RNN_flight_passanger.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# load the libraries\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "import torch\n",
    "import torch.nn as nn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# read the airlines data\n",
    "data = pd.read_csv('airline_passengers.csv',usecols=[1])"
   ]
  },
  {
   "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>Passengers</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>112</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>118</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>132</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>129</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>121</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Passengers\n",
       "0         112\n",
       "1         118\n",
       "2         132\n",
       "3         129\n",
       "4         121"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x262e65d1888>]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD4CAYAAAAXUaZHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nO3deXxcV3n4/88ZjTTaRvu+2LJlx3Zix0scx05C9pAFSiBASUghpfkSaENLaflCoD/a0i+F8isFUr4UGpqSQCFAEtKENPtOVsdOHO+LbMva931Gs5/vH/fe0UgaaUaaK2vx8369/JLmzsydo0n0zNFzn/McpbVGCCHE0uKY7wEIIYSwnwR3IYRYgiS4CyHEEiTBXQghliAJ7kIIsQQ553sAACUlJbqurm6+hyGEEIvK7t27e7TWpfHuWxDBva6ujl27ds33MIQQYlFRSp2a6j5JywghxBIkwV0IIZYgCe5CCLEESXAXQoglSIK7EEIsQRLchRBiCZLgLoQQS5AEdyGEsNG7zQPsPtU338OQ4C6EEHb61hOHuOMX7xCJzO9eGRLchRDCRgPeIB1DPnY2zu/sXYK7EELYaHA0CMCj77bN6zgkuAshhI2GzOD++L52AqHIvI1DgrsQQtgkFI7gCYTZWJPPgDfIKw3d8zaWpIK7UqpAKfWgUuqwUuqQUmqHUqpIKfWMUuqY+bXQfKxSSv2rUqpBKbVXKbVlbn8EIYRYGIZ8IQDed24l+VnpPLpn/lIzyc7c7wKe1FqvBTYCh4A7gee01quB58zbANcBq81/twM/snXEQgixQFkpmZJcF5evKeX1E73zNpaEwV0plQdcAtwDoLUOaK0HgBuA+8yH3Qd80Pz+BuBn2vAGUKCUqrR95EIIscBYF1PzMtMpdbsYGg3N21iSmbmvBLqBnyql3lFK/YdSKgco11q3A5hfy8zHVwPNMc9vMY8JIcSSNuQzg3tWOrmudEaDYYLh+bmomkxwdwJbgB9prTcDHsZSMPGoOMcmVfMrpW5XSu1SSu3q7p6/iw5CCGEXa6aen5WOO9PY6M7jn5/ZezLBvQVo0Vq/ad5+ECPYd1rpFvNrV8zja2OeXwNMuqqgtb5ba71Va721tDTuFoBCCLGoRNMyWc5ocB/2LdDgrrXuAJqVUmvMQ1cCB4FHgVvNY7cCj5jfPwp80qya2Q4MWukbIYRYyqy0TOzM3Tp2uiW7QfafA79QSmUAJ4BPYXww/EYpdRvQBHzUfOzjwPVAA+A1HyuEEEve4GgQp0ORlZ6GOzMdgJF5mrknFdy11nuArXHuujLOYzVwR4rjEkKIRWdoNEh+VjpKKXJdCzwtI4QQIjmDo0HysowZu5WWGVnAF1SFEEIkYcgXIs8M6rnRC6rzk3OX4C6EEDYZipm555k592GZuQshxOIWG9xdTgfpaUpy7kIIsdgN+YwLqkD0oqqkZYQQYhHTWhsXVM10DIA7M33eSiEluAshhA18wQjBsI7O3AFz5i7BXQghFq3Y1gMWd6ZTLqgKIcRiFtt6wOLOTJeZuxBCnA6RiOZ377bhDdgbdIdierlb3JlyQVUIIU6L3+1t48/vf4enD3Taet6xtMz44C4rVIUQYo6FwhHuevYYAAPegK3njpeWsS6oGi23Ti8J7kKIM8aj77ZxoscD2N/zZdBrpWViL6imE45ofMHTvxuTBHchxBkhFI5w13PHOLsyD5fTYfuFziHzfLFpmfnsLyPBXQhxRnjzZB+ner187opVuDPTo8HYLkOjQbIz0khPGwur1ix+PsohJbgLIc4I3cN+ANZUuMmbgyqWwdHguHw7MK9b7UlwF0KcEfrNC6iF2RlmiaLdaZnxrQcAcl1mZ0hJywghxNzo9wZRytrfNH1OZu6xq1MhZsMOmbkLIcTcGPQGyMtMJ82h5qTny9BoaFJaZj632pPgLoQ4I/R7gxRkj22BZ3spZEwvd8t8btghwV0IcUbo9wYoyM4A7O/5orWmZ8RPSa5r3PEcVxogOXchhJgzA94ghRNm7uGIPStHR/wh/KEIxTkZ44470xxkZ6RJWkYIIeZKvzdAYXTmbl7otCld0jtiVOJMnLlbryUXVIUQYo4MxuTco7lwm9IlPSNGDX2Je3Jwz3U5GfZLWkYIIWwXDEcY9ocoyBo/c7crXWIF94lpGeO15qenuwR3IcSSN2A29SrMsXLu1szdruBupGVK48zc52LBVDIkuAshljyrva9VLZMbzbnbm5Ypijtzn5+e7hLchRBLXr81c4+plgF70zIF2enjmoZZjAVTknMXQgjbRWfuE3LudnWG7B0JxK2UMV5Lcu5CCDEnrJz7XFbLlOROTsmA8UHiDYQJhU/vhh1JBXelVKNSap9Sao9Sapd5rEgp9YxS6pj5tdA8rpRS/6qUalBK7VVKbZnLH0AIsXT85OUTfOepI7afN9oR0syJu5wO0tOUrRdUi6eYuZe5MwHoNFsOny4zmblfrrXepLXeat6+E3hOa70aeM68DXAdsNr8dzvwI7sGK4RYujz+EN9/9iiP72+3/dz93iDpaYqcDKMdgFLK1s6QPSN+SqcI7tWFWQC0DYza8lrJSiUtcwNwn/n9fcAHY47/TBveAAqUUpUpvI4Q4gzwu3fb8ATCc7Kac8DsK6OUih6zq0TRFwwz7AtNmZapLjBm7gs1uGvgaaXUbqXU7eaxcq11O4D5tcw8Xg00xzy3xTw2jlLqdqXULqXUru7u7tmNXgixZNy/swmwf+NqMHLuBXF2SbLjg6TPY6R8pkrLVBUYM/eW/tMb3J2JHwLARVrrNqVUGfCMUurwNI9VcY5N6s6jtb4buBtg69at9nTvEUIsSvtbB3m3ZZAyt4uuYT/hiCbNES+UzE5sXxmLXT3do60Hpgju2RlOCrPTF+bMXWvdZn7tAh4GtgGdVrrF/NplPrwFqI15eg3QZteAhRBLz6/easLldHDT+UbosHv2PhDTV8ZibJKdes59LLjHT8uAMXtvXWjBXSmVo5RyW98D7wX2A48Ct5oPuxV4xPz+UeCTZtXMdmDQSt8IIUQ8rzX0cslZpdQUZgP29z+PN3O3K+feM01HSEtVQdZpn7knk5YpBx42L0Q4gV9qrZ9USr0F/EYpdRvQBHzUfPzjwPVAA+AFPmX7qIUQS0rnkI/L1pTFtAWwdyONeDP3PJuqZRKlZQCqC7J4raEHrfW4i7pzKWFw11qfADbGOd4LXBnnuAbusGV0Qoglb8QfwhMIU5bniu45amfFzGgwTCAcifaVsVg9X1INuD3DAXIy0sgyyyzjqS7IwhMIG/usTviQmSuyQlUIMa+6hnwAlOe5xnq+2Dhzn9hXxuLOdBLR4AmEUzp/r8c/ZaWMxap1P515dwnuQoh51WWu3CxzZ47tkGTjzL3fM74jpMVq+5vqa03XesBilUNKcBdCnDE6Y2buuS4z4No4cx+YZuYOqV+87RmeummYpWoeFjJJcBdCzKtuc+Ze6s4cu6Bq58zdG3/mbuX3U+0MmUxapiTHRYbTcVqDe7KLmIQQYk50DvnITHeQl+lEa1DK3lLIsaZhk+vcIbXXaun30usJUJmfOe3jHA5FVX4mLTJzF0IsJAPeAK8cM0r57NY17KfMnYlSCodDkZvhtPWCaueQjzSHojhn/Ow6z4YNO374QgPpDgcfOa8m4WNPd627zNyFEFPqGvLxtUf28/zhLoJhzS8/fQEX1pfY+hqdQz7K88YCb65NPV8s7YM+yt2uSe0MUt1HtbnPywO7WrjlgmXRC6bTqS7I4qWjp6+PlszchRBTeupgJ08d6OT6DUZj1/YBn+2vYc3cLbkue/cc7RzyUR4nbZLqBdUfPH8Mh0PxZ5evSurxVQVZdA378YdSK71MlgR3IcSUmvu8ZDgdfOOD64Gx1Zh26hryUxYzc7d7Q+n2QV/cnHh2RhoupyPa1XEmekf8PPR2K7dcsIzyvOnz7ZZqc3bfOXh6Nu2Q4C6EmFJzn5eawixyXU4y0x22B3ePP8SIPzR+5m7jnqNaazoGfXEDsFKKsjxXtBRzJo53ewhHNJevKUv8YFOJ26jW6fVIcBdCzLPmfi+1hdkopSjJddE7MvNZ7nSsBUyxOXe3jWmZYX8IbyA8ZTVLuTszOoaZaOrzArCsKDvp5xSZF3St6p25JsFdCDGlpl4vtUVGOqE410W3zTN3q/XAxJy7XaWQnYPWAqn4wX22M/emPi8ORVIXUi1FZp293R+QU5HgLoSIa3A0yJAvFJ2dluZmRNvb2qUzzszdzmqZdjO4V+bHD8Jls5y5N/d5qczPIsOZfAi16uxl5i6EmFfNZuqh1uyxbqRlTs/M3RMIE46kXlPfYZ6/YpqZ+7AvxOgMm4c19XlnlJIB4+dKT1P0eeztVT8VCe5CiLha+s3gbgax4twMej0BIjYEXUvXsB+X00Fe1tiSG6tE0RNIffZupWViq3FiWR8qXcMzS83MJrgrpSjMzog2MptrEtyFEHE19xmrKWNn7uGIZmDUvpln15CPsjzXuH7qdvZ0bx/yUZSTQWZ6/F7rVjqocyj5v0hGA2G6h/3RaxEzUZSTQZ+kZYQQ86mpz0tepjO6uYTV+dDO1EznkJ9y9/iUSbQVrw0VM52DvilTMjC7mXvzhL9oZkJm7kKIedfc7x0XwIrNnuV2Vsx0DfsmpUxybWrFC8YF1YppmnrNZube1DvzMkiLzNyFEPOuuc8bTckAlJozdzsrZia2HoCxtIwdC5k6h6YP7vlZ6WQ4HTOauc+mxt1SlDM2c9da4wvOXSsCCe5CiEkiEU1L/+i4vLKVlumZRelgPIFQhGFfiOKcyXubQuppGX8oTK8nMG1aRilFmdtF10xm7n1ecjLSKMqZfveleApzMhgYDRKOaAZHg6z92pP8/PXGGZ8nGdIVUggxSfeIH38oMm52mp+VTppD2bZ8fmDU6rMefxONVC+oWgF7upk7YAT3meTc+4x01Ww21S7KTkdrYw2BVY1UlmRvmpmSmbsQYhKrxr0mJrg7HIrinAx6hu1Jy/Sb9d4TZ8C5Ns3crQVM083cwVi9OqOc+yzKIC3WB1mfJ5BSeicZEtyFEJNEK0IKxweeklyXbc3D+qIbV4/fISknw56cu7WAKdEuSWXu5FsQaK1TCu7WB1m/dyy4z6bqJhkS3IUQk7T2GzXuNYXja7mLczPosamUz1qGP3HmnuZQtvR0j/aVSRTc8zKTXqXaPWymq4pnOXPPHpu5N/d5Kc7JiKah7CbBXQgxSc9IAHemc9Lin9Jcl20XVK2Ze1H25AuTdjQPO9HjwZ3pxJ0geJa5jQvF8fLuvmCYkz2e6O1TKc62iyakZeZq1g4S3IUQcfSM+KPVMbFK3EZaxo69VPujaZk4wd2GDTt2n+pjy7LChBc+rY6R8RqI/cNjB7n+rt9HSxYPtA4CsLbCPasxTQzuc5VvBwnuQixa3336CP+zt31Ozt3nCcQt9SvOycAfiuCZYaOtePq9QdwuZ9zOisbMffbBfdAb5GjnCFuXFyZ8bFl0IdP4mXvXsI8Hd7UwGgyz3wzq+1qHKMnNSHiRdiqZ6WlkZ6TRPeynbcA3p8FdSiGFWITCEc2PXzqBwwFrK93Ul+baev4+TyBuyiC21j3VXHG/NzCpDNKS6lZ7u5v6ANhaV5TwsVb7g4m17j99tZFgJALAnuYBttYVsb91kPXV+bMqg7QUZmdwoG2QcETLzF0IMV7bwCiBcARfMMIXfr2HYDhi6/l7RgKU5E4OvCVua5Vq6nn3Pk+AwgmVMpZc18x7umuto22C32rsx+lQbKotSPi8gux0MtIcdMbk3Id9Qf7rjVNcv76S6oIs3mkeYDQQ5ljXMBuq82c0romKcjLYZ/4lsCBy7kqpNKXUO0qpx8zbK5RSbyqljimlfq2UyjCPu8zbDeb9dXMzdCHOXKfM/iZ/fGEde1sG+fGLx207dySi6fdOnZYBe1oQTDdzn021zGd+vpvP/tduAHY39nNOdT5ZGfG7Qcay9lLtGBwL7r/a2cywL8RnL61nU20Be5oGONg+RETD+hSDe2FOBr6g8WE826qbZMxk5v554FDM7W8D39Narwb6gdvM47cB/VrrVcD3zMcJIWx0steo4PjspfVsqyviucNdtp17yGcsj7f2/IxVas7c7Wge1ucJxK2UAcjLSmfAG5zRhdtDHUM8c7CTZw52sqdlIKl8u6W6ICta/gnw+ole1pS72VCTz6baAloHRnnxiPEepzxzN/9aSU9Ts87dJyOp4K6UqgHeB/yHeVsBVwAPmg+5D/ig+f0N5m3M+69UqSSohBCTNPZ4yEx3UOZ2sbw4m/bB0cRPSpI1K5/Y88U6ppRR752qfs/UM/eqgixGg2H6vcmXQ1p7k37xgXcJhCKcX5d8cK8pzKZ1YOw9bO7zstycVW9aZqR2fvVWM8U5GQkXRSVi/cw1hdmkOeYuNCY7c/8+8CXASuwVAwNaa+vvphag2vy+GmgGMO8fNB8/jlLqdqXULqXUru7u7lkOX4gz06leD3XFOTgcisqCLLqG/bbl3a368+I4OXdnmoPiHFd0e7zZ8ofCeALhKZtvWYunYmfT0/EGQngDYTbVFjBobiZy3vLEF1NjX69jyEcgFEFrPa7d8fqqfNIciu5hf8oXU2Gsrn8u8+2QRHBXSr0f6NJa7449HOehOon7xg5ofbfWeqvWemtpaWlSgxVCGE72eKIzy6r8TLSeXMo3W31mY7CpAq/RaCu1mfuAOSMvnCItYwV3q7lWItas/ePblrFtRRFnledGU0jJqC7MQmtoHxyle8SPLxih1hxDVkZatK491ZQMjM3cl81iJ6eZSKaW6SLgA0qp64FMIA9jJl+glHKas/MaoM18fAtQC7QopZxAPtBn+8iFOEOFI5rmvlGuOrscgMoCI0i0D/qoKUx9Nthrzdzj5NzBqAuf6Z6jE0VXp+bEr5apKTB+jpYkZ+5W9U6JO4P//OPzZ9wnPfYvBZe5Kjf2Yuem2gIOtA2lfDEVxtJdc1kGCUnM3LXWX9Fa12it64CbgOe11rcALwAfMR92K/CI+f2j5m3M+5/XdixnE0IAY2WQdcU5wFhjrLYBe/Lu1ix42pn7DLooxjPd6lSAvCyjbUBrkj+TNeaSXBe5Lmfc1bXTsRqktfSPjm0MHvNB+Z7VpbicDrYsS1xamUixOba5Du6prEL4MvArpdQ3gHeAe8zj9wA/V0o1YMzYb0ptiEKIWI1mpczE4B5bypeKPo/RVybeylEw9h3tGfETjuhZXxDsm6JpmEUpRXVhVtJpGWvmXjzDoG6pyM/EoYw0kPVzx/4VdM055ez+2tW2NPk6b3kh3/jgeq5YW57yuaYzo5FqrV8EXjS/PwFsi/MYH/BRG8YmhIij0axxX1FiBHd3ZjpulzPavzxVvZ5A3EoZS1mei4iGXs/kLfKSZc3cp8q5gxFck865e6au8ElGepqDirxMWvpHSU9zUJLrGlcjr5SyrXtjmkPxR9uX23Ku6cgKVSEWmdgySEtlQaaNaRn/tFvIRbsoppCasUocJ/Zyj1VTmEVL/2hSte49I37crsldLGeipjCbloFRs6HX3F7sPB0kuAuxyDT2jJVBWirzs2ybufd5AtOmN0rN2Xoqte59ngB5mU7S06YOQTWFWYz4QwyNJl6p2jsSiFu6ORM1hcZCptgyyMVMgrsQi0xj71gZpKWqINO2hUwJ0zLT9D9P1lTtDWJZFSzNSaRmekb8s863W6oLs2gfHKV90DdpB6rFSIK7EIuIVQZpXUy1VORl0TMSwB9KrRVvJKLpn6Ldr6XUhrRMnycwZaWMpaYw+XLI3ikanc1ETWEWEW28x7WSlhFCnE69I34C4cik7e8qC+ypmBnyBQlF9LSz4Mz0NPKz0lNayJTMzL3arN9Pphyy15P6zD22OkZm7kKI08oKqKUTqlSq8o1A2DaQWnBPturEWKWaQlrGE5y2UgaMi605GWkJK2bCEU2fJ0DJLCtlLNaHCcx9a4DTQYK7EHNgrtbtWS0GyvPGz1KjM/eh1PLuYytHEwT3vNRaEBgz96krZcAoPzTKIaf/mfq9ASJ6rNf8bFUWZKKUUaqYanOwhUCCuxA2e2JfOxd887loAys7WQG1LG+OZu4j0/eVsZS5M5POuUcimiMdw9HbvmAYbyA8ZUfIWFY55HR6R6Zvl5AslzONcncmVQWZOKep4lksFv9PIMQCs6d5gK5hP88f7rT93NbMvXRCfjkrI42C7PSUK2astEyi5ftlbhfdw8ltlP3soU6u+f7L7DxptJh640QvAKuS2BqwujCL1inSMg/ubqFryBezOjW1tAwYWxauq8hL+TwLgQR3IWzWZl7UfHxfh+3n7ho2FhjFaw1QmZ9Fe4oz9z5zFlyYIGVSlpdJIByJdneczrGuEQB++eYpwAjKBdnpXLomcTfYmsIshnwhhnzjX6exx8MXH3iXH77QMNY0LMULqgA/uHkz3/3YppTPsxBIcBfCZu1mdcdLR7tT2uQ5nq4h37iVqbEq8zOjHyzJ+tRPd/L3jx6I3u42V3q6nNOv9ByrdU+cmrHSKo/v76Cp18vTBzu5YWNVwtcAY9MOYNKHlvVXwDMHO6Obi6RaCglGKwe72gzMNwnuQtisfdDH8uJsAqEIz9u4/R0YwXRivt1SmT+zhUxaa9482ce9rzXy1IEOjnYO8+DulujOQ9OZyUKmln4vRTkZBEIR/uyXuwmEInz4vJqkxlgZvZYw/ud60wzubYM+Xj7ajdOhyMuc/q+NM40EdyFsFI5oOoZ8XL+hkjK3iyf2tdt6/s4hH+VTzNyrCrIY8AYZDSS3kGnEb+xepBTc+dBebv/ZLnJcTr7z0Y0Jn2t9wCRzUbWlf5Qd9cVsqi1gf+sQq8tyk970osqsAmqb8KG1s7GX8+sKcSh4+Vg3xbkZ49oxCAnuQtiqa9hHOKKpLsji2vUVvHCkC2/AntRMOKLpGQlQljd1WgZIevbeaQbmz12+Cm8gTEv/KD+6ZQvlSWzanGxaJhLRtPaPUlOYxce3LQPgw+fVJL1VXZk7kzSHGpeWaRsYpblvlGvXV7J1eRFap14psxQtjeSSEAuEVYpYXZBFdWEWP3v9FO82D7KjftI2wjPW6zF6qE8VfK0URvugj5VJVKJY+6BeWF/CBSuKiWjN1rrk9h3NcTnNNsPTf5B0DRsramsLs7lhcxWDo0FuvmBZUq8BRs15uds1bub+VqORkrlgRRGRiGZnY58tlTJLjQR3IWxkBbvKgkzcZg74ePeILcHdSoFMdUE1msJIsvVv5/DYgqhkPgwmqi3Kprlv+tWjVtOvmsIsXM40Pn3Jyhm/TmXB+CqgN0/2ketysq4yj1yXk398/NCk0lAhaRkhbGUFocr8LKryM8nOSKPBLAVMlXXxcqoLqtaMPtnWv1ZaZqrzJbKsKJumCcHdGwjxd4/s5wu/3gOMbXCdyt6uEy8U7zzZx9a6QtIcirqSHG7eVhvdT1aMkZm7EDZqGxwlJyONvEwnSinqS3M53m1TcE8wc89MT6M4J2MGOXcfuS7nrEv/lhVn8/yRLiIRjcOhONo5zJ/+126OdxvbAN553Vqa+4yxTGx0NhPVBVk8fbATrY0eMg1dI9y4pTp6/7duPHfW517KZOYuhI3aB3xUFmRFLxiuKsu1beZuzbRLp+mhYuzIlNzMvWvIP+XF2WTUFhnlntZF1b9/9AD93iBfvX4tYKxEben3Uup2pbRDUmV+JoFQhF5PgD3NAwCct6xw1uc7U0hwF8JG7YOj45pOrSrLpX3QZ8tipq5hH0U5GdMu/jF2ZEp+5l4+yz1QwUjLADT1edFac6BtiGvXV3DbxSvJy3Ty+vFemvtGqU1h1g5Gzh2MD853WwZxKNhQk1wp5ZlMgrsQNmod8I1rHVtfamyqccKG1EznkH/KlIylKj8z+Zz7sC+lmXtscO8Y8jE4GmRdhZs0h2LbimJeP9FLy4A3pXw7xDRFGxxlb8sAq8vcZGdIRjkRCe5C2MQfCtMz4o+WJIIxcwdsSc10D/sSXvysLMhi2BdK+JeC1prOIX9SNe1TqS7IQikjuB9qHwJgbaXRdGtHfTGner209I+mvKuR1c64fWCUvS2DnCuz9qRIcBfCJp2DRu7ZCkYAy4tzcDqULcE9mZl7dCFTgnLIwdEggVAk4fmmk+F0UJWfRXOfl0PtRkvfNRVuAHasNEo/tU6tUgaMjUMynA7eOtVPnycgwT1JEtyFsIm10KYqZuaenuZgeXF2ysE9EtF0j/gnbdIxUbQXS4LUjHVxNpWZO0BtURZNfV4OdwxTXZAV7e+ytsJNQbbxfSqVMmBs2lGVn8kLZp+ec2sS974REtyFsE3sAqZYsy2HDIUj0e97RozVqWUJLoAmO3Mf29EpteC+zFzIdLh9iHWV7uhxh0NxwQpjtasd+5FW5mfhDYRJT1OsjXkdMTUJ7kLYxCpBjJ25g5F3P9XrJRgTrBN5cn8H6/72Sf79peMMeAN87v53AFhfPf1GEhX5xlZx8WbuWmveONFLIBSZcru+mVpWlE3XsJ8TPR7WTtjk4oZN1awuy4227U2F9YG5rjIvqVbBQhYxCWGbtoFRCrLTycoYH3xWleUSimhO9XpYVZbcrPPtpn6CYc23njjMXc8dIxiOcNdNmzhv+fS9X9LTHJTmuuLO3J/Y38Gf/eJtvnztWiLmDkqJ/hJIxNpIOhzRk2bU12+o5PoNlSmd32J9YEq+PXkycxfCJq0Do3FTELOpmGns8bCqLJdvf3gDtYXZ3PupbdywqTrxEzF7sUyYuQ/7gnz9d8amHL96q4mOQR95mc5JH0QzZZVDApNm7nayZu6Sb0+ezNyFsElL/2jcfUFXlBi17id7pm+yFetUr5e64mw+dv4yPnZ+8l0Uwah1P9o5PO7Yvzx9lK5hP39y0Qr+89WTPOHvSDnfDmPB3eV0UFecem59KhtrCsjLdEarcERiCWfuSqlMpdROpdS7SqkDSqmvm8dXKKXeVEodU0r9WimVYR53mbcbzPvr5vZHEGL+aa1p6ffGrQxxZ6ZT6nZxsie5mbvWmlN9HpYX58xqLMYqVV908+oT3SP87PVG/uiC5UzqlGMAACAASURBVHzp2jUUZqfTM5JajbulKCeDnIw0zip340ybu0TA+up89v79NdE0kEgsmf8afuAKrfVGYBNwrVJqO/Bt4Hta69VAP3Cb+fjbgH6t9Srge+bjhFgQ3mrso98TsP28vZ4AvmBkyrK/FSU5nOzxJHWurmE/vmBk1jPh5cXZeAPhaM+X3af6iWj444vqyExP48NbjC3uUlmdalFK8d5zKrh2fUXK5xL2ShjctcGacqSb/zRwBfCgefw+4IPm9zeYtzHvv1Ilu+2KEHPIHwpzy0/e5I5fvh2d1drF2gR6qgU7K2cQ3BvNxy2b5cx9Yo6/oXuEjDQHy81Z703mjkixPXBS8b2PbeKOy1fZci5hn6T+jlJKpSml9gBdwDPAcWBAa22tcW4BrKs91UAzgHn/IDApUaaUul0ptUsptau7uzu1n0KIJDT3eQmEI7x2vJf/3tNq67lbzeBePc3MvWckwOBoMOG5Tpk90mc7c58Y3I93jVBXkh1Nm6wqy+XfP3Een9heN6vzi8UhqeCutQ5rrTcBNcA2YF28h5lf483SJ02TtNZ3a623aq23lpaWJjteIWbNuqBZkuviG48dYtCbONAmy9qUYrrgbowh8ez9VK8Hp0ONa0A2E2VuF26XM7pwqqFrJBrwLdecU0GFTTN3sTDN