTEYtP57wtCQnoYdA5rqhLas1mpt2FYduzczHzGS9KngMmkzjrALwak+S7Qph+NPrEC947VhrPZHOV+iFL6tzuJPwHG2OcYY7/KGPsPjLEvrOdBbVe8NNrwjyVq1mazbfpsJmslp6tQlc402jQ7blQsErWWkSqFjhqzF7v4+sPIZbTQEJIwolRKQhx03fmYMBtt/EXGC0raoDmbGS/sqemqYWG2Lj2iEACAj3zzJFRFwf/86bvx82+4AVldw3MXykOjagKO2tMO2mhT9Ww6P0uhEnvcwmy1YaBlWKg2jMSJyWmg5z0kFJumZcO0WeoNpqzuJDwHVba5SgOaqnQWXzopfd7MzO7KpnO+tdqWp7R1UTYp5GZflLKZWcOczZDnbLrfheBCmH622Tb5f9Nzd7PRAv4eIVE1VxUF3VqhTZulttHunyzCZoxvFvDiPqP5zo0kIVphBK8JNKDdp2xmPAUV8PcxFXKeDc4Ze9J/Cy3gTz8nJXWx2kLbtDBbaWC1aSYrNt2AoLXOrD6ybxx//gv3RG6eEFPufW2h2kQhJ/Rshij/gLMZ5BWbnTZaRfHSk9Mm0QLeZ5fGQgsIfdrCArvC59j2LzCNnidOOQxyxO1/DI7cKeUyKK+G22gLWQ0ZTY1QNr2N5rB75krdiH3NuqailNM7fne51sJcpcGPdyPRVAWlfAbVZjCN1nldZB22hT7M8wvJis0DUyWsNo2OViRyQsTaaIUk/VrK0XmA/xqctGcT8FTQnDvexmKdyuZoITqkKo3YsZ2IvYMqivIt9/+riqKsCP+rKoqyEvdYSXoM0+a7k2HsmyxirtLwWR0ZY250ev92jBVFcee0+YvNNLtKuuZP8ao1082WJEtbL8VmJoVnnvdgRBQeI/lMzOiTzoCg7jbadEFJa4XOi0bbjLVe0q5xMHltvtLAg8dn8aaXXIZd7vyvt770EADnRjIs5ARrS1TAUxz0s2TroRj91abhhVj1WdkEnHO5kNX48zXanmKY2kYbsZidX2lg51i+w2ZIqpXhzksT/y6KvC+N1hkRQDdlUmqDVu2Z5bqTLBzxPffmbPYWEAQAhuVdD+gzDKa00iKj3jL5f9NzJzlXxKj9jE/ZTGij7UHZBLx+yLZhQVWc/mSxCOlZ2QwMrQ8b60BFOLfRCqpnUVA2p8v1rsVXr4iLfvGeNVdp4PSMs0y5uksSLeCdK2sJB0rDDiGHoOgWOEBnWvNcWSg2C9SzWUfe7T0UoWtCL5/5oT2j+MU33og3JQhd8z2n3qlskmrYz3mRdG3pNmNT5PuOXoZ3vP56XLHL369bzOm8mAq2UyiK4szajLHRZoQ0WjGbotKIVzbp+IM2Wi8caCLBq+o/o4UM30wigjZa8b5xvpuy6Z7DV+xyNkmD6iZtTsUHBIk9m70om97nqqW4tmrCRmOYjbbaNGOPI6dLZTOM2E+AMfZK9/9HGWNjwv9GGWPdtwolXfmLLz6LD3zpOQBOml7cgmP/VAmWzfhOJwCuPPTbzlh0h4IT1UY6ZVNT1Y6AoDR+e1117EDlWst3U05CmtEndGOPsk2NFjqVTSPSRuvsaMVZ5riNdqOKTeq/dW20USMcuAIUaIb/9LFzABjecscV/O/edteV2DGaw42XTa3PQa8D4m4jt06nuAHRgmTZ3Q3fKxSbS+7f9XPGJpHVVeydKCKjqdBVBc22xW/AaccdiYm6IvMrzdD+m2zGCyxpx/T7iogqRztkvExYX/P0ch17JoqRPXX0OfVL2SR1OhheQtfepmHx/84l7Nmk56JrXnobbbo0WsBTgqknqmXayGWc4l4smNbcs+le77iNNkTZLNfa0FTF14tOAUHNtonl1Zavb66f6EIqaE0YKzVTbuD0TAWqAlyxu/sYD7qeb1Q/vbjQLsT0bM65AV6Ojda5B9NsziD0O3oJ5lEVBT9w56HU96aw+aC0odPPUVDkmkhTbE6Ucvihu67sUKF8va4h37vxYnixafiUTc++DTgb/yv1dlc1d7yY7UijPXGpAgXdE5PXi9FCpmP0STDjQPx8LyyEJ8x6j3V+lhw5wWDLxWr3e6bPRpsy8wPwr88yKfrhuatFV6FGjD6JKzazKQIqtxNdPwFFUVRFUZ7eiIPZjpyeXcEz55cAOF/quNCSsFmbfOHZR2WTfl9wzmZU1HMYTs+mf/RJmiQxXVOxsNKEzeJ3v8Jw5n6lG30S9b6nUjYjFvMig2CjDYN28yxff4aFzz9+Hndds4cra4Bzs//bX34dXn/rwY7fM6hkM97sq24bDKGPd3+2HLTRNk3er9bvNFrAKXLoe5/Pao5imFBlDCIm6orMrzQ6woEA73xumd7MzG4Knzijl4afB4+hs2czXvWiVNielM0Qq1uUskk/22ib3HZF73GSYtO3uBEWsFoiG63New+TMjWaR0ZTeSJty7D451MQjrdXG20wjXalYUBV/As+VVH4aw1aywo5JyBoPceeAJ5FzrAYGm2TL3Bny3WcnlnBwR0jiTZm+EzVPt9LoxA/l6Jgow0uVOcqDZRyOkp5JyiPXkshFz5nF+ivotgNsu76lM3aOiqbfQjQEdciYff+8WI2vmdTV3lvH61zai0Tls26FsNhv/vkdAUHd5Q2bF0QxMmmCNhoA8om3Tf2TBSw4M6HjYKuGZftHIGCTmVz0Z1LHbcRls9qMG3mjhszUmV+AP5eXC3FtZU2PXOZ8NEnqw0jMokWSLf+3E50XW0xxmwATyqKks5bIUlEKZfhuzZGFyvVvpBZm2RvS6tydKMQZqNN27MpFputdMqmpin8gpXeRts9qIfoZqmknk1RmYgOCAq3KYp4yubGLGj4GBtX2YxaNNNCVxxQ/bVnLqFSb+MH7jzU8fNr6WnaDEKVzRTFJn0vec/mBBWbno02LuygV971xpvwk/dcA8Db6W0lVBmDhPWEWTbDQpSySYtfw078nDldhQIvkCr4846y6b95z5Tr2NclpTSra72l0Ya85iiLn1dsesrmeCmLUk7H/gQWULHAFPsvFQWJAoLSzmjVVAV7JgqYdnte28LsVVH16rVvLlioVxvOWIegwkznSXAzktwxvCd3an1ttJYbEHRwRwm6qmC23MDphOFAgKfkb5SymdU1/tkUctEBQbPlhm8sERUz4cpm7z2bvRLWs0kbOmlUyKTPkyYgKApxLRJ273eUzc45m+JGs5fgHrCZdzm+sWIWlQ5ls4wjm2ShBZzvbkdAUKDIpNd+1R7n+3RhMVrdpMeUcjp2juU7Emnnqw1MlHKx611vZrOF1ZaZelyceD1O5WKicLiY0SddlU1po+0g6ae3D8AziqJ8BwA/axhjP7AuR7WNKOW8vhazy6D5qdEccrrqsyRQQdhvG60zFNz5wtAw7zTKpqb5ve71VsqeTWHh1VsabUJlM4GN1rIZmoZnWyJlLGzOpvg7w6i1DOTdlLONgI+xcYuUqE0JL2TDswR96jtncfnOEdx6KHkM/qCS0zV+M+0WChUGKTi0iNo1XoACZ5ezYZjI6WqqzZSk3H7VLv7fXNk0e3MzhM3xW15twbJZRxIt4Kl6bVNQU7uk0Tr93ppgow1RNoXnd8IjDJ9yHkZWV9dmoxWVzVUqNiNstG2TL2gLWR1//6tvQJI6Nyr9UFGUBD2bLNWCiBBnbbYMm/dQ0rkxXsymtucSweInaqxDNqOh1jIxGigsClkNhmXjRTdMJEnB3gtcYbIdZbOUz2DXeAEnpstYWGkm6tcERGVz4xK2p0ZyqNTbKGb1UMs34Cibe4Tv53gxi7lKI/Sev5aezV4JS6Ot1FsYyWd6OqejoM+lL8qmcK0OO8aJUi42ICgj2mht7/sBdC+wqWeTMQZFUbBYbWJptYVrNiEcyDumkGIzEKhHr/3KPWN48Pgszi+s4tqIAtkQFOB9k8VOZTPBlAFvZrOJetNIrfr6ejZTbFRqgqtF10KUzaaB0ZjC11E2pY02SNKr6n9e16PYxhRzOuqkbJrxxaaqKNg7WcT00vrbaAtZnTdxN9oWbMbSK5u+OZsGj1ZP9HhBGUjbs5lLkQbGG/6jAoLc11xtGPziZ0QUK3FDuYm0RfdaoRt0s22FJnASwTCQ5y6WcWpmBe9+001Dp2KGkdU1IY3WWzCkIaOpfPd6NJ9BMaej2jRQqbWwYyy/7u9TPhOw0fYaECQUXvMr4TM2gUDPppncupvP6G6xaYV+R8Tnp0KpW1Lpj7/qaj7bNQ1hr7lcb7u2Rf+x0TWn0bYwNZJ+oULPFQx5c+ZsRj+OuSmPadNoAcea+uz5ZTDmJAZnA8pm0CqcBq9Qdw6+2jA6CkrACwkKbkbSMZyZraKU01OFe6SB3re2aaHRtlDM6dg7UcSTZxcBAFcmVDZ1VYEC9DVsrxs7RvN4Ya6KQlaH4m5odfRsVhq46XKvP56yE8KO0ys2N07ZDA0IqrX7fgw8jbYPaqm4FgnbaB4vZp1ZwYFrmFhERSqbXY5vophF27T5jOYTl9xwoE0Ye0KM5rOoNQ3fPOJOG63Xh6mpCi7EJNK2BffQvskSvnNqzvfvSeZSr1XZFD/XVKNPhBnLTs+mP5Cz2uiibLprDdpMkDjEfnqKouQBvBPA1QC+C+AvGWNm3GMk6SjlMqi3TR4r3W0ncN9kyWejXT9l07PRkpc/zWJBE9JoGWOop7XRql4oyHjKwdAZXUusbLZNG6oSvaCkBdRqw+DqT5SNNirgQSRtKu9aoTE29YRptGR9/tR3zqKU0/G6Ww5s2LGuJ7mMKkS5pw8IAtzRFm04C9KcjhE3PGqhh8TkXsjzwJU+Fptu2NiuEGVT7Nlspihw81mN2307bLSa315PFstu/Xz33nGo6/OGEaYWlVdboQthen8abbOnwo8Xm4HzSlXjA4JI9exFgdw3WUStZWK1afje76z7vZ8Y6X1h7hWbbkBQvR2qQNN5EiwCKKTohdkV7JssrtvCi943CnIrZnXsmSjw9/XqhMUm9QZv5PWZrhv0XmV11ZdGW2saqLVMn/OAipkwuy99FputbJZrrb4fg2ej7UfPpqBsRhSbgDPCRbw2ev3+ThIwzTUFPGWzW7FJ35NKvY18VsfJ6TJUBbgyoQK/HowWMmBwzjc6Pq5sBkagFLI69k0WcT7GRssdY5qG3eMFLK+2fIX7QrWJm66IDxjkG+WGhVrKaQjAGuZsUhJ5xgnlE5XNJMJLLqOBwXkP1mOu8LDS7RP4awB3wCk03wTgf677EW0zuM3RHRbfTW3ZP1XE9HKNL17I6tr3YjPnpdHSTTxNr4STzGjzY2RIlyRGoQ9To7nUvVpp5hw573n0XDW6sYlJba3A/EDxeYEuymbb3NAQABpjQ7MFo+axijHui9UmvvncNN5w9LK+n1ebRVbYgBB3p9P9DrenK6dDVRTez7u02sKOdQgHCkJR8L0GBPEAK1NUNh33QljPJilW7ZRqai7jHGeYjVYXerEBYcbmOlkso5TNsIWw6EzopfCjxwcfq3QJCCLlUO/BWi+GxrVMi58T9L3vi7JpUkBQm6ehiuQiigC6dlxYXF23sSeA937T6IaCq2wCzriQNPet77lxP15yeGf/DzICcu0UebHpT7Kk5Pk9IT2bYXZf+iz6Od+yG7zYNEUbbfdU1rSMFbOYKGX70oJS7GKjpWMvB6y0wY3mjOYpX4lttAWv2AScJNrLd432PXcjDaKDi/B6NZ3/bwn3zct3jsTO2hTtxrvGnXsjJdA2DQurTaNrenteGNvmOONS9mz2PGfTSyJ3ejYZX2/TWjhe2ey+DtyOdPv0bmCM3QwAiqL8JYDvrP8hbS+KbgFWd4vNbjvq+yaLaJm2s8AdzaNpOOpj/wOCND4gmy5AqZRNVeH2EkotS1NkUTpqL4v4rK7Bshksu3vohmFasQoXvWYxqc0wnXl2wd8d1h8WpN5Kf9FcK9R/6/Rshr9WMcb9c4+9CNtmuFcYdzLs5DIa2u4mAe+57UHZBLwkw9F8BqsNAwsrTbz8uvUvNnMZZ35Xk3o20yqbITbvuZUGClkNIyEbQZ5Sb/PHJCk2C0LP5kQpJCBIeP6Z5TpG8plUSdVpCFU2a63QGbFikdhLrxk9JrhhqHYJCCKlt1cbLeAUm23DxljBe+5777gCNxycTP07CbFQd8Y6GHyhLEKLq+C/Fd3Fos2wbmNPAM+VQdfoYlbnabwUZpKU//ADt/b34LpAyiaplMHvBxWbu4UALXqfw+4jtEG0+cpmGzdd3t9i80dffhVee3N/nDa0FsloauhGM42OCfZtivZQwFmn0LijlXobGU3tasMedzdsqG/z5HQFLzuyew2vZu14m+ptAM53lTbVg2m0WV3D3skijp1ZiLSKGkKuxW53I3Ou0sC+ySIW3Q3ObsUmvY/NtoVaK362ZRj+wLYUyqa7rsvpXraGzQBNcRxuQPxaWPw+rFfrwDDSbdXLtzkYY6b0H/efknuTqbXMRCERdNO+tFRzis117NlsGo5lwFM2e0ujpbTddAFBzrmWtl8T8O8sFbLx72eY+iJCu+LirM22Ff6YTIhyFKTWNDdEBRPJZ/Suo0/oAts0LHzmsRdxx9W7sD9kQT6sZIU48qie266/Q3N+ns7jkXwGz1xYhmHZ6zJjMwh9jrSoS/ud99JovUXhfKWBXWOF0AWDqFh4amr3m7bTs+naaLuMPpkpN9ZN1aTnAzpHn4TNiPWNK+kh+ZbOp+A1XFEU2DHSpldspi9w9wrKZvD7/f+89rrUv09EfO9ahgXDsiOKzQhlU7jer6eyScdJm6KFnMav20n7NTeLu67ZgwuLNVy207nWOgFa3vdzlopNn7LpvM9hm02kxoRtHq0XwWLTsp15kxMp21+6MTmS63lmbBByWUU5yTwbrT+RNnjvEK9nFVf577ZOpragSr2NuUoDlXob12xiEi3grXNEZZPfLy1/+0nWLSBbhoVqwwhVcsVWFTp3qWWD5sZ2S28nF1bTVTbTBvD5bbRplE0FmqpA11R+H3CEC4073OLWwuImrcSj26d3q6IoK+5/KwAK7p8VAIwxNthX8iGg6O581FsGDNPuajUVbVM3X7FjXdNoAWdXiX/BekyjrXNlM0WxScpmD4t4f7EZ/7NGl2KTdqZ8xWbEY7znje7ZrLc3tmcTcIqjRpeeTXq/v/7MJSyvtvDWkHEnw4yodosWnzTQz9M5MVLIYNlNNt3Rp0VQHIXgnM0elU1/QFAztF8T8F4vhVkA3gD3OPIZDcu1li+whtA1lY//AZzr2JUp1ac0BDeALNvGSr0dai9ds7LJezb9Cxu1i42WHCC9PGfBtcrOlOu+0Sf9QEyjXXEXoWE2Wj76JHCxFe9J67mhQDP0VgRl8/DuUdx8+RRece3edXvefrBjNI+ff8MN/M/BFpC5SgMZTfUplXHK5mtvPoA9E+GbR+sFBWLRNaLaaIPBUwcHEVI2o+79YkEo0jYtKPA2w3XBRhul/Aeh4qxca+PktBMOdGQTk2gBL+HXV2wafhutmOJOheJcpRFbbGZ1jRdmVGReWnJ6PQ/siN/Mps3UatNAy7RTu1/os1WAVEnmuqrydgSv2Exuo6UNVlHpl3QpNhljsrt1naECzFE2uwcE7R4vQFUUXHJDgqivcj3mbDq/3/SsAz2m0fLZkiltuECPxab7XiSZddTukgBcyDpBAMFehjBVLGyOYZB601w3y2AU+azGb5rRPZvO+/3k2UUcmCrhNmHkxlZA3IDoNu4mCm6jFZRNYsc6zNgMks9Qz2aPymzIjuv8SiMyQCU4+iSjJRs/ks86dt+2aYcrm7zwY5irNPDya/ekeh1pCBbYlXr0QrhX2xXh9XH1ZqNNM3hcZO9kwbHRhrzfa8FLqLa9frSQxTQVuMGRFGKAzb4uAVBrIahsFnM6Clkd/+On716351wvsrrmCwiaqzSwazzv+97Fzdm8/uAkrl+DdboXFEVBPqPxnj4aD7WRVt60iD2yYYzkdWiqEmqjzeqe9VZTvbnBlXorUYFdyuvYNZbH/d84icN7RqGrCg7vHl3Ly1kznrLpvF7TsnmBZXTYaAW1cqWJq0NSdEUbbVbXMFnKcWXzwmINuYzWdW1Ha9olt9czvbLp9dCn2XzRNZVf0/gM3xTFJl9/Jgyp3C70784k6Qm66NWayQKCdE11Bnm7u0O0COx1lloUYrFZbRjQVSVVQatrYs+mU2ymuVhk1qJsJhhBQkQVjoSiKBh1U0cJwwz/nLqNPrEZc5TNDQ7dKWR1vgCImpMonj/33nFFTzMNBxnR6kWfT9rvDJ1XxbBicwOs0fmsDsOyUWsZyAgWn6QEw3LapoVyrY2dIeFAgDD6xLDRMu3EqpkTSOUUqEFlM6N7trOlahOGZa+vxTLwmul7EKZsrrXYjA8Iii42+cKsx+CTvRNFJyAo5P1eC7pQqNMiNEzF6Gaj1VSFh4SsB9RXRcrmMIeaOT2b3iJ1rtLomIE7zmfADo4WkHPHMgFOTzSwseNX0kJrkaiNZkVRMFbMhhabGWG9kNFUPpt6pW7wzyYOVVHwnp++G4f2jOKZ88s4tHt001NLyXZNGzZkodVUpWP0SUZXeaAcjc4KEhwpt3Msjzm3V/PiUg0Hpkpd1xj0PV5cdR6XdpPeux6nu08Wshq/xwdT+pMIL9JGG87wXpW3CGTnqLfNRKNPAMeS5CmbZt/7NQGx2HRstCOF7r0IIprqXYS9YjOdDRfotWcz+c5SkgJ/xA2CIaJttPHP2+AK78YXm2T3jO7ZdN7vfEbD6289uGHHtlH4lE1XzU5bUHfYaIWb31QP52la6Hteqf3/27v3IMnO877vv+f0bXpm57a7s4vF7uJCYAFiYYAgtRRJiRIpU5AoWhYpiVJRcdGQZYW5WIkdRy7RJcd27KTMshM7cqhYYRyXaZdt+cqSWGGKIRlFjmTBRUUiIYEgAV4AY4kFdgHszu7OrW9v/jjnPX2mp+/dp/t09/dTtbU7PT0zp2dPn3Oe8zzv81SGKpds3gwJ989Xo5N/68Wslxx9EpZg9xcMhfNAa+H7pOV4llzjlHYnWv/zpObd+WbW5ehFYfKipDBElrFzZtPUJdaM75oPe8Pwzs0V/ebTL8m5wTsUd3OojHa381olv1+0Zj19MHR6ozyWDqKd5KL5mH4bJ71MYZyK+cNl5le393Tp/sNVJvecXtXPfv8b9fYH0qsIGFSpEMRlg/F7bIIdcQdVLuVl6r6UYn25qO2dw8Hmfsv1Vi44PPqk3xmgpzeW9bf+5Nv16S++0LOcdBJyQaCVUj4ul/f/l6vlgm7sVOI5vlJ4DVEq5FTIBXG2slW1VlcusDhYO7VejrvXfvu1Hd17uncmt5gPZJJei+a9D9rFvxiXwg527PkT33sh/j20ltHe2q8qMOuaMPDHQ8poDyOzOWXNzGa172Dzzs3leNbmXqWeyp3ccqLt9O296kDrNaWWzOb+4BcBvmHNcGW0/Wc2O2Upk1bLhUOjT3qV0SbLoJLicuIpdKP1J8ROZbRhuUugx990buJlvpOQHDze7/uslf8/by2jXV8uTuTOtK8sCOezDf7z/H7n7177TpdbHUqAw7VYzdEnfWc2C7l4HtmRzGYi2LzS54zNURzNbPqsS/c1m6NkNo82CFJfmc1hutFKYfOdhpOcBl/H242/AVWtN+KsYbtxFp0ymz77nub/r5fPBUOdZ7ImWWZeqdX1+u0DnV4//PsLzPTBd9yXqeN0MZ+LL663dzu/x7IiMFO5lO96Hthok9lsHekSzg0OewHc3m/fLKeTXBDoA2+7V2+9f7qdaL3VciGuYPCd2/05rlpvJNZshjdqk9nKVq1NFE+tl3V1e0+1ekMv39jVuT6aD4bjm3LxTdFBm151WkPfy+mNZV2ISoPbrdk8tpTvmngZJNmxSGb3qDwnwjWB0eiTPue7nTm+otv7Vd3aq4R32lKYz+RLoPYqNd3arwy0XlMKT/7JzGZgg5X9XLp/Sz/5Xffp3IljA/1cabA3e6XW6Pna/N09r2MZbY81m8NkeMcheTOi08VoLgj0d/7Udw31+54FyXUUlVp94OZAUjOIWEk0CJKk4xNoDiQ1g80bOwdDZbBKhZy21pZ0ORrGfa1HZjMcch+uIaskZjj23M7E+7z1a5KjHV6+vqvAOv/8cYgDplrvYDN5cTRUsJk/vM7HC3qU0Y7SjVY6nBkedB1yN2YWBz83o4vudpnNM5vLOrVePnJsMTOdWF2ayHq0cH5reHE4zI2krEjOA+5VeZAlpUJzu2/sVMI5xANeM0zaSinf9f2yvlzU11++eeixGzuHg01/neNv4E1yvum4rZaL8XIhXxLtg00//io58u3UerlLZvPwDd2ttSXtV+v6+svbqjeczvZ5nbFUzI+Q2Wy/rGEQ/rXGweZeted+3WwQRBltEsHmlJmZlkv5ZoOgPi4W7owuLl66vqu9akqZzejCYe+grtt71YFbjucDixfO7x6EHVgHKcPdWivrZ94zXOt+f5A56CezWT9a6tfq2FIhvkCXojLaNiWFvYJN35V3GplNr1vm4747ptsRL02lxD7RTza7Hb9fJedsSr1buI+Lf59v71b66nrYzvmTx3T5tbCcyWc2u21/OJ+03nVsTqftlHTkfRJmAnwZ7Z5OrpXHvt48KQ6YEms28x1GQxzKbA41+qRzN9puZbTjDDbHmdmUmmtsD2oNrZTybUvS/vilu/W+t9zV9vj+Pz7xjokEHeHvrj7weSZrkmuam2NPJjsqaxhLyTWbUfYv6+v+V0qFrhUp6yvFOEvrbe8e6M7jzfdbLheW0fqb0bMdbBbioNnfOPA3l6q1RlTRlQwgy/ry86+2/V6VlnOsv2Hy+98Mn99v6fBSIacr0RKgXpMaWsVrNoc4lnv5OLMZrdncr/acnUmDoPZm9xbgHFkuFcLMZt9rNsM36pXXd4+sIRiXOLNZren2fvVIW/teckEQtwTfPahqeYLZPH8CqfaV2eyd5Tq2VOizG233g8wwjZLGIXnxn8a+Mgt81qkSNQgaJgPUqRvtMKXew/D/dzeGXLMpSedOrOjyqztyzunVm3vaWOleAlzIBzqI7moPUkbrtWY288k1m9d3U12v6SUv4G/sHGhjpdQ2IBm5jLbDmk3r1Y22MfzoEymcP+j353Gu2ZSaNwdu7XVej2ZmHX9fW+vliTTs8csuZrk5kHR49MnVONhM/z0yqlIhUUa7019X1ml79x+5U+/o0gl7fbmk2/u1+GaQFN7oS1ZF+DLam10aaM2K1XKx2SDoSGbz6Hlza21Jr93ajwOxpGq9cei84hsKfen51ySprzJa6fD7efDRJ+0rTQbRbBCULKPtldls9jpAE8FmBiyX8to9qKpW77OMNs5s7mi/Uk+njNY3CDoIu9H2eoO1yudMtYaTc047+7WJBljJZjC99BN4rJYL2tmvxqVwnb4mHzWq6PRzp7dmMxFsTrnr3bSUEut4w2z24L+HI2s2o7u+k+hEKzWDuGq9/8Cv1bkTx7Rbqen12we6enM/vgjopJTPxXM2+w3Qk8ej1hs5vizTOacrN3Z1eiP9EsHwZ4Yn/utdLoSTd8DHuWYzsObM4Xbi0SdD3oEPrLkucpzdaKXwtVSi0SfDZtMnwWeTJ93pe9wOBZs39mSaXOXEKJLB5o2dSqY70Xofeuf9+pEu86R9ltKv2zyo1rVXqR8po63Vm2Xm/XSjzarkmk1fFeaTDBWf2UwcX06tl9VwzdEkSZVa/XAZbZSd/8qL17VaLvQdlPtziWnw66a40mSE5mS56DU0fIOgPvqXJLu4o4lgMwOWS3nd2quq4fq7ux3OKCrpyvVd7VVrqTYI2jmoaeeg1natTje+3KrhnHYOqhMNsIoD3FmqtrQyb+fYUkFOzWCx9a6dF65xa2ZRWk0r2Fzqs4x2niWzzuNas7lWLuhD77xf73r4zPg2tIul5E2DEYJNSbr82m1d297r2BzIK+aDgRsElXqs2XQKu29fv32Q6vzF5M+sJMpoOzUu8SW30nBZxk530s3C5j2djFpGKykeHzPOOZtSc43trb2q1pazuwbP/+5muTmQFFZg+Bsjr2zv6cTq0kysQT0UbO4eaGM5u82B+tUabPq/kzer8oEdnkM7y5nNpXDEW8O5ZjfapWYZbbVNZlOSrrYZf9K6VGVjpRQvZzjbZ1ZTal67lEv5obvHD9t4TUpkNpNltD3XbJLZbCf7R7EFsFLKxwerfk8sd26uhMFmpZ7KvK1cEHYn9U0KhulGK4XlBzsHtYl2zvMHxE5rJ5Oq9XrP37k/uCTXM3TK8hQSd6Zb7URrNqfZIGhRy2iPdKMdQxmtmemnv+9B3bU1mYHch8pThww2z0drZV58dUfXbu5pq0fzkVIhF44+GaRBUCG5ZvNoN1pJ8RroNGdsJn+mPxZs71a02SXr4o9bw1ygxIFqy74VBNa9jLY+Whmt1Kx2GfuazegC8eZeZeClFJPks9LlGQ82i9Hv2zmnazePztjMqjDYTN7Qye6+0i8fVLYGm8lAOu/LaONgM7s3ZHpZLRfUcOHM94ovo42ufSr1aG5yoiLInzuutmkS1LokzHevlTRYsBkdzwatrJM6zz0eRD7RjdY511cZbT4aq1Zh9MkhBJsZsFwqNIPNPi+Cz2wu66XXd6I1m+mcYMvFfDy0d+ButFFms9ZoaPdg0mW0/Q/V7auMdik86fhObd2+JnlnutXuQU2myQd8/XSjnXfJcTjt5j/29T1aGgRN2jgy1CfWlrRUyOmr376hvUpdp3qU0fqbJ5Vq/6W7hxpStSmjlaQXr4VNiiayZjNRunv99oHWu4xkGOUCpVP3w7AbbeevG7WMVlKcIU6jjNZ3o81y8xNf7pbGjddJKuQDNVx4cXt1e4aCzXygg6hqZPeglumxJ/2KM5tR8x/fyfpQZjMXqNZo6OZeVcvF/ERGYKXF30y6tVfRfu3wms1qtPykcCizGe6br7YZf9KaBZWaTYIGmSvqr12GqVjwlSqjHFeDRDfag2pd9YbrK/AtFYK+GlQuEoLNDFgp5eM7Y/3e3T6zuazXbx9oL6U1m1J44vZ3rYbNbNbrbgrBpm893f3OUjiouL81m1Iys9m59LZbGe1O1JV30l36+u1GO8+SpS3DdqO97/Sa7rtjbWqlUuO4aRCY6eyJFf3+t65J6jxjM/450SiGQedsekcym9Hv/T9EA74nMYPRNwjarYRN2LpnNocvox22QVA1KtEaJbP5wNkNFXJBz//PQRXygfYqNe1V6gMvpZikwpyU0SbPXddmKdiMymjjUtMM35joV7OMNgwyfdC5sXy0jPbGzsFMZzWlxHXOfjXOyvnH/E3a5DlguZTXsaV828xmu2skf2zqtzmQ1LzBOkxmUwrfT6McV+Pr2EYj7rbcz3kwOcIIodk+Ms+JlVI+XtPTb/nWnYk3bFod+MrFvF56PSx3G2bOphSWU+zsT7obbfiz93sEm/12gfQHutt7FTnnVK3VO2bGCrnOZbQvXLvVsyFLGvz+EZiN1AZ8lsXtyKuNIzPA+nXp/lO6NMUB3MmT3Cg3mM6fOKZvRPPjepXRhms2wxNtv+sBk5UWR9ZsxmW0t7VUyE3kotSXgt64fbQMrpU/bg1zN7zTek8zixtMtOPLaEcp97p4blP/5hd+cOzr+wq5QK9Howey3CDI/3/NfoOg8P3yyo091RpupoLNesPFMxHnoYx2tVxUYNL1aP9vu2Yz7kZbnen1mlLYg0AKb6r7kujDczaPjtnbWit3LKNtvUby1z6DZDb9uWTYZEUxn4urHoaRS5TR+uu6foLNZFk5QmQ2MyB5N3aQzKaXVllmuZhvdiUb8M6Sf5PuVWqqNdxE7zibmc4eX9HzV291fZ6/89Sr9CV5x6/ecGq4zsPTi1H3zla396v6gxde13demHyw4jObS8XcTM+gG0Vz9mq9Y4OnrAvM4oBvlAz1ucTJvtfNj2I+p91KTQ3nhststrxPfED14qu3dcfm8kT2R9/k5kaUoeh2ITxag6DOczYbXdds+gZBo/0u0mgkk88FcbfJLF9Mz0+DoObNGEkzFWxKYZAsqWup+qzIBaat9bJejl7TjZ0DFXLBoRsahZyFZbQZLzPvR7KM9uBIGW1dlerRm7Rb6+W2ZbTtmvA9du8JPXjnRtykrh/+XDLs0pVCPhjpBnty9ImvlOund0HYVZrMZhLBZgYk30j9B5sTyGwmTtzDrtn05cGDDuQd1cXzm3rm8vXu5WtRUNjPnE0pDBh9INkpWOl0kPnic1dVbzi9vctcr7T4/WNRS2il8IK/kAuiOZvDdaPNAn+nd6TM5snwZJ8LTJvHul8UFgtB3A6/3/WAyZtfre8Tf3x76fruRDrR+p9ZrTfipkTdLuBHWbNZ6NCNNjCpy2FoLN1o01LIB9qthF201zJcRusD9Vmfsxmvae5jX82SpZZgc2PGAy/vzOayXr6xKynMbK4vFw/dIMtFo0+yPhqoH6uHMpthE8TkGLl288VPrS2170ZbP7o86U33nNQv/envHug6xN8oHz6zOVoZba5lzabUHKPWTSmf67mMa9Fk7+y2gA7dKevzIni1XIgPDqmt2Ux830Fr5nPRyX97dzodWC+e29T2bkXfjsqA2+k32CwVcirkAt3aq8aBZKevKXRYs/nks69oY6WoN57d6PMVjI+/+E9rP5kVxWgERrsT4azw/5fjyGyeWF3qWS5ayue0Vwn3+X73n0LUjS/czpYGQfnmyXsSnWj99lRrDf3BC69pfbkYB9vtNLvRjnfOZvfM5uhltGlJvpYsX0z7m5uzntn0F/P+vDWJObTj4LM9r2yHgdk8NAiSwpv6V64fDjaT8rlA1agbbZYz//3w13j+OqdUyMX7Y7XeUKXeOHI8P7lW1q29qvajG1LesEtVWvnO5sNmNv/UH32jfvTt9w7985tltI0429vXms1C5+VUiyp7Z7cFlMz6DXLB4Utp0+rA5zObyYNOv+LMZpQVmWSDICnMbErSVy5f7/icOEvZx+98tVxoyWx2XrPZOnKlWm/oi9+4prddOD3x5kBSeAGTD2yhM5tSmJkLM5vjORFOgw/4Rvm/PBuVMfWTNSkWOmcpOzGzOChu/Zrk8e2OCV1I+xtAT73wuh6563jX0t3mbLbhy2jbrtnsFmzGDYKyV+KefC2rGW6AEpfRznhm0+9D335tR6vlwsxkapNltMV8MPNdgb07N5e1vVvRzkFVN3Yqh9ZrSlIhCEdc7FfrM19Gm49KhG/tVaI1+rlEZrPetumPP4dcayml7dZEcRBxGe2Q14/f/cY79PD540P//OToE78Gs5/zYCmfY85mi9m84pozw6zZlMJZm5JSHH0SvqkGXa8pNTMEzTLayV6onD95TMeW8nrmxW7BZn9rNqXwrt/tRGazcxnt0TWbf/DCa9o9qOkdUyih9ZaK+YXPbIbt+f1w6tn8XYyjJHqpkNP5EyvxzM1ukjdiBvmZfl9rvSmTPL5NMrP5yo1dXd3e06P3nOj63GY32sEDP//7aa16sD7LaKdxI6qX5GvJdmZzTuZsJtZs9hpLlCV+tNTV7T1trJTmpjeAv6F/5fVdbe8eHAkoc7kgbu4468GmFC6XurVXjUddHSqjrR6dL36qw6zNYWdZtxq1G+2ofBltrd5IlNH2k9nMMWezxWwfmedEslPrIG/Q1DOb0YXtoOs1peab1Hdwm/Qd58BMD53b7JrZ9OWu/fzOj5ULYUvwHpnNdms2f+drr6iUD/Tme0/2u/ljVy7myGxG6yhaB07PEh/EjXrj4GMffntfjcWS+0y/3Wil8NhRyFWPXHQm32uTGHvif6ZvdPbIXd3vco+yZvPE6pL+7A8/ou9+8I5Dj/dTRlvIBZm8QPe/j1Ih28cP//81K5nATvzve+egNjPrNaXEms3tPd3