6AqI1noAeBHYDhQopaxqmxqgzfy+BagFMO/PB/rsGKwQqbAuaP7g5s0MjAb54YsNtp27pX+U/Kz06PL7iVaWWsE98UXVxl4v1YVZs75AqZSi3uwj7wuGaerzxq3iEUtbMtUypUqpAvP7LOAq4BDwAvAR82G3Ao+Y3z9q3sa8/3ltd4JTiFk42eOhKCeDHfXFnLe8kN2n+m07d0u/d9qZdm1RNg4FJ7sTz9yber2zrpSx1Jcawf1Ur5eIhvoyCe5nmmSmBpXAC0qpvcBbwDNa68eALwN/pZRqwMip32M+/h6g2Dz+V8Cd9g9biJk72eOJpkfOKs/lWOewbRdWWwdGp22Q5XKmUVOYzYkEaRmtNY29npRrxleV5dI17OedJuMDrF5m7mechIuYtNZ7gc1xjp/AyL9PPO4DPmrL6ISw0ckeD+9ZbVzfWV3mZsgXonvYP+sNoi1GjfsoF6+a/trRdOWQbzf1M+ANsKm2kGFfaNzKz9mwcuxPHehAKQnuZyJZoSrOCB5/iM4hf3TmvtoMfse6RlIO7v3eIN5AOGFr2xUlObzV2IfWmtjq4J0n+/jEPW8SDEf4wlVnAVCXYlrGCu6vNvRSXZCVcpsBsfhIbxlxRmjsNWbMVnBfVW4G9wnL9GcjUaWMpb40Z9ziIoD9rYPcdu9b1BRmsbrMzb88cxSAupLUZu61hVlkpDkIhCOTKmXEmUGCuzgjWOkQa0ZcmusiPyudozbskNQaXcCUaOZuBNkTMRdVv/bIfnJcTn5+2wX8+yfOIy/TiVKp717kTHOMfZBJSuaMJMFdnBGsVZ/WjFgpxeqyXBo6Uw/uiVanWlaUjq9111pzrHOEa84pp6ogi7qSHO754/O589q1ZKannkapL8sxv0pwPxNJcBcLypP727n2+y/jSbDB80yd6PFQkZdJdsbYZabV5bkc7UqtYsaqbnFnOsnPil/jbqnMy8TldERr3fs8AUb8oXFtBs6vK+Izl9bPejyxrBm7pGXOTHJBVSwYbQOjfOnBvQz5QhzuGEq4McVMxJZBWlaVuRnwNtPrCVCSm3wTLV8wzDMHO3n03TZ2m5s2b6hOvImEw6HGVcw0mW0GUq2Mmcpla8t44Ug3Z1fOXZ91sXBJcBcLQiSi+eID7zIaDAPGknk7g3tjj4frJuwKFK2Y6RxJOrhrrfnQv73GofYhKvIyuWpdGesq87hsTVlSz19RksMR8yKuFdyXz1Ef9C3LCvndn188J+cWC58Ed7Eg/GZXM68d7+UfP7Ser//u4Ix2LUpkwBug3xtkxYTywtXlVoOtYXbUJ7cJRPewn0PtQ3zu8lV84eqzSHPMrOHpipIcnjnYSSgc4VTv3M7cxZlNcu5iQXh8fwf1pTl8fNsyVpbkcDyJZfrJOtxhzJStYG6pyMsk1+Xk2Aw+SKzH7qgvnnFgByO4hyLGoqdTvV7K81y2XDwVYiIJ7mLe+UNhdp7s5T2rS8c1vbLLwbYhAM6uGp97Vkqxqix30pZ007Hq4lfP8iLlypiKmaY+D8uLUlusJMRUJLiLeff2qQF8wQgXryoBjCqP5n4vPjP/nqpD7UOU5GZQ5p68EnVthZvDHclXzBzrGiEv00mpe3a7GEVr3Xs8NPV5WTaH+46KM5sEdzHvXm3oIc2huGClcQG1viwXrccv9knFwfYh1k1RMXJ2VR4D3iAdQ76kznWsa4TV5W5mu7lYYXY6+VnpHGofonPIL/l2MWckuIt590pDD5tqC3CbvdCt+uyG7tRTM8FwhGOdI1OWA1pB30rdJNLQNTLrlAwYqaCVpTm8fNTYfWyuKmWEkOAu5tXgaJC9LQNcZKZkwMhLK2VsD5eq490jBMKRSfl2y9oKN5BccO8d8dPnCaS8KGhFSU60v4zM3MVckeAu5tUbJ3qJaKL5doDM9DRqC7NtmbkfajeC9lRpGXdmOsuLsznUkTi4W5Uyq8vdKY1pZcxiqlQ35RBiKhLcRdKePdhJk1mbbZeXj3aTnZHGptqCccdXleXaMnM/2DZEhtMxLqBOtK4ib8qZezAc4ZdvNuENhMaCe8ozd+P5bpeTwuzpWxYIMVsS3EVSGns83P7zXfzfF47Zdk5fMMxje9u5fG0ZGc7x/yuuKsvlRI+HcCS1nZIOtQ+zptw97X6kZ1flcarPy0icfjYPv93KVx/ex///5BEaOofJyUijMsWNpa02CMuKs2d9YVaIRCS4i6Tc/fsTRDQcmWUXxY5BH70j/nHHntzfweBokFu2LZv0+PrSHAKhCM19s/9LQWvNwfahhL1V1lXmoTUcmZCa0Vrz09caUQrue72RZw91sSqFShmL1ZlSLqaKuSTBXSTUNezjwd0tpDkUxzqHicxwNj0aCHP5d17kvG88y7Z/fJb/fOUkAL/c2URdcTbbV05e+r+mwgjIVs48Ga829PCn/7U7Wh/fOWRcAF1XOX2O3LrYOjE1s/NkH4fah/jqdesozXXROjCackoGIDvDyfvOreSqdeUpn0uIqUhwFwnd+2ojwXCE2y5egTcQpnVgdEbPP9XnYTQY5kObq1lVlss/PHaQbz5+iJ0n+7hp2zIccZbxr6t0k5HmYE/zQNKv88zBTp7Y38E/PXEYgLueM3Y12lo3fQOyqvxM8rPSOdg+fqXqva81UpCdzh9tX87f/sHZAKxJ8WKq5Ycf38KNW2psOZcQ8UjjMDGtYV+Qn79xiuvWV/Des8u5++UTHOsapnYGJXyNPUZq5baLV7Cmws1nfr6bu18+QXqa4iPnxQ9wLmca66ryeGcGwd360Ln3tUa8gRC/2dXC5y5fxfoE7XiVUqyrdHMw5q+E1oFRnjrQwe2X1JOVkcb7NlSS8QkH25NsMCbEfJOZu5jW/TubGPaF+Oyl9dESwKMzzLufMvcvXVacTXqag3+7ZQtXn13OJ3fUTdtqd3NtAftaBgmFI0m9Tmv/KBfWF7OqLJff7GrhsjWlfOHqs5J67jlV+RxuHyJovtazBzuJaLjp/FrA+AB47zkV5GVKdYtYHCS4iyn5Q2HueeUkF9YXc25NAflZ6ZTnuTjaMbNNpRt7vRTlZEQDY2Z6Gj/55Fa+9v6zp33eptoCRoPhpD9M2gZHqS/N5d9u2cLN25Zx18c2J925cVNtAf5QhCPmz7aneYBSt0sueopFS4K7mNIj77TROeTnszHbvp1V7uZo18yCe1OfZ1ZB0qp9f7clcWrG4w8x4A1SVZDFWeVuvnXjBvJnUENuvZaVBtrTPMCm2gIpVRSLlgR3EVckovnxy8c5pyqP96weWz16Vrmbhq6RGdWfN/Z4qZvFSszlxdkUZqezpylxcG8z8+1VBbOrQa8pzKI4J4M9TQMMeAOc7PFMWlglxGIiwV3Etbd1kBPdHm67eMW42etZ5bn4gsnXn/tDYdoGR2fVQ0UpxcbagqQqZqyLqTWFWTN+Heu1NtUWsKe5P/p6myW4i0VMgruIy9rAYvOywnHHxy6qJpeaaekfReuxhTsztam2gKNdw3FXj8Zqjc7cZxfcrdc63u3h98d6UAo21CTe9FqIhUqCu4jreNcIGWkOaifMhKObSifZ98WqlJltg6xNtQVoDXsT5N3bBkZxOlTcDTmSfq1lxkz9gV3NrC7LjbYgFmIxkuAu4mroGmFFSc6knizuzHSqC7KSXjlq1bgvn2Vr23NrjICbqCVv24CPivzMWe1rOvG1hnwhybeLRS9hcFdK1SqlXlBKHVJKHVBKfd48XqSUekYpdcz8WmgeV0qpf1VKNSil9iqltsz1DyHsd7x7ZMq+5VvrCnm1oSep+vNTvR7cLidFORmzGkdRTgblea5xC4ziae0fTSklA5CflU69ucfpptrCBI8WYmFLZuYeAv5aa70O2A7coZQ6G7gTeE5rvRp4zrwNcB2w2vx3O/Aj20ct5pQvGKapzxsNdBO99+wK+r1Bdp/qT3iuU31elpek1v1wbUUeh2NaA2itef5wJ+//we/58oN7ASPnXpNicIexoL6xVvLtYnFLGNy11u1a67fN74eBQ0A1cANwn/mw+4APmt/fAPxMG94ACpRSlbaPXKC1pmfET8+In9GAPZtJAzT2eohoYy/TeC5dU0pGmoNnDnYmPNepXi/Li1LbkGJtpVF+aa0e/cKv9/An9+7iaMcID73dQs+In44hX8ozd4AbNlVx6VmltvWQEWK+zCjnrpSqAzYDbwLlWut2MD4AgDLzYdVAc8zTWsxjE891u1Jql1JqV3d398xHLvinJw6z9RvPsvUbz7Ljn55j2Be05bwN5sXSqdIyuS4nF64q5plDnWg9db17KGyUTKa6ynNdRR6BcISTPR76PAEeebeNm7fV8pvP7iAU0dz3WiPhiLYluF9yVin3/cm2afu/C7EYJP1/sFIqF3gI+Eut9XQJ0Hh/f0+KAFrru7XWW7XWW0tLS5Mdhojx9MFOzq3J5y+vWs2AN8gT+zpsOW9D1whKwcqSqdvbXn12Oad6veOqZrTWvHC4i7+4/x02/8PTbPqHZwhF9KwWMMVaa7bsPdQ+xOvHe9EaPrq1lo01+awszeG+1xoBqJ5ljbsQS1FSwV0plY4R2H+htf6tebjTSreYX7vM4y1AbczTa4A2e4YrLC39Xk72ePjgpmo+f+VqVpbk8ODulhmf5/F97fzNw/vGzcCPd3uoLsgiKyNtyudZvcifPjD2gfLA7hY+de9bvHysmyvXlfOHW2v508vqee85qfUtX1mSS3qa4nDHMK809OB2OTm3Oh+lFB/YWMWQz6iBr57l6lQhlqJkqmUUcA9wSGv93Zi7HgVuNb+/FXgk5vgnzaqZ7cCglb4R9nm1oQeAi1eXoJTiw+fVsLOxb0Z7nIYjmm8+fohfvNnE6yd6o8cbuqaulLGU52WyqbZgXN792YOd1BRmsfOrV/Gdj27kb//gbL587VoKsmdXKWPJcDqoL83lcPsQrzb0sL2+OJo2+cDGqujj7EjLCLFUJDNzvwj4BHCFUmqP+e964J+Aq5VSx4CrzdsAjwMngAbgJ8Cf2T9s8UpDL6VuV3RR0Yc2V6MUPPR28rP35w510tI/SppD8eOXTgBGwD/RPcKq0sQ7Dl21rox3WwbpHvYTiWjePNnHhfXFk/ZDtcO6yjzePNlHU5+Xi1eN9bpZWZrLhup8CrLTyc6Q7QmEsCT8bdBav0L8PDrAlXEer4E7UhyXmEYkonmtoYdLziqNlhhWFWRxUX0Jv32nhc9fuTru7kYT3ftaI1X5mXzs/GV879mjHGgbxO1Kxx+KJJy5A1y+tozvPH2UF490sa4yj8HRIDvmaDOLtRVuHn6nFYCLYoI7wN/9wdm09M9sdyghljopCViEDncM0+sJTApyH9pcTXPfKPvbBhOe40jHMK8d7+UTO+r44wvryMlI4+u/O8hXHjbqxpMJ7mdX5lGRl8nzh7t4w0zr7FhZkuBZs7PW3OS6Ii9zUv391roiPrh5UkGWEGc0Ce4L2FQbUVv59otWjZ8lW615Xz/eO+k5E93zyglcTgc3nV9LvrlP6M6TfRzv8vD5K1dz3vLEKzSVUly+tpTfH+vh5WM9rCjJoSJ/bi5qrqswKmYuWlUiPdaFSIIE9wWqd8TPxq8/zZP7x5c3DvuCPLi7hfrSHCrzx19ALDNntbEXR+PZ0zzAA7tbuOWC5RSabQH+6r1n8ejnLuLVO6/gC1eflXQAvXxNGSP+EC8f7Wb7yrnbX7TU7eKL7z2LT1+yYs5eQ4ilRIL7AvVO0wDD/hA/e70xeswXDPPpn+3iePcI/9/74m9Rt6O+mLdO9kVXc04UCkf46m/3UeZ28YWrV0ePu5xpnFtTMOPGWxetKiHDrFyZq3w7GH8lfO6K1aytyJuz1xBiKZHgPkfCEU0wHEl6c+eJ9rUaefPXT/TSOjCK1pq/+s0e3jjRx3c+upHL15bFfd6OlSV4AuHo8yf66auNHGwf4u//4BxbWtrmuJxcsLIIgO3mVyHE/JPasTnQNeTjyu++xLC5uOb/3HAOn9hRN6Nz7G8dpCQ3g56RAA+/3cLK0lwe39fB/75mzbQXD60A+/rxXrZM2Gjj/p1NfOuJQ1y1roxr11fM7Ieaxh2Xr2LzssKUeqkLIewlwX0OvHi0m2FfiM9cspLfH+vh/77QwB+eX4vLOfWKz4n2tQ7yntWltA2M8ptdLfhDYc6uzOMzl6yc9nnFuS7WlLt540Qvd1y+Knr8hy808M9PHeGyNaXcddNmWy9Kbl9ZPKf5diHEzElaZg682tBDSa6LO69by53XraVzyM8j7yTfgaFryEfXsJ/11fl8+Lwamvq8dA37+eaNG5JqaLWjvphdjf0EQkZKqN8T4F+ePsJ16yv4ySe3kuOSz3QhljoJ7jbTWvNqQw8XrypGKcV7VpdwTlUeP375+JSljRNZ+fIN1flcv6GS/Kx0bt1Rl/TuQNtXFjMaDPNOk9Fv/ZWGHiIaPn3JStKl26EQZwT5TbfZkc5hekbGFhgppfjMpfWc6PbwzKHE/c/BCO5KwTlVeeS6nLz8pcv52/fHr46J5+LVJbicDh7fZ7T0eeloN/lZ6Wyska3jhDhTSHC32SvHrAVGYys1r19fQUVeJv9tLp9PZH/rICtLcqLpk/ys9KTaCVhyXU6uXFfG/+xrJxSO8PLRbi5eVZLS/qJCiMVFgrvNXjvey8rSnHEdCp1pDi5YWcTbTf3Tbm5h2dc6GN2sebY+sLGKnpEA977WSNewn0vPkp75QpxJJLjbKBiO8MaJ3nFdCy1blhXSOeSnbdA37Tm6hn10DhkXU1Nx2Zoy3C4n//L0UQDec9bc9HwRQixMZ1zZxPOHO/mfvcaS/jUVudx+Sb1t597TPIA3EObC+vjBHeDtU/1UT9F3fMQf4q9/8y4A2+pSWxCUmZ7Ge8+p4KG3W1hT7p7UqkAIsbSdUTP3YDjCnQ/t4+kDHbx0tItvPn6Y/VOs5JyNnSf7ALhgxeTAvLbSTWa6g7fNCpaJuoZ93HT367x2vJd//si5bKhJbeYO8IFNxkYWl8isXYgzzhkV3J/Y30HXsJ9/vXkzz3/xMtwuJz9+6bht5999qp9VZbnRZlyx0tMcnFtTwNtNA5PuO9nj4cM/eo3jXR7+45Nb+ejW2kmPmY2LV5XwF1eu5pMzXB0rhFj8zqjgfu+rJ6krzubSs0rJy0zn49uX8fi+dk71emZ0nh88d4xb/3MnI/5Q9FgkotnV2MfWaVrlbllWyMG2QXzBcPTYkY5hPvyj1/D4w9x/+/Ype8bMRppD8VdXn0VtUbZt5xRCLA5nTHB/t3mAt5sGuPXCumhZ4W0XrcDpcPCT359I+jy+YJi7Xz7BS0e7+fR9u6KBuqF7hCFfiK3T5Mq3LCsgGNbjmnrd93oj/mCYh/70wqQXKQkhRCJnTHC/77VGcjLS+Mh5NdFjZXmZ3Lilmgd2tTDoDSZ1nqcOdDDsD3HLBct4/UQvf/mrPWiteavRyLdPO3NfPnZR1bKnaYDNywpZUZIz1dOEEGLGllRwD4Yj/Pz1xklpln5PgMf2tXPjlppJbW5v2rYMfyjCs0muHn3o7VaqC7L4Pzes58vXruXJAx08c7CT3Y39lOS6WF48dQqkJNfFsqJsdpvBfTQQ5kjnsMzYhRC2W1LB/dE9bXztkQNc+S8v8XeP7GfIZ8zGf/tOK4FQhJu3LZv0nI01+VTlZ/LE/vaE5+8c8vHKsW5u3FKNw6H49HtWUF+aw7eeOMybJ418e6JuixfWF/P68V6C4Qj7WgcJR7QEdyGE7ZZUcL9/ZxN1xdn84fm1/NebTfzvB95Fa839O5vYVFvA2VWTd/FRSnHdhkpePtrDsG/61MzD77QS0XDjFiO140xz8NXr13Gyx0PrwChb6xLvO3r52jKG/SHeauxjT7Mxg9+0TIK7EMJeSya4H+0cZtepfm65YDnf/NAGvnTNGp460MlXH95PQ9cIH48za7dcv6GCQDjC84e7pnxMKBzh/p1NnLd8fH78irVl7DB7mU93MdVysbkt3QuHu9jTPEBNYRYlua4Z/KRCCJHYkgnu9+9sIiPNwYfNC6b/6z0r2baiiPt3NpHrcvL+jZVTPndzbSHlea5oF8V4fvtOK6d6vXz20vErWpVSfOvGDfzpZfVsSKJlgLUt3fOHu9jTNCApGSHEnFgSwd0XDPPbt1u5Zn0FReYCojSH4rt/uJGC7HRuOr+W7IypOy04HIrr1lfy4pFuPDG16/2eAKFwhGA4wg+eP8aG6nyuWje5Dr2uJIcvX7s26a6LV6wt43i3h7ZBnwR3IcScWBLB/bvPHGVwNMjN28av7KwpzOaVL1/BV69fl/Ac7zu3En8owlMHjL4zXcM+Lvr281zz/Zf5u0cP0Nw3yheuXm3L9nRXxCxUkuAuhJgLiz64/+jF49z98gn+aPuyaO47Vq7LmVQv9K3LC1lWlM1Db7cA8MCuFryBMBENv3yziY21BVy+xp7Vo8uLc1hZmoPToVLu/iiEEPEs6q6Qv9rZxLefPMwHNlbxDx9Yn9KsWinFjVuqueu5Y7T0e/nVW03sWFnMz2/bxlMHOllfnWfrptK3v2clhzuGyUxPftNsIYRI1qIO7usq87hxczXf/si5M9qpaCof3lLD9589xpce3Etz3yj/+5q1ONMcvO/cqS/GztZN01TvCCFEqhKmZZRS/6mU6lJK7Y85VqSUekYpdcz8WmgeV0qpf1VKNSil9iqltszl4DfWFvDdj22ybdPn2qJstq0o4rXjvRRmp3PNOeW2nFcIIU63ZKLivcC1E47dCTyntV4NPGfeBrgOWG3+ux34kT3DPH0+Yi5Q+sh5NbickjIRQixOCYO71vploG/C4RuA+8zv7wM+GHP8Z9rwBlCglLI/pzGH/mBjFbddvIL/9Z6V8z0UIYSYtdnmM8q11u0A5lerjKQaaI55XIt5bBKl1O1KqV1KqV3d3d2zHIb9sjLS+Nr7z6Y8L3O+hyKEELNmdylkvKuaOt4DtdZ3a623aq23lpaW2jwMIYQ4s802uHda6Rbzq9WUpQWIXUlUA7TNfnhCCCFmY7bB/VHgVvP7W4FHYo5/0qya2Q4MWukbIYQQp0/COnel1P3AZUCJUqoF+Dvgn4DfKKVuA5qAj5oPfxy4HmgAvMCn5mDMQgghEkgY3LXWN09x15VxHquBO1IdlBBCiNQs+t4yQgghJpPgLoQQS5AEdyGEWIKUkSaf50Eo1Q2cmuXTS4AeG4czlxbLWBfLOEHGOhcWyzhh8Yx1rsa5XGsdd6HQggjuqVBK7dJab53vcSRjsYx1sYwTZKxzYbGMExbPWOdjnJKWEUKIJUiCuxBCLEFLIbjfPd8DmIHFMtbFMk6Qsc6FxTJOWDxjPe3jXPQ5dyGEEJMthZm7EEKICSS4CyHEErSog7tS6lql1BFzz9Y7Ez/j9FBK1SqlXlBKHVJKHVBKfd48Hnfv2YVAKZWmlHpHKfWYeXuFUupNc6y/VkplLIAxFiilHlRKHTbf2x0L9T1VSn3B/G+/Xyl1v1Iqc6G8pwt5X+QkxvnP5n//vUqph5VSBTH3fcUc5xGl1DWna5xTjTXmvi8qpbRSqsS8fVre00Ub3JVSacAPMfZtPRu4WSl19vyOKioE/LXWeh2wHbjDHNtUe88uBJ8HDsXc/jbwPXOs/cBt8zKq8e4CntRarwU2Yox3wb2nSqlq4C+ArVrr9UAacBML5z29l8WxL/K9TB7nM8B6rfW5wFHgKwDm79dNwDnmc/7NjBGny71MHitKqVrgaozuuZbT855qrRflP2AH8FTM7a8AX5nvcU0x1kfM/8BHgErzWCVwZL7HZo6lBuMX+grgMYwdtXoAZ7z3ep7GmAecxCwCiDm+4N5TxrabLMLovPoYcM1Cek+BOmB/ovcR+Hfg5niPm49xTrjvQ8AvzO/H/f4DTwE75vM9NY89iDERaQRKTud7umhn7sxgv9b5pJSqAzYDbzL13rPz7fvAl4CIebsYGNBah8zbC+G9XQl0Az8100f/oZTKYQG+p1rrVuA7GLO1dmAQ2M3Ce09jpbwv8jz4E+AJ8/sFN06l1AeAVq31uxPuOi1jXczBPen9WueLUioXeAj4S6310HyPJx6l1PuBLq317tjDcR463++tE9gC/EhrvRnwsABSMPGY+eobgBVAFZCD8af4RPP9niZjIf6/gFLqbzDSn7+wDsV52LyNUymVDfwN8Lfx7o5zzPaxLubgvqD3a1VKpWME9l9orX9rHp5q79n5dBHwAaVUI/ArjNTM94ECpZS1mctCeG9bgBat9ZvmZWSfPAAAAZxJREFU7Qcxgv1CfE+vAk5qrbu11kHgt8CFLLz3NNai2RdZKXUr8H7gFm3mNVh446zH+HB/1/zdqgHeVkpVcJrGupiD+1vAarMCIQPjYsqj8zwmwLgaDtwDHNJafzfmrqn2np03WuuvaK1rtNZ1GO/h81rrW4AXgI+YD5v3sWqtO4BmpdQa89CVwEEW4HuKkY7ZrpTKNv9fsMa6oN7TCRbFvshKqWuBLwMf0Fp7Y+56FLhJKeVSSq3AuFi5cz7GCKC13qe1LtNa15m/Wy3AFvP/49Pznp7OCw5zcAHjeowr5seBv5nv8cSM62KMP7P2AnvMf9dj5LKfA46ZX4vme6wTxn0Z8Jj5/UqMX44G4AHAtQDGtwnYZb6v/w0ULtT3FPg6cBjYD/wccC2U9xS4H+NaQBAj6Nw21fuIkUL4ofk7tg+jAmg+x9mAka+2fq9+HPP4vzHHeQS4br7f0wn3NzJ2QfW0vKfSfkAIIZagxZyWEUIIMQUJ7kIIsQRJcBdCiCVIgrsQQixBEtyFEGIJkuAuhBBLkAR3IYRYgv4flRQvJEdX1I4AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot the data\n",
    "plt.plot(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# scale the data\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "scale = MinMaxScaler(feature_range=(0,1))\n",
    "data = scale.fit_transform(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# divide the data into number of steps \n",
    "def create_data(data, seq_len):\n",
    "    N = len(data)\n",
    "    X = []\n",
    "    Y = []\n",
    "    for i in range(N-seq_len-1):\n",
    "        x = data[i:i+seq_len]\n",
    "        X.append(x)\n",
    "        y = data[i+seq_len]\n",
    "        Y.append(y)\n",
    "        \n",
    "    return X,Y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "x,y = create_data(data,10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# convert the x and y values into arrays\n",
    "x, y = np.asarray(x), np.asarray(y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(133, 10)"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x.reshape(-1,10).shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# divide the data into train and test\n",
    "def split_data(x,y, ratio):\n",
    "    assert len(x)==len(y)\n",
    "    N = len(x)\n",
    "    train_x, test_x  = x[:int(N*ratio)], x[int(N*ratio):]\n",
    "    train_y, test_y  = y[:int(N*ratio)], y[int(N*ratio):]\n",
    "    return train_x,train_y, test_x, test_y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_x,train_y, test_x, test_y = split_data(x,y,0.7)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.lines.Line2D at 0x262f6c92708>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(np.vstack((train_y,test_y)))\n",
    "plt.axvline(int(len(x)*0.7))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Implement the RNN \n",
    "class Simple_RNN(nn.Module):\n",