VpUx91sTB5vVdbe9WjpQHHyozn4Ngcy26zslZWAlVSAabbeZTn4zGmVxrWbfZrWP/IEbtRjsqvxys7lx8XdfPebCYp4y21Wxecc2Z5BtpkIuce0+tKjBpvUsr/1H4E/dwmc3DDYKWJ9wgSJIeOrepF67d1u39atvP9wock1ajzGavdZ6tczadc3ry2Vf0ljdsTfVibblEZrOYD7RzEO4Ls9ogqDSGMlopDIz6WUedfG8Mmtls10gheXF2ekLBpr/oWSsXdPep1a7PzY8QbErSD735riMXnUFgXTOb1Xpj5E60afHvkyw3B5KalTTzsmZTkk7NyHpNqXlsOKjW52LsieerL56/dkt7laOlssn37TxkNlfLxbgbbTEfKBeE6+8rvot7yzng5OqSTNK17WYZbb3RUMON5xx7cjUMZnuN6EpL7lAZbf+ZTcpoj5rtI/OcGLaM9p0Xz+hXTq3Gd5fGLS6jHeJCIx83CPJrNid/sXLxXLhu86uXr7edj9irs2zSarmgb1291bOMtnXO5jdevqlrN/f14Xc9MPD2j9PPfv9DEy9lzppiIaed/bCRwaw2CCqPoRvtIJL7+SDBZqnYfp23P74dP1aa2Gvwa4ceuftEz1LVOLM5xnm0JnXNbNYbjUw2B5KaQXfWszbzsmYzeYE+S5nN5DiItG5+T8NKqaD15aK+9u0bknRkzabvTSFl/4ZMP1bLBV27uafA7NBSiN2D9ufNfC7Q8dXSocymv/4ZR/XQPadW9Q9/7vsmtuSiVRxs1p1q6n/OZlhGS2YzKZtnuAUzbLAZmOmure536kcRl9EOk9mM3qS+QdA07jg/eHZDgXVuElStR51l+/idN9ds9mgQlM+pmrij9TvPviKTpjJfM+mt95+Kg+9FVcrn4iz3rGY2x9GNdhDFRHayn/li3lKHskv/XpvkxYP/mY/c3btRxKiZzXYCMzUaXUYw1d2hi9YsKcRZ4YwHm8G8ZDYTmf9ZCjYT7/V5ymxKYSntV78dXkO0jnRJHifmYbboql+zWWtmMYv5ID5vtruBuLVWPjT+pDpAxVg/phVoSs2bCfXGYGs2w/4QZDaTsnmGWzDFfG6kYeJpGWnNZlxGW1W5mOs5YiEN5WJe955e6xxs9jn6RArLS3YrNe1V/B2+znM2k91on3z2FT10fnNuWsHPsl4nzVkwjm60gygNmdk8ubqkzTb7vH+vTWq9ptQ8pj56d/fmQOFz7dDXjEM/DYIyW0abKEHOsnjN5owHm8n9bqYym3MebN6OKmJaA0r/vjVNr4nNOK0uFeJRJv7YX0zcpG0XQJ5aLx8qo40zmzN6jk3y1621AbvRFvM0CGo120fmObKylNeNnUqmMi4jdaNNZDanebf54rlNff6py6o3GnHTIm+QNZvHojWn13cOun6NXxjunNO1m/v6xss39aff88ZRXgLGpFTIacdnNjN0U2cQS1Msoy32Mcza+09+4GLbebP+4uzMBIPNN917Qi9d39E9PdZrSqM1COokCHo1CMpuGa0/zq3OQBltYDZQE6ssOrRmc4aCzeTxaGOOymilZpMgqU1mM7qmWC0XpnJDfdxWy0U1nNP1nYoejI73hXyQWH7SLrO5pCeffUXOOZlZfNwfR4OgafNJk0a0ZrOQC/r6fy4Vcqo1nOoNNxf7xTgQbGbESqmgGzuVTF10+LvEo3SjvbVXnWp3uovnNvXp331Bz1+9pfvuWD/0ufig2OfoE0l67ZYPNjuv2XQK74Q9+ewrkqS3PzC9kSdoKuZz8tWMs7pmc6WUl0l9dZIdh+TvaZAAt9P64PXlkt798J0THQP06N0n+spqSsky2vFdIARmcn10o80iv13rGS+j/aOPnNWp9XImO/oOwr/fSvlgphrOFOc6s9kcEXW0QdBsrGnul+/PcXu/2hzllAu6Zja31suq1Bpxt17f1yJLiZNhxZnNekOV2tHRL50k55POeofsceG3kBE++5eliw4fYA2z4D95sTbNzOZD58N1il+5fP1IsBkfFPv4na9GF1uv3w7LRTpnNnPx937y2Vd07sSKzs9RK/hZluyOmqX32SB+8LHzOn/y2MROYMnyuHGUHucC00d/7M0jf5+05FNYzmBm6rJkU7V6I7N3vwtxZjPbJYJvOL2mN5xem/ZmjMz/vrdmLHDOBxaP+JmHtYtJPrOZD+zItYy/zpmfYLP5Oor55prNbhVdW9E82Gs397WxUhr7ms1pau1G2+9SkmR3ZoLN0OzvDXPCjz/J0tqdu04e0y/++FuGykIkM7TT7IJ6er2sE6slfeXFo+s2Byqjje74Xb/dvYzWXyxs71T01POvkdXMkGSwNKvrSdaWixPdp5L7+TxcPPTig8zcOMtorXs32tosdKPNeGZzXvj9b5ZKaKXwhoqvfJi/zGYYbG6slI7cAPDvj9by2lmVnDxQKiTXbIZltO3Om35fvbYdNgmq1MfXjXbaAjMFFq3ZrDX6DjabmU060nqzvzfMiSxmNs1M33PxzFDblOyuOK2BvFL4Gh46t9m2SdBADYJaymg7BSv+e/27r72sWsNNtFwQ3S1a4DQOPkAv5YOZyrQMy79/C2O86WdzUEY7L5mbrDMzFXLBTHWi9fyF+CyV//bj+LGSSoVc29c1r2W0kuL1z8V8EHdibZ/ZDEfv+Y601TlqECSFS8L8ms1+O7L75x3QJChGfjcjlksFFXLzc0GXlTJaKVy3+VvPvKzXbu3rxGpzJml1gHlQ/iDsy2g7fY1fFP+bT1/R+nJRbzy72ONGsuRwZjObF/dZE68hm1BDoml7/NFz2lpbOtJMbBRBP2W0GapoSVqa0wAiy/7jxx+ayTFVpUKg5VJ+Zjt9d2JmOnd8RcdXj5YH+0aI85L5b5fZTJ4r2wVb68tFFXKBrt0Mr418L4x5OcfmAlO9Ea3Z7LNJnl/DXGH8SYxgMyM2VopTD8rGKZeRzKYkXYzWbT5z+bre+dCZ+PFKrd53gO/Xr27vVJQPrOMaK3+ife7Ktn7gTecyuxZrEc3Dms1JK0141Mq03bG5rPdu3jXW7xmYumY2q/VGZo/9l+7f0p/74Ud0/x2zvx5yVvzIW++Z9iYMpVTIaWNOz3d/4QOPtc3q+czmvNyMSY5vSTYI8toFkGYWjT/xmc0oCzon59hcYKo1nPYHWbMZ/Z4OKKONZfMMt4A++I436F0P3zntzRibw5nN6TaXuO+OdRVygb7SEmxW642+yynzuUBLhZz2q/Wud26T34/1mtlyaIzHnNx1TVsxMWsNwzHrPvqk3nCZXbNZzOf03jePN/jGfCoVclqZ8rk+LZ3GJs1bmXkxn4uvc0ptjv2dzptba0u6dqSMNpvHtEHlc4HqDadKtXEo89tNnNmkjDZGsJkRGyslbcxRF7dkRm9labq7WSEX6IE71480CarUGgMdEFfLBe1X612/ppBY5/CWN5wcboORinF3Vl0E/uJiUnM951HYIKjz56v1RlyOB8yq97/1noWrGNlaX9LGSlH3zVHmf225qP3tvUSDoGSvg/bnga31sn7/m69KSjZenI9zRhCY6vWGDqp1nVxb6v0FSqzZpIw2RrCJVCTv1GehROziuU196t9/K5qV5OvpGwMdEI8tFXTt5n7XrJj/fm++96SWaHmdKcn/t0W7KBqW/531u1YFR/VuEJTdbrRAv97z6Llpb8LEHT+2pF/9849PezPGanWpoKvbe/Exv5/GeltrS3rt1r5q9UZzzeacHNPygYWjT4aZs1mljNabj70BmRO2jA7v1mehtObi+U3VGk7PXdmOH6vWGwMdEH0JRbcAdTkKMOlCmz2U0Q7OzFTMBwuzZjMNQWDqEmsSbALIDH+dsxQ3COpdEbS1XpaT9Nqt/eb88jk5x/o1mwfVet8VPvGcTTKbsfnYG5BJft3mtBsESdJDUXe/ZClttda9JLaVXzzf7WvuPb2qv/KTl/T4mxbvLm/WlZKZzTk5EU5CMZ/ru+U7juo9Z9Nlar4ygMV1rOWmevKGfKebtKfWolmbN/fjzOa83NDNBdGazVo9XovZS7J6DqH52BuQSX7W5srS9DObGyslnT2+omcS8zbDMtoBgs34INz5a8xM73jw9FhHJ2A8kieKeSnxmQQym6MxMzW6LNqs1hpz07kRwGxbjca4tK7ZDEwdu+v7WZvXtvfiAGtezrF+9MlBtdH/nM2CL6Mls+nNx96ATMplKLMphdnNr1y+Hq+fSq7f7Ic/CM/LwvdFU0rcqZ2XebaTcPH8pi6cWZ/2ZsyswLqX0e4e1KbesRsApHDNpqQj3WiL+VzH8+bWepjZvHpzT9Vao2tgOmtygalaC9ei9nvTtRwtp9o5qKW5aTMlG1EA5pLPbGahQZAUXjR//qnLunJ9V3ceX1G13tBSof9t8wfheSkPWTT+/40S2sH8pQ9+x7Q3YaZ1K6Ot1Oqq1hsqF7mBBWD61pajYLMls9ntuqdczIcNFLf3VCrm5+qGbj4XaLcSBo2lPhvl5XOBji0VtL17kOamzRSuupAan9nMTLDp121GpbQDl9H6NZtzUh6yaNoNqQbS1q0b7V4lLLPKyjESwGK7dN8pveeRs9o8FlZyNce5db8hdmq9HK7ZrDUONRWadbnAtBtlKAepattYKerGTiWtzZo5XHUhNfnAVMgFmSk7vWvrmJZL+XjdZnWIOZuSWL82o5rlQBz2MDmBWcc5m/4ihjJaAFlwz6lV/YUPPBb3nSj0WRG0tbakq9t7AzdezLpksDnIvOmNlZJu7JDZ9OZnj0Dm5IJAK0vZuWMfmIXrNqOOtIOOPjlGGe1Mo4wW02Bdymj3oosYymgBZFExFx6bejXH2fKZzfpgFWNZN2xmc325qO1dMpve/OwRyJxCLsjEjM2ki+c29fzVW9rZrw7cIMh3o52nEpFF4rvR+pMnMAndymj9WiAymwCyqN+btKfWyrq9X9X2bmWulqrkgiAONgepaqOM9rD52SOQObmcZW4t0kPnNuUkffXbNwZes0mDoNlGZhPTkOtSRrtzUJWUnY7dAJDUT4MgqTn+5KXXdzOzdGoc8jmLZ4f22yBIkjaWS7q5W1G9y9irRcJVF1JTKuTidY5Z8eDZdQUWNgkads0mM/FmU2DhGmKCTUySb8rYLrvZXLNJsAkgewotI1A68eNPXr6xO1eZzSDRVXeQzOb6SlFO0q09spsSo0+Qov/8Bx9WPmMHnZVSQfecWtMzl6+Hmc0Btm+5VFC5mNPacjHFLUSaSoWAzDQmyo8AaDinXMs4AN+N1s9lA4AsKQyY2aw33FydY5PXsIN1oy1Jkm7sVOJ/L7Kp7BFmdtzMPmdmz0V/b3Z43nvN7Gtm9nUz+2ji8Z8ws6fNrGFmlya35RjE/WfWdc+p1WlvxhEXz4dNghrODbT+MheYPv6z36M/funuFLcOaSrmc3N11xXZ52ebt6um2o3KaMlsAsgif0O+V6B1YnUpPtbNU/VQLmjeIBysG22YlKAjbWhae8RHJX3BOXdB0heijw8xs5ykX5b0Q5IuSvopM7sYffoPJf2YpH87mc3FPLl4blP71TCjMOgduLMnVrREFmJmlQq5uVpPguzzmc1OZbQmaYlutAAyqN81m/lcoOOrYXZznm7oJoPNQa4XN5Z9sEkZrTS9YPP9kj4Z/fuTkj7Q5jnfKenrzrlvOucqkn41+jo5555xzn1tEhuK+XPxXDORPk934NDbux++U2+7cGram4EF4i9W2mU29yp1lYv5Q+uCACArBplP7Utp5+m6ys8blQbtRhuV0e6S2ZSmt2bztHPuiiQ5566YWburv7OSXkx8fFnS2wb9QWb2EUkfkaS77rpriE3FvDm9UdbmsZKu3z4gy7Vgnvi+B6e9CVgw3RoE7RxUKaEFkFmFuIt772ulU2tlPaMbc5vZHCTYPFYuKDAjsxlJbY8ws8+b2R+2+fP+fr9Fm8cG7iHsnPuEc+6Sc+7S1tbWoF+OOWRmcXZzng6KALInbhDUJrW5e1Aj2ASQWT6jWeojW3ky6kg7Tzfx87lkGW3/rysw0/pyUdu7BJtSiplN59z3d/qcmb1iZmeirOYZSVfbPO2ypPOJj89JemnMm4kFdfHcpn77qy/PVdc0ANkTWPcyWoJNAFnV7+gTSTo1l2W04fG7kAsOZTn7sb5S1DYNgiRNb83mr0t6Ivr3E5J+rc1zvijpgpnda2ZFSR+Kvg4Y2aP3nJAkxpgASFXQdc5mlbEnADKrlA/0XQ+e1sN3He/53K01n9mcp2CzvwZJ7WysFCmjjUxrj/iYpMfN7DlJj0cfy8zuNLPPSJJzribp5yR9VtIzkv6Fc+7p6Hk/amaXJb1D0v9hZp+dwmvADLtwZl1/7yPfo8eioBMA0pCcs9mKMloAWWZm+ss/eUlvecPJns/dispo52l5ks9mDrJe09tYLtEgKDKVs5xz7jVJ72nz+EuS3pf4+DOSPtPmeZ+S9Kk0txHz797Ta9PeBABzLohHnxz9HMEmgHlxag6DzfwowSaZzdj87BEAAGSM70bbLrO5V6lRRgtgLqyVC/oT33tB73zozLQ3ZWx8GW1piKZHGysl7R7UVKnVx71ZM4ezHAAAKQk6lNE658hsApgbZqYPv+uBaW/GWAVRZrNYGDw3t74S9gTZ3qnEJcaLiswmAAAp8RcrrYnNg2pdDSetEGwCQCb50SfDrtmUpBuMPyHYBAAgLZ3KaHcrNUmijBYAMipuEDRUGW2Y2bzB+BOCTQAA0hIoKqNtGbS5exAGm5TRAkA25f2azSEym+vLzTLaRUewCQBASiyes3n4cYJNAMi2eM3mUHM2fRktmU2CTQAAUtKpQRDBJgBkWz4a47I0RGazXMypmA8YfyKCTQAAUtNsEHQ42NxjzSYAZFou7kY7eLBpZtpYKRFsitEnAACkxuLM5uHHd6LMJt1oASCbcjZ8gyBJevzRc9paXxrnJs0kznIAAKQk6NSNljJaAMi03AijTyTpw++er7mjw6KMFgCAlPjMJmW0ADBbcnE3WsKlUfDbAwAgJUGHMtrdg5pygQ3V5RAAkL583I12uMwmQpzlAABISRCPPjlaRrtcyseZTwBAtvgGQcN0o0UTwSYAAClpltEefnyvUmO9JgBkWC4afUJmczQEmwAApMRnNuttMpus1wSA7PKZzWEbBCFEsAkAQEo6NQjyZbQAgGwqF8Mg89gSx+pREGwCAJCSIGhfRrtzUGPGJgBk2IN3bui//dAlPXzX8Wlvykwj2AQAICWd5mzuVSijBYAsMzO97cLpuKs4hkOwCQBASnwZbaNBGS0AYPEQbAIAkJJuczYJNgEA845gEwCAlFibOZv1htN+tU4ZLQBg7hFsAgCQkqBNN9r9Sk2SyGwCAOYewSYAAClpV0a7e0CwCQBYDASbAACkxNp0o92NMpsrlNECAOYcwSYAACmxNmW0PrNZJrMJAJhzBJsAAKQkRxktAGCBEWwCAJCSdt1o42CTMloAwJwj2AQAICXWJrO5V6GMFgCwGAg2AQBISeAbBDWOZjZXCDYBAHOOYBMAgJS0axC04xsEUUYLAJhzBJsAAKQkF/gy2mawuVepqZQPlM9xCgYAzDfOdAAApKSZ2Ww+tntQY70mAGAhEGwCAJAS34220dKNlrEnAIBFQLAJAEBKAmtfRsvYEwDAIiDYBAAgJc05m83HKKMFACwKgk0AAFLSLrO5e0BmEwCwGAg2AQBISdCuQVCFNZsAgMVAsAkAQEpoEAQAWGQEmwAApIQyWgDAIiPYBAAgJa1zNqv1hqr1BplNAMBCINgEACAlQXSW9ZnNvYOaJBFsAgAWAsEmAAApictoG2GwuRsFm4w+AQAsAoJNAABS0lpG64NN1mwCABYBwSYAACkJWrrR7lZ8GW1hWpsEAMDEEGwCAJCS5pzNw2W0y6Xc1LYJAIBJIdgEACAlFo8+CT+mjBYAsEgINgEASAlltACARUawCQBASqyljJbRJwCARUKwCQBAigI7Wka7VGTNJgBg/hFsAgCQosCs2SCoUtNyMR83DgIAYJ4RbAIAkCIzOzRnkxJaAMCiINgEACBFgUn1xOiTMiW0AIAFQbAJAECKrLWMlk60AIAFQbAJAECKgkNltFXKaAEAC4NgEwCAFAVBc87m3kGdMloAwMIg2AQAIEVmpkaDMloAwOIh2AQAIEWHRp9QRgsAWCAEmwAApMhMajjJOafdgzrBJgBgYRBsAgCQIp/ZPKg11HBOy0WCTQDAYiDYBAAgRYGZGi4soZVEZhMAsDAINgEASFFYRuu0d1CXRLAJAFgcBJsAAKTIojLa3UpNklSmjBYAsCAINgEASFEQNQiijBYAsGgINgEASFEQRJnNgzCzSbAJAFgUBJsAAKTI5BsEhcHmCsEmAGBBEGwCAJCiwKRGgzWbAIDFQ7AJAECK4gZBlNECABYMwSYAACkK52w67R3UlAtMxTynXgDAYuCMBwBAisIGQdJupaZyMS8zm/YmAQAwEQSbAACkyExqRGW0lNACABYJwSYAACnyZbQEmwCARUOwCQBAiswUl9Eu04kWALBAphJsmtlxM/ucmT0X/b3Z4XnvNbOvmdnXzeyjicf/lpl91cyeMrNPmdnGxDYeAIABkNkEACyqaWU2PyrpC865C5K+EH18iJnlJP2ypB+SdFHST5nZxejTn5P0R5xzj0p6VtJfnMhWAwAwoMCiBkEEmwCABTOtYPP9kj4Z/fuTkj7Q5jnfKenrzrlvOucqkn41+jo55/4v51wtet6Tks6lu7kAAAzHNwjaqxBsAgAWy7SCzdPOuSuSFP19qs1zzkp6MfHx5eixVj8j6f/s9IPM7CNm9rtm9rvXrl0bYZMBABhcsoy2zJpNAMACSe2sZ2afl3RHm0/9Yr/fos1jruVn/KKkmqR/0umbOOc+IekTknTp0iXX6XkAAKTBTKo1nPYqdTKbAICFktpZzzn3/Z0+Z2avmNkZ59wVMzsj6Wqbp12WdD7x8TlJLyW+xxOSfljSe5xzBJEAgEwKAtPuXlWSCDYBAAtlWmW0vy7piejfT0j6tTbP+aKkC2Z2r5kVJX0o+jqZ2Xsl/YKkH3HO7U5gewEAGEpgpp39KNikjBYAsECmFWx+TNLjZvacpMejj2Vmd5rZZyQpagD0c5I+K+kZSf/COfd09PUfl7Qq6XNm9iUz+5VJvwAAAPphZtrZD3vakdkEACySqZz1nHOvSXpPm8dfkvS+xMefkfSZNs+7P9UNBABgTAKTdisEmwCAxTOtzCYAAAvBrNnvjmATALBICDYBAEhRkAg2GX0CAFgkBJsAAKQoSAzyIrMJAFgkBJsAAKQoWUa7QrAJAFggBJsAAKQoEWtSRgsAWCgEmwAApCgXRZvFfKB8jtMuAGBxcNYDACBFvoyW9ZoAgEVDsAkAQIp8GS0ltACARUOwCQBAigIymwCABUWwCQBAiiijBQAsKoJNAABSFERn2mXKaAEAC4ZgEwCAFFFGCwBYVASbAACkiDJaAMCiItgEACBFQdSNljJaAMCiIdgEACBFPrNZJrMJAFgwBJsAAKSINZsAgEVFsAkAQIriMlqCTQDAgiHYBAAgRXGDINZsAgAWDMEmAAApIrMJAFhUBJsAAKSINZsAgEVFsAkAQIooowUALCqCTQAAUuTLaBl9AgBYNASbAACkyCijBQAsKIJNAABStL5c1LGlvMqU0QIAFgxnPgAAUvS+77hL73zoTNwoCACARUGwCQBAior5nE6u5aa9GQAATBxltAAAAACAsSPYBAAAAACMHcEmAAAAAGDsCDYBAAAAAGNHsAkAAAAAGDuCTQAAAADA2BFsAgAAAADGjmATAAAAADB2BJsAAAAAgLEj2AQAAAAAjB3BJgAAAABg7Ag2AQAAAABjR7AJAAAAABg7gk0AAAAAwNgRbAIAAAAAxs6cc9Pehokxs2uSXpjSjz8p6dUp/WxkD/sD2mG/QCfsG+iG/QPtsF+gnZOSVpxzW5P4YQsVbE6Tmf2uc+7StLcD2cD+gHbYL9AJ+wa6Yf9AO+wXaGfS+wVltAAAAACAsSPYBAAAAACMHcHm5Hxi2huATGF/QDvsF+iEfQPdsH+gHfYLtDPR/YI1mwAAAACAsSOzCQAAAAAYO4JNAAAAAMDYEWx2YGbnzew3zOwZM3vazP5s9PhxM/ucmT0X/b0ZPX4iev5tM/t4y/cqmtknzOxZM/uqmf14h5/5HWb2B2b2dTP7u2Zm0ePfa2a/Z2Y1M/tg2q8dR2Vsf/hPo8e/ZGa/ZWYX0379aC9j+8VPm9m1aL/4kpn9bNqvH+1lbL/4O4l94lkzu5Hyy0cfMraP3G1mXzCzp8zs/zGzc2m/frQ3pf3ivzezF83sdsvjXHtmxLj2CzNbTZwPvmRmr5rZ/9ThZ44tJiHY7Kwm6b92zj0k6e2S/oyFF/UflfQF59wFSV+IPpakfUn/jaSfb/O9flHSVefcA5IuSvrNDj/z70n6iKQL0Z/3Ro//B0k/LemfjviaMLws7Q//1Dn3iHPuMUl/U9LfHvG1YXhZ2i8k6Z875x6L/vz90V4aRpCZ/cI591/5fULS/yzp34z+8jAGmdlHJP0Pkv6Rc+5RSX9N0t8Y8bVheNPYLz4t6TvbPM61Z3aMZb9wzt1KXCM8JukFdT4njC0mIdjswDl3xTn3e9G/b0l6RtJZSe+X9MnoaZ+U9IHoOTvOud9S+B/c6mcUHbydcw3n3KutTzCzM5LWnHO/48KuTf8o8b2fd849JakxtheIgWRsf7iZeOqKJLp8TUmW9gtkR4b3i5+S9M+Gf2UYl4ztIxcVXqhK0m9E24ApmPR+EX3uSefclTaPc+2ZEWPeLyRJZnZB0ilJ/2+bz401JiHY7IOZ3SPpzZL+vaTT/k0Z/X2qx9duRP/861Ha+V+a2ek2Tz0r6XLi48vRY8iYLOwPZvZnzOwbCjOb/+WQLwVjlIX9QtKPR6Vw/8rMzg/3SjBOGdkvZGZ3S7pX0v89xMtAijKwj3xZki+x/FFJq2Z2YoiXgjGa0H6BGTPKftHipxRWQ7VLWIw1JiHY7MHMjkn615L+XEtGqV95Seck/bZz7i2SfkdhycqRH9XmMTJWGZOV/cE598vOufsk/YKkvzTEdmCMMrJffFrSPVEp3OfVvNuJKcnIfuF9SNK/cs7Vh9gOpCQj+8jPS3qXmf2+pHdJ+rbCsj1MyQT3C8yQMewXSR9S50qXscYkBJtdmFlB4X/qP3HO+ZrmV6L0sk8zX+3xbV6TtCvpU9HH/1LSW8wsl1ig+9cU3jVILso/J+mlMb0UjEFG94dfFWWUU5WV/cI595pz7iB6/H+T9B0jvjSMICv7RUK3CwtMQVb2EefcS865H3POvVnhOj8557ZHf4UYxoT3C8yIMe0X/nu9SVLeOff/RR+nGpMQbHYQdV363yU945xLNmD5dUlPRP9+QtKvdfs+UXr605LeHT30Hklfcc7VE4t0/3KU/r5lZm+Pfvaf7PW9MTlZ2h+iOnvvj0l6brRXh2FlbL84k/iWP6JwTQemIEv7RbQ9D0raVJjdQAZkaR8xs5Nm5q8H/6KkfzD6K8QwJr1fjHXjkZpx7RcJh9bvpx6TOOf40+aPpHcqTBk/JelL0Z/3STqhcCH9c9HfxxNf87yk1yXdVnhX4GL0+N2S/m30vb4g6a4OP/OSpD+U9A1JH5dk0eNvjb7fjsK7VU9P+/ezaH8ytj/8kqSno234DUkPT/v3s6h/MrZf/I1ov/hytF+8cdq/n0X9k6X9IvrcX5X0sWn/XviTzX1E0gejn/espL8vqTTt38+i/pnSfvE3o69rRH//1ehxrj0z8mec+0X0uW+qxzVCl+PFwPuF/0IAAAAAAMaGMloAAAAAwNgRbAIAAAAAxo5gEwAAAAAwdgSbAAAAAICxI9gEAAAAAIwdwSYAACMys3o0EPtpM/uymf35xOzCTl9zj5n9R5PaRgAAJo1gEwCA0e25cCD2w5IeVzgD7a/0+Jp7JBFsAgDmFnM2AQAYkZndds4dS3z8BklflHRS4XD1fyxpJfr0zznn/p2ZPSnpIUnfkvRJSX9X0sckvVtSSdIvO+f+14m9CAAAxoxgEwCAEbUGm9Fj1yW9UdItSQ3n3L6ZXZD0z5xzl8zs3ZJ+3jn3w9HzPyLplHPuvzOzkqTflvQTzrlvTfK1AAAwLvlpbwAAAHPKor8Lkj5uZo9Jqkt6oMPzf0DSo2b2wejjdUkXFGY+AQCYOQSbAACMWVRGW5d0VeHazVckvUlhr4T9Tl8m6b9wzn12IhsJAEDKaBAEAMAYmdmWpF+R9HEXrlVZl3TFOdeQ9GFJueiptyStJr70s5L+MzMrRN/nATNbEQAAM4rMJgAAoyub2ZcUlszWFDYE+tvR5/4XSf/azH5C0m9I2okef0pSzcy+LOkfSvolhR1qf8/MTNI1SR+YzOYDADB+NAgCAAAAAIwdZbQAAAAAgLEj2AQAAAAAjB3BJgAAAABg7Ag2AQAAAABjR7AJAAAAABg7gk0AAAAAwNgRbAIAAAAAxu7/BxYMJ61jhpDQAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 1080x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "i = 4\n",
    "df_plot = df_price_final[df_price_final['City'] == list_of_cities[i]]\n",
    "df_plot = df_plot[pd.DatetimeIndex(df_plot['Date']).year == 2016]\n",
    "df_plot = df_plot.sort_values(by='Date')\n",
    "#--------------------- Visualizing the RNN model results--------------------#\n",
    "plt.figure(figsize=(15, 7))\n",
    "plt.plot(df_plot['Date']\n",
    "         , df_plot['Return']\n",
    "         , color = '#4782B4'\n",
    "         , label = 'Return in {}'.format(list_of_cities[i])\n",
    "        )\n",
    "plt.xlabel(\"Date\")\n",
    "plt.ylabel('Price')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Preparing data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Inputs parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "output_metrics = 'Return'\n",
    "input_metrics = 'Raw Temperature'\n",
    "#input_metrics = 'Rolling 5'\n",
    "window_days = 10\n",
    "test_days = 50"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train_Corrected_Final = []\n",
    "X_test_Corrected_Final = []\n",
    "y_train_Final = []\n",
    "y_test_Final = []\n",
    "\n",
    "\n",
    "#for each city making train and test data sets\n",
    "for city in list_of_cities:\n",
    "    df_price_city = df_price_final[df_price_final['City'] == city]\n",
    "    df_price_city = df_price_city[['Date', output_metrics]]\n",
    "    df_price_city.reset_index(inplace=True, drop=True)\n",
    "\n",
    "    df_weather_city = df_weather_final[df_weather_final['Is{}'.format(city)] == 1]\n",
    "    df_weather_city.reset_index(inplace=True, drop=True)\n",
    "\n",
    "    # Dropping 'IsDallas' => because other 0 is IsDallas\n",
    "    df_weather_city = df_weather_city[['Date', input_metrics, 'IsSacramento', 'IsMinneapolis'\n",
    "                                   , 'IsLas Vegas', 'IsNew York', 'IsAmsterdam', 'IsCincinnati']]\n",
    "\n",
    "\n",
    "    # Train/Test start dates\n",
    "    # Find Train output start day\n",
    "    train_date_output_start = df_price_city['Date'].min()\n",
    "    # Find Train intput start day (remember input is LAST 'window_days' of days)\n",
    "    train_index_input_start = len(df_weather_city[df_weather_city['Date'] < train_date_output_start]) - window_days\n",
    "    train_date_input_start = df_weather_city['Date'][train_index_input_start]\n",
    "\n",
    "    # Test is LAST 'test_days' DAYS\n",
    "    test_date_output_start = df_price_city['Date'][len(df_price_city) - test_days]\n",
    "    # Find Train intput start day (remember input is LAST 'window_days' of days)\n",
    "    test_index_input_start = len(df_weather_city[df_weather_city['Date'] < test_date_output_start]) - window_days\n",
    "    test_date_input_start = df_weather_city['Date'][test_index_input_start]\n",
    "\n",
    "    test_date_end = df_price_city['Date'].max()\n",
    "\n",
    "\n",
    "    ##### Spliting datasets\n",
    "    ### Train\n",
    "    # input train data\n",
    "    df_train_input = df_weather_city[df_weather_city['Date'] >= train_date_input_start]\n",
    "    df_train_input = df_train_input[df_train_input['Date'] < test_date_output_start]\n",
    "    training_input_set = df_train_input.iloc[:,1:2].values\n",
    "    training_input_set_Dates = df_train_input.iloc[:,0:1].values\n",
    "    training_input_set_DUMMY = df_train_input.iloc[:,2:].values\n",
    "\n",
    "    # output train data\n",
    "    df_train_output = df_price_city[df_price_city['Date'] >= train_date_output_start]\n",
    "    df_train_output = df_train_output[df_train_output['Date'] < test_date_output_start]\n",
    "    training_output_set = df_train_output.iloc[:,1:2].values\n",
    "    training_output_set_Dates = df_train_output.iloc[:,0:1].values\n",
    "\n",
    "\n",
    "    ### Test\n",
    "    # input test data\n",
    "    df_test_input = df_weather_city[df_weather_city['Date'] >= test_date_input_start]\n",
    "    df_test_input = df_test_input[df_test_input['Date'] <= test_date_end]\n",
    "    testing_input_set = df_test_input.iloc[:,1:2].values\n",
    "    testing_input_set_Dates = df_test_input.iloc[:,0:1].values\n",
    "    testing_input_set_DUMMY = df_test_input.iloc[:,2:].values\n",
    "\n",
    "    # output test data\n",
    "    df_test_output = df_price_city[df_price_city['Date'] >= test_date_output_start]\n",
    "    df_test_output = df_test_output[df_test_output['Date'] <= test_date_end]\n",
    "    testing_output_set = df_test_output.iloc[:,1:2].values\n",
    "    testing_output_set_Dates = df_test_output.iloc[:,0:1].values\n",
    "\n",
    "\n",
    "    ### Feature scaling\n",
    "    sc_input, sc_output = MinMaxScaler(feature_range=(0,1)), MinMaxScaler(feature_range=(0,1))\n",
    "    # scaling inputs\n",
    "    training_input_set_scaled = sc_input.fit_transform(training_input_set)\n",
    "    testing_input_set_scaled = sc_input.transform(testing_input_set)\n",
    "\n",
    "    # scaling inputs\n",
    "    training_output_set_scaled = sc_output.fit_transform(training_output_set)\n",
    "    testing_output_set_scaled = sc_output.transform(testing_output_set)\n",
    "\n",
    "    \n",
    "    #creating a data structure with 90 timesteps + Dummy for city 8 (7 columns) and 1 output\n",
    "    X_train = []\n",
    "    X_train_Dates = []\n",
    "    y_train = []\n",
    "    y_train_Dates = []\n",
    "    X_test  = []\n",
    "    X_test_Dates  = []\n",
    "    y_test  = []\n",
    "    y_test_Dates  = []\n",
    "\n",
    "\n",
    "    # transforming dataset\n",
    "    for i in range(window_days, training_input_set_scaled.size):\n",
    "        X_train.append(np.append(training_input_set_DUMMY[0]\n",
    "                                 , training_input_set_scaled[i-window_days:i, 0], axis=0))\n",
    "        X_train_Dates.append(np.append(training_input_set_DUMMY[0]\n",
    "                                       , training_input_set_Dates[i-window_days:i, 0], axis=0))\n",
    "\n",
    "\n",
    "    for i in range(0, training_output_set_scaled.size):\n",
    "        y_train.append(training_output_set_scaled[i, 0])\n",
    "        y_train_Dates.append(training_output_set_Dates[i, 0])\n",
    "\n",
    "\n",
    "    for i in range(window_days, testing_input_set_scaled.size):\n",
    "        X_test.append(np.append(testing_input_set_DUMMY[0]\n",
    "                                , testing_input_set_scaled[i-window_days:i, 0], axis=0))\n",
    "        X_test_Dates.append(np.append(testing_input_set_DUMMY[0]\n",
    "                                      , testing_input_set_Dates[i-window_days:i, 0], axis=0))\n",
    "\n",
    "\n",
    "    for i in range(0, testing_output_set_scaled.size):\n",
    "        y_test.append(testing_output_set_scaled[i, 0])\n",
    "        y_test_Dates.append(testing_output_set_Dates[i, 0])\n",
    "\n",
    "\n",
    "    # Financial data has gaps in dates therefore we have to drop missed days from weahter (input) datsets\n",
    "    # Fixing of train set\n",
    "    y_train_Dates_Corrected = [] # output is forecast therefore input - output is one day and we \n",
    "                                 # should correct for that\n",
    "    X_train_Dates_Corrected = []\n",
    "    X_train_Corrected = []\n",
    "\n",
    "    for i in range(len(y_train_Dates)):\n",
    "        # taking the day before\n",
    "        y_train_Dates_Corrected.append(y_train_Dates[i] - dt.timedelta(days=1))\n",
    "\n",
    "    y_train_Dates_Corrected_set = set(y_train_Dates_Corrected)\n",
    "\n",
    "\n",
    "    for i in range(len(X_train_Dates)):\n",
    "        # cheking whether we have an output for that dates \n",
    "        if X_train_Dates[i][-1] in y_train_Dates_Corrected_set:\n",
    "            X_train_Dates_Corrected.append(X_train_Dates[i])\n",
    "            X_train_Corrected.append(X_train[i])\n",
    "\n",
    "    # Fixing of test set\n",
    "    y_test_Dates_Corrected = [] # output is forecast therefore input - output is one day and we \n",
    "                                # should correct for that\n",
    "    X_test_Dates_Corrected = []\n",
    "    X_test_Corrected = []\n",
    "\n",
    "\n",
    "    for i in range(len(y_test_Dates)):\n",
    "        # taking the day before\n",
    "        y_test_Dates_Corrected.append(y_test_Dates[i] - dt.timedelta(days=1))\n",
    "\n",
    "    y_test_Dates_Corrected_set = set(y_test_Dates_Corrected)\n",
    "\n",
    "    for i in range(len(X_test_Dates)):\n",
    "        # cheking whether we have an output for that dates \n",
    "        if X_test_Dates[i][-1] in y_test_Dates_Corrected_set:\n",
    "            X_test_Dates_Corrected.append(X_test_Dates[i])\n",
    "            X_test_Corrected.append(X_test[i]) \n",
    "    \n",
    "    # Adding Train/Test data sets of the city to the final dataset\n",
    "    X_train_Corrected_Final += X_train_Corrected\n",
    "    X_test_Corrected_Final += X_test_Corrected\n",
    "    y_train_Final += y_train\n",
    "    y_test_Final += y_test\n",
    "\n",
    "    \n",
    "    \n",
    "# transforming pandas lists to numpy arrays required for the RNN\n",
    "X_train, y_train = np.array(X_train_Corrected_Final), np.array(y_train_Final)\n",
    "X_test,  y_test  = np.array(X_test_Corrected_Final),  np.array(y_test_Final)\n",
    "\n",
    "# Shaping/adding new dimensions to allow adding more indicators: from 2D to 3D\n",
    "# 3 input arguments\n",
    "# batch_size: number of observations\n",
    "# timesteps: number of columns\n",
    "# input_dim: number of predictors\n",
    "X_train = np.reshape(X_train, [X_train.shape[0], X_train.shape[1], 1])\n",
    "X_test  = np.reshape(X_test,  [X_test.shape[0],  X_test.shape[1],  1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# All models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "74/74 [==============================] - 1s 3ms/step - loss: 0.0276\n",
      "Epoch 2/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 3/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 4/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 5/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0111\n",
      "Epoch 6/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 7/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0111\n",
      "Epoch 8/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 9/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0111\n",
      "Epoch 10/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 11/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0111\n",
      "Epoch 12/20\n",
      "74/74 [==============================] - 0s 2ms/step - loss: 0.0111\n",
      "Epoch 13/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 14/20\n",
      "74/74 [==============================] - 0s 2ms/step - loss: 0.0110\n",
      "Epoch 15/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0111\n",
      "Epoch 16/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0113\n",
      "Epoch 17/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 18/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0111\n",
      "Epoch 19/20\n",
      "74/74 [==============================] - 0s 2ms/step - loss: 0.0111\n",
      "Epoch 20/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0112\n",
      "11/11 [==============================] - 0s 2ms/step - loss: 0.0072\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0111\n",
      "Epoch 1/20\n",
      "74/74 [==============================] - 2s 6ms/step - loss: 0.0482\n",
      "Epoch 2/20\n",
      "74/74 [==============================] - 1s 7ms/step - loss: 0.0131\n",
      "Epoch 3/20\n",
      "74/74 [==============================] - 1s 7ms/step - loss: 0.0117\n",
      "Epoch 4/20\n",
      "74/74 [==============================] - 0s 6ms/step - loss: 0.0119\n",
      "Epoch 5/20\n",
      "74/74 [==============================] - 0s 7ms/step - loss: 0.0113\n",
      "Epoch 6/20\n",
      "74/74 [==============================] - 0s 6ms/step - loss: 0.0112\n",
      "Epoch 7/20\n",
      "74/74 [==============================] - 0s 6ms/step - loss: 0.0111\n",
      "Epoch 8/20\n",
      "74/74 [==============================] - 0s 6ms/step - loss: 0.0111\n",
      "Epoch 9/20\n",
      "74/74 [==============================] - 0s 6ms/step - loss: 0.0112\n",
      "Epoch 10/20\n",
      "74/74 [==============================] - 1s 7ms/step - loss: 0.0111\n",
      "Epoch 11/20\n",
      "74/74 [==============================] - 0s 6ms/step - loss: 0.0111\n",
      "Epoch 12/20\n",
      "74/74 [==============================] - 0s 7ms/step - loss: 0.0110\n",
      "Epoch 13/20\n",
      "74/74 [==============================] - 0s 7ms/step - loss: 0.0112\n",
      "Epoch 14/20\n",
      "74/74 [==============================] - 0s 6ms/step - loss: 0.0111\n",
      "Epoch 15/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0111\n",
      "Epoch 16/20\n",
      "74/74 [==============================] - 1s 7ms/step - loss: 0.0111\n",
      "Epoch 17/20\n",
      "74/74 [==============================] - 1s 9ms/step - loss: 0.0111\n",
      "Epoch 18/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0111\n",
      "Epoch 19/20\n",
      "74/74 [==============================] - 1s 9ms/step - loss: 0.0112\n",
      "Epoch 20/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0111\n",
      "11/11 [==============================] - 0s 4ms/step - loss: 0.0071\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 1/20\n",
      "74/74 [==============================] - 4s 14ms/step - loss: 0.0348\n",
      "Epoch 2/20\n",
      "74/74 [==============================] - 1s 14ms/step - loss: 0.0148\n",
      "Epoch 3/20\n",
      "74/74 [==============================] - 1s 14ms/step - loss: 0.0127\n",
      "Epoch 4/20\n",
      "74/74 [==============================] - 1s 14ms/step - loss: 0.0126\n",
      "Epoch 5/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0124\n",
      "Epoch 6/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0123\n",
      "Epoch 7/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0125\n",
      "Epoch 8/20\n",
      "74/74 [==============================] - 1s 15ms/step - loss: 0.0122\n",
      "Epoch 9/20\n",
      "74/74 [==============================] - 1s 15ms/step - loss: 0.0123\n",
      "Epoch 10/20\n",
      "74/74 [==============================] - 1s 14ms/step - loss: 0.0121\n",
      "Epoch 11/20\n",
      "74/74 [==============================] - 1s 14ms/step - loss: 0.0121\n",
      "Epoch 12/20\n",
      "74/74 [==============================] - 1s 12ms/step - loss: 0.0119\n",
      "Epoch 13/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0119\n",
      "Epoch 14/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0117\n",
      "Epoch 15/20\n",
      "74/74 [==============================] - 1s 12ms/step - loss: 0.0119\n",
      "Epoch 16/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0118\n",
      "Epoch 17/20\n",
      "74/74 [==============================] - 1s 14ms/step - loss: 0.0117\n",
      "Epoch 18/20\n",
      "74/74 [==============================] - 1s 14ms/step - loss: 0.0116\n",
      "Epoch 19/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0119\n",
      "Epoch 20/20\n",
      "74/74 [==============================] - 1s 12ms/step - loss: 0.0117\n",
      "11/11 [==============================] - 1s 5ms/step - loss: 0.0071\n",
      "74/74 [==============================] - 0s 5ms/step - loss: 0.0109\n",
      "Epoch 1/20\n",
      "74/74 [==============================] - 7s 24ms/step - loss: 0.0243\n",
      "Epoch 2/20\n",
      "74/74 [==============================] - 2s 24ms/step - loss: 0.0131\n",
      "Epoch 3/20\n",
      "74/74 [==============================] - 2s 28ms/step - loss: 0.0124\n",
      "Epoch 4/20\n",
      "74/74 [==============================] - 2s 26ms/step - loss: 0.0128\n",
      "Epoch 5/20\n",
      "74/74 [==============================] - 2s 27ms/step - loss: 0.0122\n",
      "Epoch 6/20\n",
      "74/74 [==============================] - 2s 26ms/step - loss: 0.0121\n",
      "Epoch 7/20\n",
      "74/74 [==============================] - 2s 26ms/step - loss: 0.0124\n",
      "Epoch 8/20\n",
      "74/74 [==============================] - 2s 25ms/step - loss: 0.0122\n",
      "Epoch 9/20\n",
      "74/74 [==============================] - 2s 23ms/step - loss: 0.0126\n",
      "Epoch 10/20\n",
      "74/74 [==============================] - 2s 26ms/step - loss: 0.0124\n",
      "Epoch 11/20\n",
      "74/74 [==============================] - 2s 27ms/step - loss: 0.0119\n",
      "Epoch 12/20\n",
      "74/74 [==============================] - 2s 28ms/step - loss: 0.0118\n",
      "Epoch 13/20\n",
      "74/74 [==============================] - 2s 26ms/step - loss: 0.0119\n",
      "Epoch 14/20\n",
      "74/74 [==============================] - 2s 29ms/step - loss: 0.0120\n",
      "Epoch 15/20\n",
      "74/74 [==============================] - 2s 26ms/step - loss: 0.0116\n",
      "Epoch 16/20\n",
      "74/74 [==============================] - 2s 26ms/step - loss: 0.0115\n",
      "Epoch 17/20\n",
      "74/74 [==============================] - 2s 28ms/step - loss: 0.0116\n",
      "Epoch 18/20\n",
      "74/74 [==============================] - 2s 27ms/step - loss: 0.0118\n",
      "Epoch 19/20\n",
      "74/74 [==============================] - 2s 24ms/step - loss: 0.0117\n",
      "Epoch 20/20\n",
      "74/74 [==============================] - 2s 26ms/step - loss: 0.0118\n",
      "11/11 [==============================] - 2s 9ms/step - loss: 0.0071\n",
      "74/74 [==============================] - 1s 10ms/step - loss: 0.0109\n",
      "Epoch 1/20\n",
      "74/74 [==============================] - 17s 59ms/step - loss: 0.0245\n",
      "Epoch 2/20\n",
      "74/74 [==============================] - 5s 66ms/step - loss: 0.0135\n",
      "Epoch 3/20\n",
      "74/74 [==============================] - 5s 64ms/step - loss: 0.0125\n",
      "Epoch 4/20\n",
      "74/74 [==============================] - 5s 67ms/step - loss: 0.0125\n",
      "Epoch 5/20\n",
      "74/74 [==============================] - 5s 64ms/step - loss: 0.0129\n",
      "Epoch 6/20\n",
      "74/74 [==============================] - 5s 64ms/step - loss: 0.0121\n",
      "Epoch 7/20\n",
      "74/74 [==============================] - 5s 69ms/step - loss: 0.0124\n",
      "Epoch 8/20\n",
      "74/74 [==============================] - 5s 67ms/step - loss: 0.0120\n",
      "Epoch 9/20\n",
      "74/74 [==============================] - 5s 67ms/step - loss: 0.0121\n",
      "Epoch 10/20\n",
      "74/74 [==============================] - 5s 66ms/step - loss: 0.0119\n",
      "Epoch 11/20\n",
      "74/74 [==============================] - 5s 69ms/step - loss: 0.0125\n",
      "Epoch 12/20\n",
      "74/74 [==============================] - 5s 64ms/step - loss: 0.0119\n",
      "Epoch 13/20\n",
      "74/74 [==============================] - 5s 64ms/step - loss: 0.0118\n",
      "Epoch 14/20\n",