    "    def __init__(self, in_dim, hid_dim, out_dim, num_layers):\n",
    "        super().__init__()\n",
    "        \n",
    "        # define the dimensions\n",
    "        self.in_dim = in_dim\n",
    "        self.hid_dim = in_dim\n",
    "        self.out_dim = out_dim\n",
    "        self.layer = num_layers\n",
    "        \n",
    "        # define the rnn layer\n",
    "        self.rnn = nn.RNN(self.in_dim, self.hid_dim, self.layer, nonlinearity='tanh', batch_first=True)\n",
    "        \n",
    "        # define fully connected layer for output\n",
    "        self.fc = nn.Linear(self.hid_dim, self.out_dim)\n",
    "        \n",
    "    def forward(self,x):\n",
    "        \n",
    "        # initialize the hidden layer\n",
    "        h0 = torch.zeros(self.layer, x.size(0),self.hid_dim)\n",
    "        \n",
    "        #initialize the rnn\n",
    "        out, _ = self.rnn(x,h0)\n",
    "        \n",
    "        out = self.fc(out[:,-1,:])\n",
    "        return out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# convert the data from numpy to tensor\n",
    "train_x = torch.from_numpy(train_x.astype(np.float32))\n",
    "train_y = torch.from_numpy(train_y.astype(np.float32))\n",
    "test_x = torch.from_numpy(test_x.astype(np.float32))\n",
    "test_y = torch.from_numpy(test_y.astype(np.float32))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(133, 10, 1)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# define the network\n",
    "in_size = x.shape[-1]\n",
    "out_size = in_size\n",
    "hid_size = 3\n",
    "num_layers = 1\n",
    "\n",
    "model = Simple_RNN(in_size,hid_size,out_size, num_layers)\n",
    "\n",
    "optimizer = torch.optim.Adam(model.parameters(), lr=0.01)\n",
    "loss_fun = nn.MSELoss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train_model(model,\n",
    "               criterion,\n",
    "               optimizer,\n",
    "               x_train,\n",
    "               x_test,\n",
    "               y_train,\n",
    "               y_test,\n",
    "               epochs=500):\n",
    "    \n",
    "    train_loss = np.zeros(epochs)\n",
    "    test_loss = np.zeros(epochs)\n",
    "    \n",
    "    for epoch in range(epochs):\n",
    "        \n",
    "        # put default model grads to zero\n",
    "        optimizer.zero_grad()\n",
    "        \n",
    "        # predict the output\n",
    "        pred = model(x_train)\n",
    "        \n",
    "        # calculate the loss \n",
    "        error = criterion(pred,y_train)\n",
    "        \n",
    "        # backpropagate the error\n",
    "        error.backward()\n",
    "        \n",
    "        # update the model parameters\n",
    "        optimizer.step()\n",
    "        \n",
    "        # save the losses \n",
    "        train_loss[epoch] = error.item()\n",
    "        \n",
    "        # test loss \n",
    "        test_pred = model(x_test)\n",
    "        test_error = criterion(y_test,test_pred)\n",
    "        test_loss[epoch] = test_error.item()\n",
    "        \n",
    "        if (epoch+1) % 5 ==0:\n",
    "            print('Epoch :{}    Train Loss :{}    Test Loss :{}'.format((epoch+1)/epochs, error.item(), test_error.item()))\n",
    "            \n",
    "    return train_loss, test_loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch :0.01    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.02    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.03    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.04    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.05    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.06    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.07    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.08    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.09    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.1    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.11    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.12    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.13    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.14    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.15    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.16    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.17    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.18    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.19    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.2    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.21    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.22    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.23    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.24    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.25    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.26    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.27    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.28    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.29    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.3    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.31    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.32    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.33    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.34    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.35    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.36    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.37    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.38    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.39    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.4    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.41    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.42    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.43    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.44    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.45    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.46    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.47    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.48    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.49    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.5    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.51    