      "74/74 [==============================] - 5s 65ms/step - loss: 0.0118\n",
      "Epoch 15/20\n",
      "74/74 [==============================] - 5s 64ms/step - loss: 0.0118\n",
      "Epoch 16/20\n",
      "74/74 [==============================] - 5s 64ms/step - loss: 0.0116\n",
      "Epoch 17/20\n",
      "74/74 [==============================] - 5s 72ms/step - loss: 0.0118\n",
      "Epoch 18/20\n",
      "74/74 [==============================] - 5s 73ms/step - loss: 0.0116\n",
      "Epoch 19/20\n",
      "74/74 [==============================] - 5s 74ms/step - loss: 0.0119\n",
      "Epoch 20/20\n",
      "74/74 [==============================] - 6s 75ms/step - loss: 0.0118\n",
      "11/11 [==============================] - 4s 20ms/step - loss: 0.0071\n",
      "74/74 [==============================] - 1s 19ms/step - loss: 0.0111\n",
      "Epoch 1/20\n",
      "74/74 [==============================] - 1s 3ms/step - loss: 0.0368\n",
      "Epoch 2/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0112\n",
      "Epoch 3/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 4/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0111\n",
      "Epoch 5/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 6/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 7/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0111\n",
      "Epoch 8/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 9/20\n",
      "74/74 [==============================] - 0s 2ms/step - loss: 0.0110\n",
      "Epoch 10/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0111\n",
      "Epoch 11/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0111\n",
      "Epoch 12/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0111\n",
      "Epoch 13/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0111\n",
      "Epoch 14/20\n",
      "74/74 [==============================] - 0s 2ms/step - loss: 0.0110\n",
      "Epoch 15/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 16/20\n",
      "74/74 [==============================] - 0s 2ms/step - loss: 0.0114\n",
      "Epoch 17/20\n",
      "74/74 [==============================] - 0s 2ms/step - loss: 0.0110\n",
      "Epoch 18/20\n",
      "74/74 [==============================] - 0s 2ms/step - loss: 0.0110\n",
      "Epoch 19/20\n",
      "74/74 [==============================] - 0s 2ms/step - loss: 0.0112\n",
      "Epoch 20/20\n",
      "74/74 [==============================] - 0s 2ms/step - loss: 0.0110\n",
      "11/11 [==============================] - 0s 2ms/step - loss: 0.0071\n",
      "74/74 [==============================] - 0s 2ms/step - loss: 0.0110\n",
      "Epoch 1/20\n",
      "74/74 [==============================] - 2s 7ms/step - loss: 0.0392\n",
      "Epoch 2/20\n",
      "74/74 [==============================] - 0s 6ms/step - loss: 0.0127\n",
      "Epoch 3/20\n",
      "74/74 [==============================] - 0s 7ms/step - loss: 0.0116\n",
      "Epoch 4/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0114\n",
      "Epoch 5/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0114\n",
      "Epoch 6/20\n",
      "74/74 [==============================] - 1s 7ms/step - loss: 0.0112\n",
      "Epoch 7/20\n",
      "74/74 [==============================] - 1s 7ms/step - loss: 0.0117\n",
      "Epoch 8/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0111\n",
      "Epoch 9/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0111\n",
      "Epoch 10/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0111\n",
      "Epoch 11/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0112\n",
      "Epoch 12/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0110\n",
      "Epoch 13/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0111\n",
      "Epoch 14/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0112\n",
      "Epoch 15/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0113\n",
      "Epoch 16/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0111\n",
      "Epoch 17/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0111\n",
      "Epoch 18/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0111\n",
      "Epoch 19/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0111\n",
      "Epoch 20/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0110\n",
      "11/11 [==============================] - 0s 5ms/step - loss: 0.0071\n",
      "74/74 [==============================] - 0s 4ms/step - loss: 0.0111\n",
      "Epoch 1/20\n",
      "74/74 [==============================] - 4s 13ms/step - loss: 0.0361\n",
      "Epoch 2/20\n",
      "74/74 [==============================] - 1s 12ms/step - loss: 0.0147\n",
      "Epoch 3/20\n",
      "74/74 [==============================] - 1s 14ms/step - loss: 0.0127\n",
      "Epoch 4/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0127\n",
      "Epoch 5/20\n",
      "74/74 [==============================] - 1s 12ms/step - loss: 0.0126\n",
      "Epoch 6/20\n",
      "74/74 [==============================] - 1s 12ms/step - loss: 0.0124\n",
      "Epoch 7/20\n",
      "74/74 [==============================] - 1s 12ms/step - loss: 0.0120\n",
      "Epoch 8/20\n",
      "74/74 [==============================] - 1s 12ms/step - loss: 0.0122\n",
      "Epoch 9/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0121\n",
      "Epoch 10/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0122\n",
      "Epoch 11/20\n",
      "74/74 [==============================] - 1s 14ms/step - loss: 0.0120\n",
      "Epoch 12/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0118\n",
      "Epoch 13/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0119\n",
      "Epoch 14/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0120\n",
      "Epoch 15/20\n",
      "74/74 [==============================] - 1s 14ms/step - loss: 0.0116\n",
      "Epoch 16/20\n",
      "74/74 [==============================] - 1s 19ms/step - loss: 0.0120\n",
      "Epoch 17/20\n",
      "74/74 [==============================] - 1s 17ms/step - loss: 0.0118\n",
      "Epoch 18/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0117\n",
      "Epoch 19/20\n",
      "74/74 [==============================] - 1s 14ms/step - loss: 0.0118\n",
      "Epoch 20/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0116\n",
      "11/11 [==============================] - 1s 5ms/step - loss: 0.0073\n",
      "74/74 [==============================] - 0s 4ms/step - loss: 0.0111\n",
      "Epoch 1/20\n",
      "74/74 [==============================] - 8s 27ms/step - loss: 0.0284\n",
      "Epoch 2/20\n",
      "74/74 [==============================] - 2s 27ms/step - loss: 0.0135\n",
      "Epoch 3/20\n",
      "74/74 [==============================] - 2s 24ms/step - loss: 0.0128\n",
      "Epoch 4/20\n",
      "74/74 [==============================] - 2s 24ms/step - loss: 0.0122\n",
      "Epoch 5/20\n",
      "74/74 [==============================] - 2s 28ms/step - loss: 0.0127\n",
      "Epoch 6/20\n",
      "74/74 [==============================] - 2s 26ms/step - loss: 0.0123\n",
      "Epoch 7/20\n",
      "74/74 [==============================] - 2s 26ms/step - loss: 0.0121\n",
      "Epoch 8/20\n",
      "74/74 [==============================] - 2s 27ms/step - loss: 0.0123\n",
      "Epoch 9/20\n",
      "74/74 [==============================] - 2s 29ms/step - loss: 0.0117\n",
      "Epoch 10/20\n",
      "74/74 [==============================] - 2s 27ms/step - loss: 0.0118\n",
      "Epoch 11/20\n",
      "74/74 [==============================] - 2s 25ms/step - loss: 0.0119\n",
      "Epoch 12/20\n",
      "74/74 [==============================] - 2s 28ms/step - loss: 0.0119\n",
      "Epoch 13/20\n",
      "74/74 [==============================] - 2s 24ms/step - loss: 0.0118\n",
      "Epoch 14/20\n",
      "74/74 [==============================] - 2s 26ms/step - loss: 0.0121\n",
      "Epoch 15/20\n",
      "74/74 [==============================] - 2s 23ms/step - loss: 0.0118\n",
      "Epoch 16/20\n",
      "74/74 [==============================] - 2s 24ms/step - loss: 0.0116\n",
      "Epoch 17/20\n",
      "74/74 [==============================] - 2s 28ms/step - loss: 0.0116\n",
      "Epoch 18/20\n",
      "74/74 [==============================] - 2s 25ms/step - loss: 0.0117\n",
      "Epoch 19/20\n",
      "74/74 [==============================] - 2s 23ms/step - loss: 0.0117\n",
      "Epoch 20/20\n",
      "74/74 [==============================] - 2s 22ms/step - loss: 0.0115\n",
      "11/11 [==============================] - 3s 9ms/step - loss: 0.0071\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0109\n",
      "Epoch 1/20\n",
      "74/74 [==============================] - 16s 65ms/step - loss: 0.0268\n",
      "Epoch 2/20\n",
      "74/74 [==============================] - 5s 64ms/step - loss: 0.0131\n",
      "Epoch 3/20\n",
      "74/74 [==============================] - 5s 61ms/step - loss: 0.0124\n",
      "Epoch 4/20\n",
      "74/74 [==============================] - 5s 62ms/step - loss: 0.0127\n",
      "Epoch 5/20\n",
      "74/74 [==============================] - 5s 63ms/step - loss: 0.0123\n",
      "Epoch 6/20\n",
      "74/74 [==============================] - 4s 59ms/step - loss: 0.0123\n",
      "Epoch 7/20\n",
      "74/74 [==============================] - 5s 62ms/step - loss: 0.0120\n",
      "Epoch 8/20\n",
      "74/74 [==============================] - 5s 64ms/step - loss: 0.0120\n",
      "Epoch 9/20\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "74/74 [==============================] - 4s 60ms/step - loss: 0.0121\n",
      "Epoch 10/20\n",
      "74/74 [==============================] - 4s 60ms/step - loss: 0.0120\n",
      "Epoch 11/20\n",
      "74/74 [==============================] - 4s 56ms/step - loss: 0.0119\n",
      "Epoch 12/20\n",
      "74/74 [==============================] - 4s 58ms/step - loss: 0.0120\n",
      "Epoch 13/20\n",
      "74/74 [==============================] - 5s 70ms/step - loss: 0.0121\n",
      "Epoch 14/20\n",
      "74/74 [==============================] - 5s 61ms/step - loss: 0.0122\n",
      "Epoch 15/20\n",
      "74/74 [==============================] - 5s 66ms/step - loss: 0.0116\n",
      "Epoch 16/20\n",
      "74/74 [==============================] - 4s 59ms/step - loss: 0.0120\n",
      "Epoch 17/20\n",
      "74/74 [==============================] - 4s 58ms/step - loss: 0.0120\n",
      "Epoch 18/20\n",
      "74/74 [==============================] - 4s 58ms/step - loss: 0.0118\n",
      "Epoch 19/20\n",
      "74/74 [==============================] - 5s 61ms/step - loss: 0.0119\n",
      "Epoch 20/20\n",
      "74/74 [==============================] - 5s 64ms/step - loss: 0.0120\n",
      "11/11 [==============================] - 4s 18ms/step - loss: 0.0070\n",
      "74/74 [==============================] - 1s 18ms/step - loss: 0.0110\n",
      "Epoch 1/20\n",
      "74/74 [==============================] - 3s 3ms/step - loss: 0.0632\n",
      "Epoch 2/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0125\n",
      "Epoch 3/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0111\n",
      "Epoch 4/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 5/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 6/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 7/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 8/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 9/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 10/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 11/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 12/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 13/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 14/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 15/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 16/20\n",
      "74/74 [==============================] - 0s 3ms/step - loss: 0.0110\n",
      "Epoch 17/20\n",
      "74/74 [==============================] - 0s 2ms/step - loss: 0.0111\n",
      "Epoch 18/20\n",
      "74/74 [==============================] - 0s 2ms/step - loss: 0.0110\n",
      "Epoch 19/20\n",
      "74/74 [==============================] - 0s 2ms/step - loss: 0.0111\n",
      "Epoch 20/20\n",
      "74/74 [==============================] - 0s 2ms/step - loss: 0.0110\n",
      "11/11 [==============================] - 0s 2ms/step - loss: 0.0071\n",
      "74/74 [==============================] - 0s 2ms/step - loss: 0.0110\n",
      "Epoch 1/20\n",
      "74/74 [==============================] - 1s 6ms/step - loss: 0.0369\n",
      "Epoch 2/20\n",
      "74/74 [==============================] - 0s 6ms/step - loss: 0.0126\n",
      "Epoch 3/20\n",
      "74/74 [==============================] - 0s 6ms/step - loss: 0.0115\n",
      "Epoch 4/20\n",
      "74/74 [==============================] - 0s 7ms/step - loss: 0.0112\n",
      "Epoch 5/20\n",
      "74/74 [==============================] - 0s 6ms/step - loss: 0.0116\n",
      "Epoch 6/20\n",
      "74/74 [==============================] - 0s 6ms/step - loss: 0.0114\n",
      "Epoch 7/20\n",
      "74/74 [==============================] - 1s 7ms/step - loss: 0.0112\n",
      "Epoch 8/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0112\n",
      "Epoch 9/20\n",
      "74/74 [==============================] - 1s 7ms/step - loss: 0.0114\n",
      "Epoch 10/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0111\n",
      "Epoch 11/20\n",
      "74/74 [==============================] - 1s 7ms/step - loss: 0.0111\n",
      "Epoch 12/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0112\n",
      "Epoch 13/20\n",
      "74/74 [==============================] - 1s 7ms/step - loss: 0.0113\n",
      "Epoch 14/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0113\n",
      "Epoch 15/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0111\n",
      "Epoch 16/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0110\n",
      "Epoch 17/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0111\n",
      "Epoch 18/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0111\n",
      "Epoch 19/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0112\n",
      "Epoch 20/20\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0111\n",
      "11/11 [==============================] - 0s 4ms/step - loss: 0.0072\n",
      "74/74 [==============================] - 0s 5ms/step - loss: 0.0110\n",
      "Epoch 1/20\n",
      "74/74 [==============================] - 4s 13ms/step - loss: 0.0308\n",
      "Epoch 2/20\n",
      "74/74 [==============================] - 1s 12ms/step - loss: 0.0144\n",
      "Epoch 3/20\n",
      "74/74 [==============================] - 1s 12ms/step - loss: 0.0128\n",
      "Epoch 4/20\n",
      "74/74 [==============================] - 1s 14ms/step - loss: 0.0124\n",
      "Epoch 5/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0126\n",
      "Epoch 6/20\n",
      "74/74 [==============================] - 1s 15ms/step - loss: 0.0124\n",
      "Epoch 7/20\n",
      "74/74 [==============================] - 1s 14ms/step - loss: 0.0122\n",
      "Epoch 8/20\n",
      "74/74 [==============================] - 1s 14ms/step - loss: 0.0123\n",
      "Epoch 9/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0122\n",
      "Epoch 10/20\n",
      "74/74 [==============================] - 1s 12ms/step - loss: 0.0119\n",
      "Epoch 11/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0119\n",
      "Epoch 12/20\n",
      "74/74 [==============================] - 1s 16ms/step - loss: 0.0120\n",
      "Epoch 13/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0121\n",
      "Epoch 14/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0121\n",
      "Epoch 15/20\n",
      "74/74 [==============================] - 1s 15ms/step - loss: 0.0120\n",
      "Epoch 16/20\n",
      "74/74 [==============================] - 1s 15ms/step - loss: 0.0117\n",
      "Epoch 17/20\n",
      "74/74 [==============================] - 1s 15ms/step - loss: 0.0119\n",
      "Epoch 18/20\n",
      "74/74 [==============================] - 1s 14ms/step - loss: 0.0117\n",
      "Epoch 19/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0117\n",
      "Epoch 20/20\n",
      "74/74 [==============================] - 1s 13ms/step - loss: 0.0118\n",
      "11/11 [==============================] - 1s 5ms/step - loss: 0.0072\n",
      "74/74 [==============================] - 0s 5ms/step - loss: 0.0109\n",
      "Epoch 1/20\n",
      "74/74 [==============================] - 7s 24ms/step - loss: 0.0381\n",
      "Epoch 2/20\n",
      "74/74 [==============================] - 2s 22ms/step - loss: 0.0136\n",
      "Epoch 3/20\n",
      "74/74 [==============================] - 2s 23ms/step - loss: 0.0126\n",
      "Epoch 4/20\n",
      "74/74 [==============================] - 2s 22ms/step - loss: 0.0123\n",
      "Epoch 5/20\n",
      "74/74 [==============================] - 2s 23ms/step - loss: 0.0124\n",
      "Epoch 6/20\n",
      "74/74 [==============================] - 2s 23ms/step - loss: 0.0123\n",
      "Epoch 7/20\n",
      "74/74 [==============================] - 2s 23ms/step - loss: 0.0123\n",
      "Epoch 8/20\n",
      "74/74 [==============================] - 2s 24ms/step - loss: 0.0121\n",
      "Epoch 9/20\n",
      "74/74 [==============================] - 2s 25ms/step - loss: 0.0119\n",
      "Epoch 10/20\n",
      "74/74 [==============================] - 2s 28ms/step - loss: 0.0117\n",
      "Epoch 11/20\n",
      "74/74 [==============================] - 2s 24ms/step - loss: 0.0121\n",
      "Epoch 12/20\n",
      "74/74 [==============================] - 2s 23ms/step - loss: 0.0121\n",
      "Epoch 13/20\n",
      "74/74 [==============================] - 2s 27ms/step - loss: 0.0120\n",
      "Epoch 14/20\n",
      "74/74 [==============================] - 2s 26ms/step - loss: 0.0119\n",
      "Epoch 15/20\n",
      "74/74 [==============================] - 2s 23ms/step - loss: 0.0118\n",
      "Epoch 16/20\n",
      "74/74 [==============================] - 2s 23ms/step - loss: 0.0120\n",
      "Epoch 17/20\n",
      "74/74 [==============================] - 2s 24ms/step - loss: 0.0117\n",
      "Epoch 18/20\n",
      "74/74 [==============================] - 2s 25ms/step - loss: 0.0120\n",
      "Epoch 19/20\n",
      "74/74 [==============================] - 2s 25ms/step - loss: 0.0119\n",
      "Epoch 20/20\n",
      "74/74 [==============================] - 2s 24ms/step - loss: 0.0116\n",
      "11/11 [==============================] - 1s 9ms/step - loss: 0.0071\n",
      "74/74 [==============================] - 1s 8ms/step - loss: 0.0109\n",
      "Epoch 1/20\n",
      "74/74 [==============================] - 18s 66ms/step - loss: 0.0273\n",
      "Epoch 2/20\n",
      "74/74 [==============================] - 5s 62ms/step - loss: 0.0129\n",
      "Epoch 3/20\n",
      "74/74 [==============================] - 5s 61ms/step - loss: 0.0123\n",
      "Epoch 4/20\n",
      "74/74 [==============================] - 4s 59ms/step - loss: 0.0126\n",
      "Epoch 5/20\n",
      "74/74 [==============================] - 4s 55ms/step - loss: 0.0125\n",
      "Epoch 6/20\n",
      "74/74 [==============================] - 4s 58ms/step - loss: 0.0121\n",
      "Epoch 7/20\n",
      "74/74 [==============================] - 4s 56ms/step - loss: 0.0124\n",
      "Epoch 8/20\n",
      "74/74 [==============================] - 4s 55ms/step - loss: 0.0124\n",
      "Epoch 9/20\n",
      "74/74 [==============================] - 4s 55ms/step - loss: 0.0121\n",
      "Epoch 10/20\n",
      "74/74 [==============================] - 4s 57ms/step - loss: 0.0120\n",
      "Epoch 11/20\n",
      "74/74 [==============================] - 4s 60ms/step - loss: 0.0118\n",
      "Epoch 12/20\n",
      "74/74 [==============================] - 4s 59ms/step - loss: 0.0119\n",
      "Epoch 13/20\n",
      "74/74 [==============================] - 4s 59ms/step - loss: 0.0124\n",
      "Epoch 14/20\n",
      "74/74 [==============================] - 4s 59ms/step - loss: 0.0123\n",
      "Epoch 15/20\n",
      "74/74 [==============================] - 4s 60ms/step - loss: 0.0119\n",
      "Epoch 16/20\n",
      "74/74 [==============================] - 4s 59ms/step - loss: 0.0120\n",
      "Epoch 17/20\n",
      "74/74 [==============================] - 4s 57ms/step - loss: 0.0117\n",
      "Epoch 18/20\n",
      "74/74 [==============================] - 5s 61ms/step - loss: 0.0115\n",
      "Epoch 19/20\n",
      "74/74 [==============================] - 4s 56ms/step - loss: 0.0119\n",
      "Epoch 20/20\n",
      "74/74 [==============================] - 4s 55ms/step - loss: 0.0117\n",
      "11/11 [==============================] - 3s 17ms/step - loss: 0.0075\n",
      "74/74 [==============================] - 1s 18ms/step - loss: 0.0116\n"
     ]
    }
   ],
   "source": [
    "#epochs_num_list = [5, 10, 20, 60, 100, 150]\n",
    "epochs_num_list = [20, 20, 20]\n",
    "\n",
    "Results = pd.DataFrame({ 'Forecast metric': []\n",
    "                        , 'Input metric': []\n",
    "                        , 'Number of epochs': []\n",
    "                        , 'Number of input days': []\n",
    "                        , 'Model Type': []\n",
    "                        , 'Loss train': []\n",
    "                        , 'Loss test': []\n",
    "                        , 'End Date-Time': []\n",
    "                        , 'Model path': []\n",
    "                       })\n",
    "\n",
    "for epochs_num in epochs_num_list:\n",
    "\n",
    "    ##########################    Model NN    ##########################\n",
    "    ####################################################################\n",
    "    regressor_NN = Sequential()\n",
    "\n",
    "    # Laier plus dropout\n",
    "    regressor_NN.add(Dense(units = 50, input_shape = (X_train.shape[1], 1)))\n",
    "\n",
    "    # Output\n",
    "    regressor_NN.add(Dense(units = 1))\n",
    "\n",
    "    # Comipling\n",
    "    regressor_NN.compile(optimizer = 'adam', loss = 'mean_squared_error')\n",
    "\n",
    "    # Fitting\n",
    "    regressor_NN.fit(X_train, y_train, epochs = epochs_num, batch_size = 32)\n",
    "\n",
    "    # Testing\n",
    "    NN_Loss_test  = round(regressor_NN.evaluate(X_test,  y_test ), 4)\n",
    "    NN_Loss_train = round(regressor_NN.evaluate(X_train, y_train), 4)\n",
    "\n",
    "\n",
    "    # Saving\n",
    "    datetime_now = dt.datetime.now().strftime(\"%Y-%m-%d %H:%M\")\n",
    "    NN_model_path = 'Models/NN model | Input {} | Epochs {} | Days {} | Loss train {} | Loss test {} | Created at {}.h5'.format(\n",
    "        input_metrics, epochs_num, window_days, NN_Loss_train, NN_Loss_test, datetime_now)\n",
    "\n",
    "    # Saving model\n",
    "    regressor_NN.save(NN_model_path)\n",
    "    # Saving resultes\n",
    "    Results = Results.append(pd.DataFrame({ 'Forecast metric': [output_metrics]\n",
    "                                            , 'Input metric': [input_metrics]\n",
    "                                            , 'Number of epochs': [epochs_num]\n",
    "                                            , 'Number of input days': [window_days]\n",
    "                                            , 'Model Type': ['NN']\n",
    "                                            , 'Loss train': [NN_Loss_train]\n",
    "                                            , 'Loss test': [NN_Loss_test]\n",
    "                                            , 'End Date-Time': [datetime_now]\n",
    "                                            , 'Model path': [NN_model_path]\n",
    "                                           }))\n",
    "    \n",
    "    \n",
    "    \n",
    "    ##########################    Model SimpleRNN    ##########################\n",
    "    ###########################################################################\n",
    "    regressor_RNN = Sequential()\n",
    "\n",
    "    # Laier plus dropout\n",
    "    regressor_RNN.add(SimpleRNN(units = 50, return_sequences=True, input_shape = (X_train.shape[1], 1)))\n",
    "    #regressor_RNN.add(Dropout(0.2))\n",
    "\n",
    "    # Output\n",
    "    regressor_RNN.add(Dense(units = 1))\n",
    "\n",
    "    # Comipling\n",
    "    regressor_RNN.compile(optimizer = 'adam', loss = 'mean_squared_error')\n",
    "\n",
    "    # Fitting\n",
    "    regressor_RNN.fit(X_train, y_train, epochs = epochs_num, batch_size = 32)\n",
    "\n",
    "    # Testing\n",
    "    RNN_Loss_test  = round(regressor_RNN.evaluate(X_test,  y_test ), 4)\n",
    "    RNN_Loss_train = round(regressor_RNN.evaluate(X_train, y_train), 4)\n",
    "\n",
    "    # Saving\n",
    "    datetime_now = dt.datetime.now().strftime(\"%Y-%m-%d %H:%M\")\n",
    "    RNN_model_path = 'Models/RNN model | Input {} | Epochs {} | Days {} | Loss train {} | Loss test {} | Created at {}.h5'.format(\n",
    "        input_metrics, epochs_num, window_days, RNN_Loss_train, RNN_Loss_test, datetime_now)\n",
    "\n",
    "    # Saving model\n",
    "    regressor_RNN.save(RNN_model_path)\n",
    "    # Saving resultes\n",
    "    Results = Results.append(pd.DataFrame({ 'Forecast metric': [output_metrics]\n",
    "                                            , 'Input metric': [input_metrics]\n",
    "                                            , 'Number of epochs': [epochs_num]\n",
    "                                            , 'Number of input days': [window_days]\n",
    "                                            , 'Model Type': ['SimpleRNN']\n",
    "                                            , 'Loss train': [RNN_Loss_train]\n",
    "                                            , 'Loss test': [RNN_Loss_test]\n",
    "                                            , 'End Date-Time': [datetime_now]\n",
    "                                            , 'Model path': [RNN_model_path]\n",
    "                                           }))\n",
    "\n",
    "    \n",
    "    \n",
    "    ##########################    Model LSTM    ##########################\n",
    "    ######################################################################\n",
    "    regressor_LSTM = Sequential()\n",
    "\n",
    "    # Laier plus dropout\n",
    "    regressor_LSTM.add(LSTM(units = 50, input_shape = (X_train.shape[1], 1)))\n",
    "    regressor_LSTM.add(Dropout(0.2))\n",
    "\n",
    "    # Output\n",
    "    regressor_LSTM.add(Dense(units = 1))\n",
    "\n",
    "    # Comipling\n",
    "    regressor_LSTM.compile(optimizer = 'adam', loss = 'mean_squared_error')\n",
    "\n",
    "    # Fitting\n",
    "    regressor_LSTM.fit(X_train, y_train, epochs = epochs_num, batch_size = 32)\n",
    "\n",
    "    # Testing\n",
    "    LSTM_Loss_test  = round(regressor_LSTM.evaluate(X_test,  y_test ), 4)\n",
    "    LSTM_Loss_train = round(regressor_LSTM.evaluate(X_train, y_train), 4)\n",
    "\n",
    "    # Saving\n",
    "    datetime_now = dt.datetime.now().strftime(\"%Y-%m-%d %H:%M\")\n",
    "    LSTM_model_path = 'Models/LSTM model | Input {} | Epochs {} | Days {} | Loss train {} | Loss test {} | Created at {}.h5'.format(\n",
    "        input_metrics, epochs_num, window_days, LSTM_Loss_train, LSTM_Loss_test, datetime_now)\n",
    "\n",
    "    # Saving model\n",
    "    regressor_LSTM.save(LSTM_model_path)\n",
    "    # Saving resultes\n",
    "    Results = Results.append(pd.DataFrame({ 'Forecast metric': [output_metrics]\n",
    "                                            , 'Input metric': [input_metrics]\n",
    "                                            , 'Number of epochs': [epochs_num]\n",
    "                                            , 'Number of input days': [window_days]\n",
    "                                            , 'Model Type': ['LSTM']\n",
    "                                            , 'Loss train': [LSTM_Loss_train]\n",
    "                                            , 'Loss test': [LSTM_Loss_test]\n",
    "                                            , 'End Date-Time': [datetime_now]\n",
    "                                            , 'Model path': [LSTM_model_path]\n",
    "                                           }))\n",
    "    \n",
    "    \n",
    "    \n",
    "    ##########################    Model DLSTM    ##########################\n",
    "    #######################################################################\n",
    "    regressor_DLSTM = Sequential()\n",
    "\n",
    "    # Laier plus dropout\n",
    "    regressor_DLSTM.add(LSTM(units = 50, return_sequences=True, input_shape = (X_train.shape[1], 1)))\n",
    "    regressor_DLSTM.add(Dropout(0.2))\n",
    "\n",
    "    # Second Layer plus dropout\n",
    "    regressor_DLSTM.add(LSTM(units = 50))\n",
    "    regressor_DLSTM.add(Dropout(0.2))\n",
    "\n",
    "    # Output\n",
    "    regressor_DLSTM.add(Dense(units = 1))\n",
    "\n",
    "    # Comipling\n",
    "    regressor_DLSTM.compile(optimizer = 'adam', loss = 'mean_squared_error')\n",
    "\n",
    "    # Fitting\n",
    "    regressor_DLSTM.fit(X_train, y_train, epochs = epochs_num, batch_size = 32)\n",
    "\n",
    "    # Testing\n",
    "    DLSTM_Loss_test  = round(regressor_DLSTM.evaluate(X_test,  y_test ), 4)\n",
    "    DLSTM_Loss_train = round(regressor_DLSTM.evaluate(X_train, y_train), 4)\n",
    "\n",
    "    # Saving\n",
    "    datetime_now = dt.datetime.now().strftime(\"%Y-%m-%d %H:%M\")\n",
    "    DLSTM_model_path = 'Models/DLSTM model | Input {} | Epochs {} | Days {} | Loss train {} | Loss test {} | Created at {}.h5'.format(\n",
    "        input_metrics, epochs_num, window_days, DLSTM_Loss_train, DLSTM_Loss_test, datetime_now)\n",
    "\n",
    "    # Saving model\n",
    "    regressor_DLSTM.save(DLSTM_model_path)\n",
    "    # Saving resultes\n",
    "    Results = Results.append(pd.DataFrame({ 'Forecast metric': [output_metrics]\n",
    "                                            , 'Input metric': [input_metrics]\n",
    "                                            , 'Number of epochs': [epochs_num]\n",
    "                                            , 'Number of input days': [window_days]\n",
    "                                            , 'Model Type': ['DLSTM']\n",
    "                                            , 'Loss train': [DLSTM_Loss_train]\n",
    "                                            , 'Loss test': [DLSTM_Loss_test]\n",
    "                                            , 'End Date-Time': [datetime_now]\n",
    "                                            , 'Model path': [DLSTM_model_path]\n",
    "                                           }))\n",
    "    \n",
    "    \n",
    "    \n",
    "    ##########################    Model D-DLSTM    ##########################\n",
    "    #########################################################################\n",
    "    regressor_DDLSTM = Sequential()\n",
    "\n",
    "    # First Layer plus dropout\n",
    "    regressor_DDLSTM.add(LSTM(units = 50, return_sequences=True, input_shape = (X_train.shape[1], 1)))\n",
    "    regressor_DDLSTM.add(Dropout(0.2))\n",
    "\n",
    "    # Second Layer plus dropout\n",
    "    regressor_DDLSTM.add(LSTM(units = 50, return_sequences=True))\n",
    "    regressor_DDLSTM.add(Dropout(0.2))\n",
    "\n",
    "    # Third Layer plus dropout\n",
    "    regressor_DDLSTM.add(LSTM(units = 50, return_sequences=True))\n",
    "    regressor_DDLSTM.add(Dropout(0.2))\n",
    "\n",
    "    # Forth Layer plus dropout\n",
    "    regressor_DDLSTM.add(LSTM(units = 50, return_sequences=True))\n",