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.52    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.53    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.54    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.55    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.56    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.57    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.58    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.59    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.6    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.61    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.62    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.63    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.64    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.65    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.66    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.67    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.68    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.69    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.7    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.71    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.72    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.73    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.74    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.75    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.76    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.77    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.78    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.79    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.8    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.81    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.82    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.83    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.84    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.85    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.86    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.87    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.88    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.89    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.9    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.91    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.92    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.93    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.94    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.95    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.96    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.97    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.98    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :0.99    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n",
      "Epoch :1.0    Train Loss :0.0019059231271967292    Test Loss :0.007607039995491505\n"
     ]
    }
   ],
   "source": [
    "train_loss, test_loss = train_model(model,\n",
    "               loss_fun,\n",
    "               optimizer,\n",
    "               train_x,\n",
    "               test_x,\n",
    "               train_y,\n",
    "               test_y,\n",
    "               epochs=500)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x262f6d3ee88>"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(train_loss,label='Train_loss')\n",
    "plt.plot(test_loss,label='Test Loss')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x262f6dc1f08>"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# testing the prediction quality if the trained model\n",
    "# 1) Multi Step prediction\n",
    "#plt.plot(train_y.detach().numpy())\n",
    "plt.plot(np.vstack((train_y.detach().numpy(),test_y.detach().numpy())),label='Target')\n",
    "plt.plot(np.vstack((train_y.detach().numpy(),model(test_x).detach().numpy())),label='Predition')\n",
    "plt.axvline(int(len(x)*0.7))\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# testing the predction model on multiple time series\n",
    "last_x = train_x[-1].view(10)\n",
    "\n",
    "prediction_val = []\n",
    "\n",
    "while len(prediction_val)<len(test_y):\n",
    "    prediction = model(last_x.view(1,10,1))\n",
    "    prediction_val.append(prediction[0,0].item())\n",
    "    \n",
    "    \n",
    "    ## replace the predicted value in last x\n",
    "    last_x = torch.cat((last_x[1:],prediction[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x262f6e5de48>"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot the result\n",
    "plt.plot(np.vstack((train_y.detach().numpy(),test_y.detach().numpy())),label='Target')\n",
    "plt.plot(np.vstack((train_y.detach().numpy(),np.asarray(prediction_val).reshape(-1,1))),label='Predition')\n",
    "plt.axvline(int(len(x)*0.7),color='red')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Implement LSTM\n",
    "class LSTM(nn.Module):\n",
    "    def __init__(self,in_dim, hid_dim, out_dim, num_layers):\n",
    "        super().__init__()\n",
    "        self.in_dim = in_dim\n",
    "        self.hid_dim = hid_dim\n",
    "        self.out_dim = out_dim\n",
    "        self.layer = num_layers\n",
    "        \n",
    "        self.lstm = nn.LSTM(self.in_dim, self.hid_dim, self.layer, batch_first=True)\n",
    "        \n",
    "        self.fc = nn.Linear(self.hid_dim, self.out_dim)\n",
    "        \n",
    "    def forward(self,x):\n",
    "        \n",
    "        # initialize the hidden state\n",
    "        h0 = torch.zeros(self.layer, x.size(0), self.hid_dim)\n",
    "        c0 = torch.zeros(self.layer, x.size(0), self.hid_dim)\n",
    "        \n",
    "        # forward propagate\n",
    "        out, (h_,c_) = self.lstm(x,(h0,c0))\n",
    "        \n",
    "        out = self.fc(out[:,-1,:])\n",
    "        return out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "# define the network\n",
    "in_size = x.shape[-1]\n",
    "out_size = in_size\n",
    "hid_size = 5\n",
    "num_layers = 1\n",
    "\n",
    "lstm = LSTM(in_size,hid_size,out_size, num_layers)\n",
    "\n",
    "optimizer = torch.optim.Adam(lstm.parameters(), lr=0.01)\n",