    "    regressor_DDLSTM.add(Dropout(0.2))\n",
    "\n",
    "    # Fifth Layer plus dropout\n",
    "    regressor_DDLSTM.add(LSTM(units = 50))\n",
    "    regressor_DDLSTM.add(Dropout(0.2))\n",
    "\n",
    "    # Output\n",
    "    regressor_DDLSTM.add(Dense(units = 1))\n",
    "\n",
    "    # Comipling\n",
    "    regressor_DDLSTM.compile(optimizer = 'adam', loss = 'mean_squared_error')\n",
    "\n",
    "    # Fitting\n",
    "    regressor_DDLSTM.fit(X_train, y_train, epochs = epochs_num, batch_size = 32)\n",
    "    \n",
    "    # Testing\n",
    "    DDLSTM_Loss_test  = round(regressor_DDLSTM.evaluate(X_test,  y_test ), 4)\n",
    "    DDLSTM_Loss_train = round(regressor_DDLSTM.evaluate(X_train, y_train), 4)\n",
    "\n",
    "    # Saving\n",
    "    datetime_now = dt.datetime.now().strftime(\"%Y-%m-%d %H:%M\")\n",
    "    DDLSTM_model_path = 'Models/DDLSTM model | Input {} | Epochs {} | Days {} | Loss train {} | Loss test {} | Created at {}.h5'.format(\n",
    "        input_metrics, epochs_num, window_days, DDLSTM_Loss_train, DDLSTM_Loss_test, datetime_now)\n",
    "\n",
    "    # Saving model\n",
    "    regressor_DDLSTM.save(DDLSTM_model_path)\n",
    "    # Saving resultes\n",
    "    Results = Results.append(pd.DataFrame({ 'Forecast metric': [output_metrics]\n",
    "                                            , 'Input metric': [input_metrics]\n",
    "                                            , 'Number of epochs': [epochs_num]\n",
    "                                            , 'Number of input days': [window_days]\n",
    "                                            , 'Model Type': ['DDLSTM']\n",
    "                                            , 'Loss train': [DDLSTM_Loss_train]\n",
    "                                            , 'Loss test': [DDLSTM_Loss_test]\n",
    "                                            , 'End Date-Time': [datetime_now]\n",
    "                                            , 'Model path': [DDLSTM_model_path]\n",
    "                                           }))\n",
    "    \n",
    "Results.reset_index(inplace=True, drop=True)\n",
    "datetime_now = dt.datetime.now().strftime(\"%Y-%m-%d %H:%M\")\n",
    "Results.to_excel('Results Excel/Input {} | Epochs 20 | Days {} | Created at {}.xlsx'.format(\n",
    "    input_metrics, window_days, datetime_now))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Unnamed: 0</th>\n",
       "      <th>Forecast metric</th>\n",
       "      <th>Input metric</th>\n",
       "      <th>Number of epochs</th>\n",
       "      <th>Number of input days</th>\n",
       "      <th>Model Type</th>\n",
       "      <th>Loss train</th>\n",
       "      <th>Loss test</th>\n",
       "      <th>End Date-Time</th>\n",
       "      <th>Model path</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>9</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>0.0110</td>\n",
       "      <td>0.0070</td>\n",
       "      <td>2022-05-30 19:47</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "      <td>SimpleRNN</td>\n",
       "      <td>0.0110</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>2022-05-30 19:40</td>\n",
       "      <td>Models/RNN model | Input Raw Temperature | Epo...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "      <td>LSTM</td>\n",
       "      <td>0.0109</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>2022-05-30 19:41</td>\n",
       "      <td>Models/LSTM model | Input Raw Temperature | Ep...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "      <td>DLSTM</td>\n",
       "      <td>0.0109</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>2022-05-30 19:42</td>\n",
       "      <td>Models/DLSTM model | Input Raw Temperature | E...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>0.0111</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>2022-05-30 19:43</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>5</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "      <td>NN</td>\n",
       "      <td>0.0110</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>2022-05-30 19:44</td>\n",
       "      <td>Models/NN model | Input Raw Temperature | Epoc...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>6</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "      <td>SimpleRNN</td>\n",
       "      <td>0.0111</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>2022-05-30 19:44</td>\n",
       "      <td>Models/RNN model | Input Raw Temperature | Epo...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>8</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "      <td>DLSTM</td>\n",
       "      <td>0.0109</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>2022-05-30 19:45</td>\n",
       "      <td>Models/DLSTM model | Input Raw Temperature | E...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>10</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "      <td>NN</td>\n",
       "      <td>0.0110</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>2022-05-30 19:47</td>\n",
       "      <td>Models/NN model | Input Raw Temperature | Epoc...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>13</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "      <td>DLSTM</td>\n",
       "      <td>0.0109</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>2022-05-30 19:48</td>\n",
       "      <td>Models/DLSTM model | Input Raw Temperature | E...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "      <td>NN</td>\n",
       "      <td>0.0111</td>\n",
       "      <td>0.0072</td>\n",
       "      <td>2022-05-30 19:40</td>\n",
       "      <td>Models/NN model | Input Raw Temperature | Epoc...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>11</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "      <td>SimpleRNN</td>\n",
       "      <td>0.0110</td>\n",
       "      <td>0.0072</td>\n",
       "      <td>2022-05-30 19:47</td>\n",
       "      <td>Models/RNN model | Input Raw Temperature | Epo...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>12</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "      <td>LSTM</td>\n",
       "      <td>0.0109</td>\n",
       "      <td>0.0072</td>\n",
       "      <td>2022-05-30 19:48</td>\n",
       "      <td>Models/LSTM model | Input Raw Temperature | Ep...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>7</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "      <td>LSTM</td>\n",
       "      <td>0.0111</td>\n",
       "      <td>0.0073</td>\n",
       "      <td>2022-05-30 19:44</td>\n",
       "      <td>Models/LSTM model | Input Raw Temperature | Ep...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>14</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>0.0116</td>\n",
       "      <td>0.0075</td>\n",
       "      <td>2022-05-30 19:50</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    Unnamed: 0 Forecast metric     Input metric  Number of epochs  \\\n",
       "9            9          Return  Raw Temperature                20   \n",
       "1            1          Return  Raw Temperature                20   \n",
       "2            2          Return  Raw Temperature                20   \n",
       "3            3          Return  Raw Temperature                20   \n",
       "4            4          Return  Raw Temperature                20   \n",
       "5            5          Return  Raw Temperature                20   \n",
       "6            6          Return  Raw Temperature                20   \n",
       "8            8          Return  Raw Temperature                20   \n",
       "10          10          Return  Raw Temperature                20   \n",
       "13          13          Return  Raw Temperature                20   \n",
       "0            0          Return  Raw Temperature                20   \n",
       "11          11          Return  Raw Temperature                20   \n",
       "12          12          Return  Raw Temperature                20   \n",
       "7            7          Return  Raw Temperature                20   \n",
       "14          14          Return  Raw Temperature                20   \n",
       "\n",
       "    Number of input days Model Type  Loss train  Loss test     End Date-Time  \\\n",
       "9                     10     DDLSTM      0.0110     0.0070  2022-05-30 19:47   \n",
       "1                     10  SimpleRNN      0.0110     0.0071  2022-05-30 19:40   \n",
       "2                     10       LSTM      0.0109     0.0071  2022-05-30 19:41   \n",
       "3                     10      DLSTM      0.0109     0.0071  2022-05-30 19:42   \n",
       "4                     10     DDLSTM      0.0111     0.0071  2022-05-30 19:43   \n",
       "5                     10         NN      0.0110     0.0071  2022-05-30 19:44   \n",
       "6                     10  SimpleRNN      0.0111     0.0071  2022-05-30 19:44   \n",
       "8                     10      DLSTM      0.0109     0.0071  2022-05-30 19:45   \n",
       "10                    10         NN      0.0110     0.0071  2022-05-30 19:47   \n",
       "13                    10      DLSTM      0.0109     0.0071  2022-05-30 19:48   \n",
       "0                     10         NN      0.0111     0.0072  2022-05-30 19:40   \n",
       "11                    10  SimpleRNN      0.0110     0.0072  2022-05-30 19:47   \n",
       "12                    10       LSTM      0.0109     0.0072  2022-05-30 19:48   \n",
       "7                     10       LSTM      0.0111     0.0073  2022-05-30 19:44   \n",
       "14                    10     DDLSTM      0.0116     0.0075  2022-05-30 19:50   \n",
       "\n",
       "                                           Model path  \n",
       "9   Models/DDLSTM model | Input Raw Temperature | ...  \n",
       "1   Models/RNN model | Input Raw Temperature | Epo...  \n",
       "2   Models/LSTM model | Input Raw Temperature | Ep...  \n",
       "3   Models/DLSTM model | Input Raw Temperature | E...  \n",
       "4   Models/DDLSTM model | Input Raw Temperature | ...  \n",
       "5   Models/NN model | Input Raw Temperature | Epoc...  \n",
       "6   Models/RNN model | Input Raw Temperature | Epo...  \n",
       "8   Models/DLSTM model | Input Raw Temperature | E...  \n",
       "10  Models/NN model | Input Raw Temperature | Epoc...  \n",
       "13  Models/DLSTM model | Input Raw Temperature | E...  \n",
       "0   Models/NN model | Input Raw Temperature | Epoc...  \n",
       "11  Models/RNN model | Input Raw Temperature | Epo...  \n",
       "12  Models/LSTM model | Input Raw Temperature | Ep...  \n",
       "7   Models/LSTM model | Input Raw Temperature | Ep...  \n",
       "14  Models/DDLSTM model | Input Raw Temperature | ...  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Loading models\n",
    "excel_file_path = 'Results Excel/Input Raw Temperature | Epochs 20 | Days 10 | Created at 2022-05-30 19:54.xlsx'\n",
    "Models = pd.read_excel(excel_file_path)\n",
    "Models = Models.sort_values(by='Loss test')\n",
    "Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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>Forecast metric</th>\n",
       "      <th>Input metric</th>\n",
       "      <th>Number of epochs</th>\n",
       "      <th>Number of input days</th>\n",
       "      <th>Model Type</th>\n",
       "      <th>Loss train</th>\n",
       "      <th>Loss test</th>\n",
       "      <th>End Date-Time</th>\n",
       "      <th>Model path</th>\n",
       "      <th>Unnamed: 0</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>NN</td>\n",
       "      <td>0.0110</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>2022-05-30 19:44</td>\n",
       "      <td>Models/NN model | Input Raw Temperature | Epoc...</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>SimpleRNN</td>\n",
       "      <td>0.0110</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>2022-05-30 19:40</td>\n",
       "      <td>Models/RNN model | Input Raw Temperature | Epo...</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>DLSTM</td>\n",
       "      <td>0.0109</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>2022-05-30 19:42</td>\n",
       "      <td>Models/DLSTM model | Input Raw Temperature | E...</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>LSTM</td>\n",
       "      <td>0.0109</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>2022-05-30 19:41</td>\n",
       "      <td>Models/LSTM model | Input Raw Temperature | Ep...</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>20.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>0.0110</td>\n",
       "      <td>0.0070</td>\n",
       "      <td>2022-05-30 19:47</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "      <td>9.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Forecast metric     Input metric  Number of epochs  Number of input days  \\\n",
       "0          Return  Raw Temperature              20.0                  10.0   \n",
       "1          Return  Raw Temperature              20.0                  10.0   \n",
       "2          Return  Raw Temperature              20.0                  10.0   \n",
       "3          Return  Raw Temperature              20.0                  10.0   \n",
       "4          Return  Raw Temperature              20.0                  10.0   \n",
       "\n",
       "  Model Type  Loss train  Loss test     End Date-Time  \\\n",
       "0         NN      0.0110     0.0071  2022-05-30 19:44   \n",
       "1  SimpleRNN      0.0110     0.0071  2022-05-30 19:40   \n",
       "2      DLSTM      0.0109     0.0071  2022-05-30 19:42   \n",
       "3       LSTM      0.0109     0.0071  2022-05-30 19:41   \n",
       "4     DDLSTM      0.0110     0.0070  2022-05-30 19:47   \n",
       "\n",
       "                                          Model path  Unnamed: 0  \n",
       "0  Models/NN model | Input Raw Temperature | Epoc...         5.0  \n",
       "1  Models/RNN model | Input Raw Temperature | Epo...         1.0  \n",
       "2  Models/DLSTM model | Input Raw Temperature | E...         3.0  \n",
       "3  Models/LSTM model | Input Raw Temperature | Ep...         2.0  \n",
       "4  Models/DDLSTM model | Input Raw Temperature | ...         9.0  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "best_models = pd.DataFrame({ 'Forecast metric': []\n",
    "                        , 'Input metric': []\n",
    "                        , 'Number of epochs': []\n",
    "                        , 'Number of input days': []\n",
    "                        , 'Model Type': []\n",
    "                        , 'Loss train': []\n",
    "                        , 'Loss test': []\n",
    "                        , 'End Date-Time': []\n",
    "                        , 'Model path': []\n",
    "                       })\n",
    "\n",
    "for model in set(Models['Model Type']):\n",
    "    \n",
    "    Results_model = Models[Models['Model Type'] == model]\n",
    "    min_score = Results_model['Loss test'].min()\n",
    "    Results_model = Results_model[Results_model['Loss test'] == min_score]\n",
    "    min_index = Results_model.index.min()\n",
    "        \n",
    "    best_models = best_models.append(Results_model[Results_model.index == min_index])\n",
    "\n",
    "best_models.reset_index(inplace=True, drop=True)\n",
    "best_models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_9\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " lstm_11 (LSTM)              (None, 16, 50)            10400     \n",
      "                                                                 \n",
      " dropout_11 (Dropout)        (None, 16, 50)            0         \n",
      "                                                                 \n",
      " lstm_12 (LSTM)              (None, 16, 50)            20200     \n",