    "loss_fun = nn.MSELoss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch :0.01    Train Loss :0.08809886872768402    Test Loss :0.28696876764297485\n",
      "Epoch :0.02    Train Loss :0.015262849628925323    Test Loss :0.07489967346191406\n",
      "Epoch :0.03    Train Loss :0.036908067762851715    Test Loss :0.036181848496198654\n",
      "Epoch :0.04    Train Loss :0.022008005529642105    Test Loss :0.06553913652896881\n",
      "Epoch :0.05    Train Loss :0.015110929496586323    Test Loss :0.11560776084661484\n",
      "Epoch :0.06    Train Loss :0.018217362463474274    Test Loss :0.13421541452407837\n",
      "Epoch :0.07    Train Loss :0.0161757729947567    Test Loss :0.11360581964254379\n",
      "Epoch :0.08    Train Loss :0.013309880159795284    Test Loss :0.08209600299596786\n",
      "Epoch :0.09    Train Loss :0.013338676653802395    Test Loss :0.06477804481983185\n",
      "Epoch :0.1    Train Loss :0.012481522746384144    Test Loss :0.06501168012619019\n",
      "Epoch :0.11    Train Loss :0.01127034705132246    Test Loss :0.07164265215396881\n",
      "Epoch :0.12    Train Loss :0.010755680501461029    Test Loss :0.07118548452854156\n",
      "Epoch :0.13    Train Loss :0.009889790788292885    Test Loss :0.06143394112586975\n",
      "Epoch :0.14    Train Loss :0.009098663926124573    Test Loss :0.05105060338973999\n",
      "Epoch :0.15    Train Loss :0.008405645377933979    Test Loss :0.045948125422000885\n",
      "Epoch :0.16    Train Loss :0.007649722509086132    Test Loss :0.0439288504421711\n",
      "Epoch :0.17    Train Loss :0.00700091477483511    Test Loss :0.04024260491132736\n",
      "Epoch :0.18    Train Loss :0.006359138526022434    Test Loss :0.03450551629066467\n",
      "Epoch :0.19    Train Loss :0.005796161945909262    Test Loss :0.02978162094950676\n",
      "Epoch :0.2    Train Loss :0.005292602814733982    Test Loss :0.02690579555928707\n",
      "Epoch :0.21    Train Loss :0.0048720696941018105    Test Loss :0.024326683953404427\n",
      "Epoch :0.22    Train Loss :0.0045265560038387775    Test Loss :0.021546104922890663\n",
      "Epoch :0.23    Train Loss :0.004261827562004328    Test Loss :0.019459405913949013\n",
      "Epoch :0.24    Train Loss :0.004067794419825077    Test Loss :0.018160875886678696\n",
      "Epoch :0.25    Train Loss :0.00393678667023778    Test Loss :0.017122400924563408\n",
      "Epoch :0.26    Train Loss :0.0038532610051333904    Test Loss :0.01631942205131054\n",
      "Epoch :0.27    Train Loss :0.0038043097592890263    Test Loss :0.015858223661780357\n",
      "Epoch :0.28    Train Loss :0.0037767875473946333    Test Loss :0.01557148713618517\n",
      "Epoch :0.29    Train Loss :0.003760602790862322    Test Loss :0.015356360003352165\n",
      "Epoch :0.3    Train Loss :0.0037493393756449223    Test Loss :0.015227245166897774\n",
      "Epoch :0.31    Train Loss :0.003739241510629654    Test Loss :0.015147899277508259\n",
      "Epoch :0.32    Train Loss :0.0037288519088178873    Test Loss :0.015079277567565441\n",
      "Epoch :0.33    Train Loss :0.003717884188517928    Test Loss :0.015028010122478008\n",
      "Epoch :0.34    Train Loss :0.0037064929492771626    Test Loss :0.014989686198532581\n",
      "Epoch :0.35    Train Loss :0.0036949424538761377    Test Loss :0.014949080534279346\n",
      "Epoch :0.36    Train Loss :0.003683389164507389    Test Loss :0.014909912832081318\n",
      "Epoch :0.37    Train Loss :0.0036718850024044514    Test Loss :0.01487182267010212\n",
      "Epoch :0.38    Train Loss :0.0036604120396077633    Test Loss :0.014827914535999298\n",
      "Epoch :0.39    Train Loss :0.00364893302321434    Test Loss :0.014781099744141102\n",
      "Epoch :0.4    Train Loss :0.0036374046467244625    Test Loss :0.01473215501755476\n",
      "Epoch :0.41    Train Loss :0.003625804791226983    Test Loss :0.014678806066513062\n",
      "Epoch :0.42    Train Loss :0.0036141176242381334    Test Loss :0.014623461291193962\n",
      "Epoch :0.43    Train Loss :0.0036023417487740517    Test Loss :0.014566848985850811\n",
      "Epoch :0.44    Train Loss :0.003590479725971818    Test Loss :0.014508381485939026\n",
      "Epoch :0.45    Train Loss :0.003578530391678214    Test Loss :0.014449439942836761\n",
      "Epoch :0.46    Train Loss :0.0035665002651512623    Test Loss :0.014390222728252411\n",
      "Epoch :0.47    Train Loss :0.003554390277713537    Test Loss :0.014330485835671425\n",
      "Epoch :0.48    Train Loss :0.0035422041546553373    Test Loss :0.014270812273025513\n",
      "Epoch :0.49    Train Loss :0.0035299404989928007    Test Loss :0.014211068861186504\n",
      "Epoch :0.5    Train Loss :0.0035176032688468695    Test Loss :0.014151133596897125\n",
      "Epoch :0.51    Train Loss :0.003505192929878831    Test Loss :0.014091181568801403\n",
      "Epoch :0.52    Train Loss :0.003492712741717696    Test Loss :0.014031043276190758\n",
      "Epoch :0.53    Train Loss :0.003480160841718316    Test Loss :0.013970673084259033\n",
      "Epoch :0.54    Train Loss :0.003467540955170989    Test Loss :0.01391011755913496\n",
      "Epoch :0.55    Train Loss :0.0034548528492450714    Test Loss :0.013849271461367607\n",
      "Epoch :0.56    Train Loss :0.0034421009477227926    Test Loss :0.013788148760795593\n",
      "Epoch :0.57    Train Loss :0.003429282922297716    Test Loss :0.013726751320064068\n",
      "Epoch :0.58    Train Loss :0.003416401334106922    Test Loss :0.013665053062140942\n",
      "Epoch :0.59    Train Loss :0.00340345804579556    Test Loss :0.013603064231574535\n",
      "Epoch :0.6    Train Loss :0.003390454687178135    Test Loss :0.01354079507291317\n",
      "Epoch :0.61    Train Loss :0.003377391491085291    Test Loss :0.013478239998221397\n",
      "Epoch :0.62    Train Loss :0.003364270320162177    Test Loss :0.013415413908660412\n",
      "Epoch :0.63    Train Loss :0.0033510939683765173    Test Loss :0.013352297246456146\n",
      "Epoch :0.64    Train Loss :0.003337862202897668    Test Loss :0.013288915157318115\n",
      "Epoch :0.65    Train Loss :0.0033245764207094908    Test Loss :0.013225262053310871\n",
      "Epoch :0.66    Train Loss :0.00331123941577971    Test Loss :0.013161331415176392\n",
      "Epoch :0.67    Train Loss :0.0032978535164147615    Test Loss :0.013097132556140423\n",
      "Epoch :0.68    Train Loss :0.0032844184897840023    Test Loss :0.013032659888267517\n",