      "                                                                 \n",
      " dropout_12 (Dropout)        (None, 16, 50)            0         \n",
      "                                                                 \n",
      " lstm_13 (LSTM)              (None, 16, 50)            20200     \n",
      "                                                                 \n",
      " dropout_13 (Dropout)        (None, 16, 50)            0         \n",
      "                                                                 \n",
      " lstm_14 (LSTM)              (None, 16, 50)            20200     \n",
      "                                                                 \n",
      " dropout_14 (Dropout)        (None, 16, 50)            0         \n",
      "                                                                 \n",
      " lstm_15 (LSTM)              (None, 50)                20200     \n",
      "                                                                 \n",
      " dropout_15 (Dropout)        (None, 50)                0         \n",
      "                                                                 \n",
      " dense_11 (Dense)            (None, 1)                 51        \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 91,251\n",
      "Trainable params: 91,251\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model_path = best_models['Model path'][4]\n",
    "Regressor_Model = load_model(model_path)\n",
    "Regressor_Model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Calculating MSE, RMSE, MSPE, MAPE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "def MSE(y_test_true, y_test_predicted):\n",
    "    if len(y_test_true) != len(y_test_predicted):\n",
    "        return 'Error: Different length'\n",
    "    \n",
    "    elif len(y_test_true) == 0:\n",
    "        return 'Error: Zero length'\n",
    "    \n",
    "    else:\n",
    "        n = len(y_test_true)\n",
    "        total_sqr_error = 0.0\n",
    "        for i in range(n):\n",
    "            sqr_error = (float(y_test_true[i]) - float(y_test_predicted[i]))**2\n",
    "            total_sqr_error += sqr_error\n",
    "            \n",
    "        return total_sqr_error/n\n",
    "\n",
    "\n",
    "from math import  sqrt\n",
    "\n",
    "def RMSE(y_test_true, y_test_predicted):\n",
    "    if len(y_test_true) != len(y_test_predicted):\n",
    "        return 'Error: Different length'\n",
    "    \n",
    "    elif len(y_test_true) == 0:\n",
    "        return 'Error: Zero length'\n",
    "    \n",
    "    else:\n",
    "        n = len(y_test_true)\n",
    "        total_sqr_error = 0.0\n",
    "        for i in range(n):\n",
    "            sqr_error = (float(y_test_true[i]) - float(y_test_predicted[i]))**2\n",
    "            total_sqr_error += sqr_error\n",
    "            \n",
    "        return sqrt(total_sqr_error/n) \n",
    "\n",
    "    \n",
    "def MSPE(y_test_true, y_test_predicted):\n",
    "    if len(y_test_true) != len(y_test_predicted):\n",
    "        return 'Error: Different length'\n",
    "    \n",
    "    elif len(y_test_true) == 0:\n",
    "        return 'Error: Zero length'\n",
    "    \n",
    "    else:\n",
    "        n = len(y_test_true)\n",
    "        total_sqr_percentage_error = 0.0\n",
    "        for i in range(n):\n",
    "            sqr_percentage_error = (float(y_test_predicted[i])/float(y_test_true[i]) - 1)**2\n",
    "            total_sqr_percentage_error += sqr_percentage_error\n",
    "            \n",
    "        return total_sqr_percentage_error/n\n",
    "\n",
    "\n",
    "def MAPE(y_test_true, y_test_predicted):\n",
    "    if len(y_test_true) != len(y_test_predicted):\n",
    "        return 'Error: Different length'\n",
    "    \n",
    "    elif len(y_test_true) == 0:\n",
    "        return 'Error: Zero length'\n",
    "    \n",
    "    else:\n",
    "        n = len(y_test_true)\n",
    "        total_percentage_error = 0.0\n",
    "        for i in range(n):\n",
    "            percentage_error = abs(float(y_test_predicted[i])/float(y_test_true[i]) - 1)\n",
    "            total_percentage_error += percentage_error\n",
    "            \n",
    "        return total_percentage_error/n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "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>Model Type</th>\n",
       "      <th>MSE Error</th>\n",
       "      <th>RMSE Error</th>\n",
       "      <th>MSPE Error</th>\n",
       "      <th>MAPE Error</th>\n",
       "      <th>Model path</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NN</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>0.0844</td>\n",
       "      <td>0.0846</td>\n",
       "      <td>0.1717</td>\n",
       "      <td>Models/NN model | Input Raw Temperature | Epoc...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>SimpleRNN</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>0.0841</td>\n",
       "      <td>0.0830</td>\n",
       "      <td>0.1703</td>\n",
       "      <td>Models/RNN model | Input Raw Temperature | Epo...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>LSTM</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>0.0841</td>\n",
       "      <td>0.0806</td>\n",
       "      <td>0.1686</td>\n",
       "      <td>Models/LSTM model | Input Raw Temperature | Ep...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>DLSTM</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>0.0844</td>\n",
       "      <td>0.0837</td>\n",
       "      <td>0.1709</td>\n",
       "      <td>Models/DLSTM model | Input Raw Temperature | E...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>0.0070</td>\n",
       "      <td>0.0839</td>\n",
       "      <td>0.0782</td>\n",
       "      <td>0.1663</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Model Type  MSE Error  RMSE Error  MSPE Error  MAPE Error  \\\n",
       "0         NN     0.0071      0.0844      0.0846      0.1717   \n",
       "1  SimpleRNN     0.0071      0.0841      0.0830      0.1703   \n",
       "2       LSTM     0.0071      0.0841      0.0806      0.1686   \n",
       "3      DLSTM     0.0071      0.0844      0.0837      0.1709   \n",
       "4     DDLSTM     0.0070      0.0839      0.0782      0.1663   \n",
       "\n",
       "                                          Model path  \n",
       "0  Models/NN model | Input Raw Temperature | Epoc...  \n",
       "1  Models/RNN model | Input Raw Temperature | Epo...  \n",
       "2  Models/LSTM model | Input Raw Temperature | Ep...  \n",
       "3  Models/DLSTM model | Input Raw Temperature | E...  \n",
       "4  Models/DDLSTM model | Input Raw Temperature | ...  "
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_types = ['NN', 'SimpleRNN', 'LSTM', 'DLSTM', 'DDLSTM']\n",
    "\n",
    "Results = pd.DataFrame({ 'Model Type': []\n",
    "                        , 'MSE Error': []\n",
    "                        , 'RMSE Error': []\n",
    "                        , 'MSPE Error': []\n",
    "                        , 'MAPE Error': []\n",
    "                        , 'Model path': []\n",
    "                       })\n",
    "\n",
    "for model_type in model_types:\n",
    "    model_data = best_models[best_models['Model Type'] == model_type]\n",
    "    model_data.reset_index(inplace=True, drop=True)\n",
    "    \n",
    "    model_path = model_data['Model path'][0]\n",
    "    Regressor_Model = load_model(model_path)\n",
    "    y_test_predicted = Regressor_Model.predict(X_test)\n",
    "\n",
    "    y_test_predicted_only = []\n",
    "    \n",
    "    # getting only the prediction value\n",
    "    for i in range(len(y_test_predicted)):\n",
    "        y_test_predicted_only.append(float(y_test_predicted[i][0]))\n",
    "    \n",
    "    MSE_error = round(MSE(y_test, y_test_predicted_only), 4)\n",
    "    RMSE_error = round(RMSE(y_test, y_test_predicted_only), 4)\n",
    "    MSPE_error = round(MSPE(y_test, y_test_predicted_only), 4)\n",
    "    MAPE_error = round(MAPE(y_test, y_test_predicted_only), 4)\n",
    "    \n",
    "    Results = Results.append(pd.DataFrame({ 'Model Type': [model_type]\n",
    "                                            , 'MSE Error': [MSE_error]\n",
    "                                            , 'RMSE Error': [RMSE_error]\n",
    "                                            , 'MSPE Error': [MSPE_error]\n",
    "                                            , 'MAPE Error': [MAPE_error]\n",
    "                                            , 'Model path': [model_path]\n",
    "                                           }))\n",
    "\n",
    "Results.reset_index(inplace=True, drop=True)\n",
    "datetime_now = dt.datetime.now().strftime(\"%Y-%m-%d %H:%M\")\n",
    "Results.to_excel('Results Excel/Best models | Other error calculations | Input Raw temparature | Created at {}.xlsx'.format(\n",
    "    input_metrics, window_days, datetime_now))\n",
    "\n",
    "Results"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Testing by Cities"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Unnamed: 0</th>\n",
       "      <th>Forecast metric</th>\n",
       "      <th>Input metric</th>\n",
       "      <th>Number of epochs</th>\n",
       "      <th>Number of input days</th>\n",
       "      <th>Model Type</th>\n",
       "      <th>Loss train</th>\n",
       "      <th>Loss test</th>\n",
       "      <th>End Date-Time</th>\n",
       "      <th>Model path</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>10</td>\n",
       "      <td>10</td>\n",
       "      <td>NN</td>\n",
       "      <td>0.0110</td>\n",
       "      <td>0.0070</td>\n",
       "      <td>2022-05-29 16:25</td>\n",
       "      <td>Models/NN model | Input Raw Temperature | Epoc...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>10</td>\n",
       "      <td>10</td>\n",
       "      <td>SimpleRNN</td>\n",
       "      <td>0.0110</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>2022-05-29 16:25</td>\n",
       "      <td>Models/RNN model | Input Raw Temperature | Epo...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>10</td>\n",
       "      <td>10</td>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>0.0110</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>2022-05-29 16:27</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>150</td>\n",
       "      <td>10</td>\n",
       "      <td>SimpleRNN</td>\n",
       "      <td>0.0110</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>2022-05-29 16:29</td>\n",
       "      <td>Models/RNN model | Input Raw Temperature | Epo...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>0</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>150</td>\n",
       "      <td>10</td>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>0.0109</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>2022-05-29 16:49</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>150</td>\n",
       "      <td>10</td>\n",
       "      <td>NN</td>\n",
       "      <td>0.0112</td>\n",
       "      <td>0.0072</td>\n",
       "      <td>2022-05-29 16:28</td>\n",
       "      <td>Models/NN model | Input Raw Temperature | Epoc...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>0</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>150</td>\n",
       "      <td>10</td>\n",
       "      <td>LSTM</td>\n",
       "      <td>0.0109</td>\n",
       "      <td>0.0072</td>\n",
       "      <td>2022-05-29 16:31</td>\n",
       "      <td>Models/LSTM model | Input Raw Temperature | Ep...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>0</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>150</td>\n",
       "      <td>10</td>\n",
       "      <td>DLSTM</td>\n",
       "      <td>0.0109</td>\n",
       "      <td>0.0072</td>\n",
       "      <td>2022-05-29 16:36</td>\n",
       "      <td>Models/DLSTM model | Input Raw Temperature | E...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>10</td>\n",
       "      <td>10</td>\n",
       "      <td>DLSTM</td>\n",
       "      <td>0.0112</td>\n",
       "      <td>0.0073</td>\n",
       "      <td>2022-05-29 16:26</td>\n",
       "      <td>Models/DLSTM model | Input Raw Temperature | E...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>10</td>\n",
       "      <td>10</td>\n",
       "      <td>LSTM</td>\n",
       "      <td>0.0115</td>\n",
       "      <td>0.0074</td>\n",
       "      <td>2022-05-29 16:26</td>\n",
       "      <td>Models/LSTM model | Input Raw Temperature | Ep...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0 Forecast metric     Input metric  Number of epochs  \\\n",
       "0           0          Return  Raw Temperature                10   \n",
       "1           0          Return  Raw Temperature                10   \n",
       "4           0          Return  Raw Temperature                10   \n",
       "6           0          Return  Raw Temperature               150   \n",
       "9           0          Return  Raw Temperature               150   \n",
       "5           0          Return  Raw Temperature               150   \n",
       "7           0          Return  Raw Temperature               150   \n",
       "8           0          Return  Raw Temperature               150   \n",
       "3           0          Return  Raw Temperature                10   \n",
       "2           0          Return  Raw Temperature                10   \n",
       "\n",
       "   Number of input days Model Type  Loss train  Loss test     End Date-Time  \\\n",
       "0                    10         NN      0.0110     0.0070  2022-05-29 16:25   \n",
       "1                    10  SimpleRNN      0.0110     0.0071  2022-05-29 16:25   \n",
       "4                    10     DDLSTM      0.0110     0.0071  2022-05-29 16:27   \n",
       "6                    10  SimpleRNN      0.0110     0.0071  2022-05-29 16:29   \n",
       "9                    10     DDLSTM      0.0109     0.0071  2022-05-29 16:49   \n",
       "5                    10         NN      0.0112     0.0072  2022-05-29 16:28   \n",
       "7                    10       LSTM      0.0109     0.0072  2022-05-29 16:31   \n",
       "8                    10      DLSTM      0.0109     0.0072  2022-05-29 16:36   \n",
       "3                    10      DLSTM      0.0112     0.0073  2022-05-29 16:26   \n",
       "2                    10       LSTM      0.0115     0.0074  2022-05-29 16:26   \n",
       "\n",
       "                                          Model path  \n",
       "0  Models/NN model | Input Raw Temperature | Epoc...  \n",
       "1  Models/RNN model | Input Raw Temperature | Epo...  \n",
       "4  Models/DDLSTM model | Input Raw Temperature | ...  \n",
       "6  Models/RNN model | Input Raw Temperature | Epo...  \n",
       "9  Models/DDLSTM model | Input Raw Temperature | ...  \n",
       "5  Models/NN model | Input Raw Temperature | Epoc...  \n",
       "7  Models/LSTM model | Input Raw Temperature | Ep...  \n",
       "8  Models/DLSTM model | Input Raw Temperature | E...  \n",
       "3  Models/DLSTM model | Input Raw Temperature | E...  \n",
       "2  Models/LSTM model | Input Raw Temperature | Ep...  "
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Loading models\n",
    "excel_file_path = 'Results Excel/Input Raw Temperature | Epochs 5-150 | Days 10 | Created at 2022-05-29 16:49.xlsx'\n",
    "Models = pd.read_excel(excel_file_path)\n",
    "Models = Models.sort_values(by='Loss test')\n",
    "Models"
   ]
  },
  {
   "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>index</th>\n",
       "      <th>Unnamed: 0</th>\n",
       "      <th>Forecast metric</th>\n",
       "      <th>Input metric</th>\n",
       "      <th>Number of epochs</th>\n",
       "      <th>Number of input days</th>\n",
       "      <th>Model Type</th>\n",
       "      <th>Loss train</th>\n",
       "      <th>Loss test</th>\n",
       "      <th>End Date-Time</th>\n",
       "      <th>Model path</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>Return</td>\n",
       "      <td>Raw Temperature</td>\n",
       "      <td>10</td>\n",
       "      <td>10</td>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>0.011</td>\n",
       "      <td>0.0071</td>\n",
       "      <td>2022-05-29 16:27</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   index  Unnamed: 0 Forecast metric     Input metric  Number of epochs  \\\n",
       "0      4           0          Return  Raw Temperature                10   \n",
       "\n",
       "   Number of input days Model Type  Loss train  Loss test     End Date-Time  \\\n",
       "0                    10     DDLSTM       0.011     0.0071  2022-05-29 16:27   \n",
       "\n",
       "                                          Model path  \n",
       "0  Models/DDLSTM model | Input Raw Temperature | ...  "
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_data = Models[Models['Model Type'] == 'DDLSTM']\n",
    "model_data = model_data[model_data['Number of epochs'] == 10]\n",
    "model_data.reset_index(inplace=True)\n",
    "model_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_type = model_data['Model Type'][0]\n",
    "model_path = model_data['Model path'][0]\n",
    "\n",
    "Regressor_Model = load_model(model_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2/2 [==============================] - 0s 23ms/step - loss: 0.0083\n",
      "2/2 [==============================] - 0s 20ms/step - loss: 0.0060\n",
      "2/2 [==============================] - 0s 19ms/step - loss: 0.0084\n",
      "2/2 [==============================] - 0s 20ms/step - loss: 0.0060\n",
      "2/2 [==============================] - 0s 23ms/step - loss: 0.0088\n",
      "2/2 [==============================] - 0s 19ms/step - loss: 0.0060\n",
      "2/2 [==============================] - 0s 20ms/step - loss: 0.0060\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Model Type</th>\n",
       "      <th>Model path</th>\n",
       "      <th>City</th>\n",
       "      <th>Loss test</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "      <td>Minneapolis</td>\n",
       "      <td>0.0060</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "      <td>New York</td>\n",
       "      <td>0.0060</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "      <td>Cincinnati</td>\n",
       "      <td>0.0060</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "      <td>Dallas</td>\n",
       "      <td>0.0060</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "      <td>Sacramento</td>\n",
       "      <td>0.0083</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "      <td>Las Vegas</td>\n",
       "      <td>0.0084</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "      <td>Amsterdam</td>\n",
       "      <td>0.0088</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Model Type                                         Model path         City  \\\n",
       "0     DDLSTM  Models/DDLSTM model | Input Raw Temperature | ...  Minneapolis   \n",
       "0     DDLSTM  Models/DDLSTM model | Input Raw Temperature | ...     New York   \n",
       "0     DDLSTM  Models/DDLSTM model | Input Raw Temperature | ...   Cincinnati   \n",
       "0     DDLSTM  Models/DDLSTM model | Input Raw Temperature | ...       Dallas   \n",
       "0     DDLSTM  Models/DDLSTM model | Input Raw Temperature | ...   Sacramento   \n",
       "0     DDLSTM  Models/DDLSTM model | Input Raw Temperature | ...    Las Vegas   \n",
       "0     DDLSTM  Models/DDLSTM model | Input Raw Temperature | ...    Amsterdam   \n",
       "\n",
       "   Loss test  \n",
       "0     0.0060  \n",
       "0     0.0060  \n",
       "0     0.0060  \n",
       "0     0.0060  \n",
       "0     0.0083  \n",
       "0     0.0084  \n",
       "0     0.0088  "
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Results_by_City = pd.DataFrame({ \n",
    "                         'Model Type': []\n",
    "                         , 'Model path': []\n",
    "                         , 'City': []\n",
    "                         , 'Loss test': []\n",
    "                       })\n",
    "\n",
    "for city in list_of_cities:\n",
    "    \n",
    "    df_price_city = df_price_final[df_price_final['City'] == city]\n",
    "    df_price_city = df_price_city[['Date', output_metrics]]\n",
    "    df_price_city.reset_index(inplace=True, drop=True)\n",
    "\n",
    "    df_weather_city = df_weather_final[df_weather_final['Is{}'.format(city)] == 1]\n",
    "    df_weather_city.reset_index(inplace=True, drop=True)\n",
    "\n",
    "    # Dropping 'IsDallas' => because other 0 is IsDallas\n",
    "    df_weather_city = df_weather_city[['Date', input_metrics, 'IsSacramento', 'IsMinneapolis'\n",
    "                                   , 'IsLas Vegas', 'IsNew York', 'IsAmsterdam', 'IsCincinnati']]\n",
    "\n",
    "\n",
    "\n",
    "    # Train/Test start dates\n",
    "    # Find Train output start day\n",
    "    train_date_output_start = df_price_city['Date'].min()\n",
    "    # Find Train intput start day (remember input is LAST 'window_days' of days)\n",
    "    train_index_input_start = len(df_weather_city[df_weather_city['Date'] < train_date_output_start]) - window_days\n",
    "    train_date_input_start = df_weather_city['Date'][train_index_input_start]\n",
    "\n",
    "    # Test is LAST 'test_days' DAYS\n",
    "    test_date_output_start = df_price_city['Date'][len(df_price_city) - test_days]\n",
    "    # Find Train intput start day (remember input is LAST 'window_days' of days)\n",
    "    test_index_input_start = len(df_weather_city[df_weather_city['Date'] < test_date_output_start]) - window_days\n",
    "    test_date_input_start = df_weather_city['Date'][test_index_input_start]\n",
    "\n",
    "    test_date_end = df_price_city['Date'].max()\n",
    "\n",
    "\n",
    "    ##### Spliting datasets\n",
    "    ### Train\n",
    "    # input train data\n",
    "    df_train_input = df_weather_city[df_weather_city['Date'] >= train_date_input_start]\n",
    "    df_train_input = df_train_input[df_train_input['Date'] < test_date_output_start]\n",
    "    training_input_set = df_train_input.iloc[:,1:2].values\n",
    "    training_input_set_Dates = df_train_input.iloc[:,0:1].values\n",
    "    training_input_set_DUMMY = df_train_input.iloc[:,2:].values\n",
    "\n",
    "    # output train data\n",
    "    df_train_output = df_price_city[df_price_city['Date'] >= train_date_output_start]\n",
    "    df_train_output = df_train_output[df_train_output['Date'] < test_date_output_start]\n",
    "    training_output_set = df_train_output.iloc[:,1:2].values\n",
    "    training_output_set_Dates = df_train_output.iloc[:,0:1].values\n",
    "\n",
    "\n",
    "    ### Test\n",
    "    # input test data\n",
    "    df_test_input = df_weather_city[df_weather_city['Date'] >= test_date_input_start]\n",
    "    df_test_input = df_test_input[df_test_input['Date'] <= test_date_end]\n",
    "    testing_input_set = df_test_input.iloc[:,1:2].values\n",
    "    testing_input_set_Dates = df_test_input.iloc[:,0:1].values\n",
    "    testing_input_set_DUMMY = df_test_input.iloc[:,2:].values\n",
    "\n",
    "    # output test data\n",
    "    df_test_output = df_price_city[df_price_city['Date'] >= test_date_output_start]\n",
    "    df_test_output = df_test_output[df_test_output['Date'] <= test_date_end]\n",
    "    testing_output_set = df_test_output.iloc[:,1:2].values\n",
    "    testing_output_set_Dates = df_test_output.iloc[:,0:1].values\n",
    "\n",
    "\n",
    "    ### Feature scaling\n",
    "    sc_input, sc_output = MinMaxScaler(feature_range=(0,1)), MinMaxScaler(feature_range=(0,1))\n",
    "    # scaling inputs\n",
    "    training_input_set_scaled = sc_input.fit_transform(training_input_set)\n",
    "    testing_input_set_scaled = sc_input.transform(testing_input_set)\n",
    "\n",
    "    # scaling inputs\n",
    "    training_output_set_scaled = sc_output.fit_transform(training_output_set)\n",
    "    testing_output_set_scaled = sc_output.transform(testing_output_set)\n",
    "\n",
    "\n",
    "    X_test  = []\n",
    "    X_test_Dates  = []\n",
    "    y_test  = []\n",
    "    y_test_Dates  = []\n",
    "\n",
    "\n",
    "    for i in range(window_days, testing_input_set_scaled.size):\n",
    "        X_test.append(np.append(testing_input_set_DUMMY[0]\n",
    "                                , testing_input_set_scaled[i-window_days:i, 0], axis=0))\n",
    "        X_test_Dates.append(np.append(testing_input_set_DUMMY[0]\n",
    "                                      , testing_input_set_Dates[i-window_days:i, 0], axis=0))\n",
    "\n",
    "\n",
    "    for i in range(0, testing_output_set_scaled.size):\n",
    "        y_test.append(testing_output_set_scaled[i, 0])\n",
    "        y_test_Dates.append(testing_output_set_Dates[i, 0])\n",
    "\n",
    "\n",
    "    # Fixing of test set\n",
    "    y_test_Dates_Corrected = [] # output is forecast therefore input - output is one day and we \n",
    "                                # should correct for that\n",
    "    X_test_Dates_Corrected = []\n",
    "    X_test_Corrected = []\n",
    "\n",
    "\n",
    "    for i in range(len(y_test_Dates)):\n",
    "        # taking the day before\n",
    "        y_test_Dates_Corrected.append(y_test_Dates[i] - dt.timedelta(days=1))\n",
    "\n",
    "    y_test_Dates_Corrected_set = set(y_test_Dates_Corrected)\n",
    "\n",
    "    for i in range(len(X_test_Dates)):\n",
    "        # cheking whether we have an output for that dates \n",
    "        if X_test_Dates[i][-1] in y_test_Dates_Corrected_set:\n",
    "            X_test_Dates_Corrected.append(X_test_Dates[i])\n",
    "            X_test_Corrected.append(X_test[i]) \n",
    "\n",
    "\n",
    "    X_test,  y_test  = np.array(X_test_Corrected),  np.array(y_test)\n",
    "\n",
    "    X_test  = np.reshape(X_test,  [X_test.shape[0],  X_test.shape[1],  1])\n",
    "\n",
    "    Model_Loss_test  = round(Regressor_Model.evaluate(X_test,  y_test ), 4)\n",
    "\n",
    "    Results_by_City = Results_by_City.append(pd.DataFrame({ 'Model Type': [model_type]\n",
    "                                                            , 'City': [city]\n",
    "                                                            , 'Loss test': [Model_Loss_test]\n",
    "                                                            , 'Model path': [model_path]\n",
    "                                                           }))\n",
    "    \n",
    "Results_by_City = Results_by_City.sort_values(by='Loss test')\n",
    "Results_by_City"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Model Type</th>\n",
       "      <th>Model path</th>\n",
       "      <th>City</th>\n",
       "      <th>Loss test</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "      <td>Minneapolis</td>\n",
       "      <td>0.0060</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "      <td>New York</td>\n",
       "      <td>0.0060</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "      <td>Cincinnati</td>\n",
       "      <td>0.0060</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "      <td>Dallas</td>\n",
       "      <td>0.0060</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "      <td>Sacramento</td>\n",
       "      <td>0.0083</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "      <td>Las Vegas</td>\n",
       "      <td>0.0084</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>DDLSTM</td>\n",
       "      <td>Models/DDLSTM model | Input Raw Temperature | ...</td>\n",
       "      <td>Amsterdam</td>\n",
       "      <td>0.0088</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Model Type                                         Model path         City  \\\n",
       "0     DDLSTM  Models/DDLSTM model | Input Raw Temperature | ...  Minneapolis   \n",
       "0     DDLSTM  Models/DDLSTM model | Input Raw Temperature | ...     New York   \n",
       "0     DDLSTM  Models/DDLSTM model | Input Raw Temperature | ...   Cincinnati   \n",
       "0     DDLSTM  Models/DDLSTM model | Input Raw Temperature | ...       Dallas   \n",
       "0     DDLSTM  Models/DDLSTM model | Input Raw Temperature | ...   Sacramento   \n",
       "0     DDLSTM  Models/DDLSTM model | Input Raw Temperature | ...    Las Vegas   \n",
       "0     DDLSTM  Models/DDLSTM model | Input Raw Temperature | ...    Amsterdam   \n",
       "\n",
       "   Loss test  \n",
       "0     0.0060  \n",
       "0     0.0060  \n",
       "0     0.0060  \n",
       "0     0.0060  \n",
       "0     0.0083  \n",
       "0     0.0084  \n",
       "0     0.0088  "
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#DDLSTM_Results_by_City = Results_by_City.copy()\n",
    "DDLSTM_Results_by_City "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "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>Model Type</th>\n",
       "      <th>Model path</th>\n",
       "      <th>City</th>\n",
       "      <th>Loss test</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NN</td>\n",
       "      <td>Models/NN model | Input Raw Temperature | Epoc...</td>\n",
       "      <td>Minneapolis</td>\n",
       "      <td>0.0060</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NN</td>\n",
       "      <td>Models/NN model | Input Raw Temperature | Epoc...</td>\n",
       "      <td>New York</td>\n",
       "      <td>0.0060</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NN</td>\n",
       "      <td>Models/NN model | Input Raw Temperature | Epoc...</td>\n",
       "      <td>Cincinnati</td>\n",
       "      <td>0.0060</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NN</td>\n",
       "      <td>Models/NN model | Input Raw Temperature | Epoc...</td>\n",
       "      <td>Dallas</td>\n",
       "      <td>0.0060</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NN</td>\n",
       "      <td>Models/NN model | Input Raw Temperature | Epoc...</td>\n",
       "      <td>Sacramento</td>\n",
       "      <td>0.0082</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NN</td>\n",
       "      <td>Models/NN model | Input Raw Temperature | Epoc...</td>\n",
       "      <td>Las Vegas</td>\n",
       "      <td>0.0082</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NN</td>\n",
       "      <td>Models/NN model | Input Raw Temperature | Epoc...</td>\n",
       "      <td>Amsterdam</td>\n",
       "      <td>0.0087</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Model Type                                         Model path         City  \\\n",
       "0         NN  Models/NN model | Input Raw Temperature | Epoc...  Minneapolis   \n",
       "0         NN  Models/NN model | Input Raw Temperature | Epoc...     New York   \n",
       "0         NN  Models/NN model | Input Raw Temperature | Epoc...   Cincinnati   \n",
       "0         NN  Models/NN model | Input Raw Temperature | Epoc...       Dallas   \n",
       "0         NN  Models/NN model | Input Raw Temperature | Epoc...   Sacramento   \n",
       "0         NN  Models/NN model | Input Raw Temperature | Epoc...    Las Vegas   \n",
       "0         NN  Models/NN model | Input Raw Temperature | Epoc...    Amsterdam   \n",
       "\n",
       "   Loss test  \n",
       "0     0.0060  \n",
       "0     0.0060  \n",
       "0     0.0060  \n",
       "0     0.0060  \n",
       "0     0.0082  \n",
       "0     0.0082  \n",
       "0     0.0087  "
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#NN_Results_by_City = Results_by_City.copy()\n",
    "NN_Results_by_City "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}