      "Epoch :0.69    Train Loss :0.0032709375955164433    Test Loss :0.012967916205525398\n",
      "Epoch :0.7    Train Loss :0.0032574129290878773    Test Loss :0.012902897782623768\n",
      "Epoch :0.71    Train Loss :0.0032438465859740973    Test Loss :0.012837611138820648\n",
      "Epoch :0.72    Train Loss :0.003230239264667034    Test Loss :0.01277205627411604\n",
      "Epoch :0.73    Train Loss :0.003216594224795699    Test Loss :0.012706239707767963\n",
      "Epoch :0.74    Train Loss :0.003202914260327816    Test Loss :0.012640158645808697\n",
      "Epoch :0.75    Train Loss :0.0031891996040940285    Test Loss :0.012573826126754284\n",
      "Epoch :0.76    Train Loss :0.0031754537485539913    Test Loss :0.012507235631346703\n",
      "Epoch :0.77    Train Loss :0.0031616801861673594    Test Loss :0.012440408580005169\n",
      "Epoch :0.78    Train Loss :0.0031478789169341326    Test Loss :0.012373336590826511\n",
      "Epoch :0.79    Train Loss :0.0031340564601123333    Test Loss :0.012306037358939648\n",
      "Epoch :0.8    Train Loss :0.0031202109530568123    Test Loss :0.012238514609634876\n",
      "Epoch :0.81    Train Loss :0.003106348915025592    Test Loss :0.012170783244073391\n",
      "Epoch :0.82    Train Loss :0.003092471044510603    Test Loss :0.012102853506803513\n",
      "Epoch :0.83    Train Loss :0.0030785799026489258    Test Loss :0.012034730054438114\n",
      "Epoch :0.84    Train Loss :0.0030646787490695715    Test Loss :0.011966431513428688\n",
      "Epoch :0.85    Train Loss :0.003050771774724126    Test Loss :0.011897968128323555\n",
      "Epoch :0.86    Train Loss :0.003036860143765807    Test Loss :0.011829348281025887\n",
      "Epoch :0.87    Train Loss :0.0030229485128074884    Test Loss :0.011760594323277473\n",
      "Epoch :0.88    Train Loss :0.00300903944298625    Test Loss :0.011691724881529808\n",
      "Epoch :0.89    Train Loss :0.0029951357282698154    Test Loss :0.01162273995578289\n",
      "Epoch :0.9    Train Loss :0.0029812403954565525    Test Loss :0.011553668417036533\n",
      "Epoch :0.91    Train Loss :0.0029673564713448286    Test Loss :0.011484524235129356\n",
      "Epoch :0.92    Train Loss :0.0029534881468862295    Test Loss :0.011415321379899979\n",
      "Epoch :0.93    Train Loss :0.0029396384488791227    Test Loss :0.011346083134412766\n",
      "Epoch :0.94    Train Loss :0.002925810171291232    Test Loss :0.011276825331151485\n",
      "Epoch :0.95    Train Loss :0.0029120068065822124    Test Loss :0.011207564733922482\n",
      "Epoch :0.96    Train Loss :0.002898231614381075    Test Loss :0.011138318106532097\n",
      "Epoch :0.97    Train Loss :0.002884488319978118    Test Loss :0.011069112457334995\n",
      "Epoch :0.98    Train Loss :0.002870778553187847    Test Loss :0.010999967344105244\n",
      "Epoch :0.99    Train Loss :0.0028571072034537792    Test Loss :0.010930889286100864\n",
      "Epoch :1.0    Train Loss :0.0028434765990823507    Test Loss :0.010861916467547417\n"
     ]
    }
   ],
   "source": [
    "train_loss, test_loss = train_model(lstm,\n",
    "               loss_fun,\n",
    "               optimizer,\n",
    "               train_x,\n",
    "               test_x,\n",
    "               train_y,\n",
    "               test_y,\n",
    "               epochs=500)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x262f82ddac8>"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(train_loss,label='Train_loss')\n",
    "plt.plot(test_loss,label='Test Loss')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x262f8369708>"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# testing the prediction quality if the trained model\n",
    "# 1) Multi Step prediction\n",
    "#plt.plot(train_y.detach().numpy())\n",
    "plt.plot(np.vstack((train_y.detach().numpy(),test_y.detach().numpy())),label='Target')\n",
    "plt.plot(np.vstack((train_y.detach().numpy(),lstm(test_x).detach().numpy())),label='Predition')\n",
    "plt.axvline(int(len(x)*0.7))\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x262f83e1d48>"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# testing the predction model on multiple time series\n",
    "last_x = train_x[-1].view(10)\n",
    "\n",
    "prediction_val = []\n",
    "\n",
    "while len(prediction_val)<len(test_y):\n",
    "    prediction = lstm(last_x.view(1,10,1))\n",
    "    prediction_val.append(prediction[0,0].item())\n",
    "    \n",
    "    \n",
    "    ## replace the predicted value in last x\n",
    "    last_x = torch.cat((last_x[1:],prediction[0]))\n",
    "    \n",
    "# plot the result\n",
    "plt.plot(np.vstack((train_y.detach().numpy(),test_y.detach().numpy())),label='Target')\n",
    "plt.plot(np.vstack((train_y.detach().numpy(),np.asarray(prediction_val).reshape(-1,1))),label='Predition')\n",
    "plt.axvline(int(len(x)*0.7),color='red')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "# implement GRU\n",
    "class GRU(nn.Module):\n",
    "    def __init__(self,in_dim, hid_dim, out_dim, num_layer):\n",
    "        super().__init__()\n",
    "        \n",
    "        self.in_dim = in_dim\n",
    "        self.hid_dim = hid_dim\n",
    "        self.out_dim = out_dim\n",
    "        self.layer = num_layers\n",
    "        \n",
    "        self.gru = nn.GRU(self.in_dim, self.hid_dim, self.layer, batch_first=True)\n",
    "        self.fc = nn.Linear(self.hid_dim, self.out_dim)\n",
    "        \n",
    "    def forward(self,x):\n",
    "        \n",
    "        # initalize the hidden layers\n",
    "        h0 = torch.zeros(self.layer, x.size(0), self.hid_dim)\n",
    "        \n",
    "        out, _ = self.gru(x,h0)\n",
    "        out = self.fc(out[:,-1,:])\n",
    "        return out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "# define the network\n",
    "in_size = x.shape[-1]\n",
    "out_size = in_size\n",
    "hid_size = 5\n",
    "num_layers = 1\n",
    "\n",
    "gru = GRU(in_size,hid_size,out_size, num_layers)\n",
    "\n",
    "optimizer = torch.optim.Adam(gru.parameters(), lr=0.01)\n",
    "loss_fun = nn.MSELoss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch :0.01    Train Loss :0.10431549698114395    Test Loss :0.4031112790107727\n",
      "Epoch :0.02    Train Loss :0.02498302236199379    Test Loss :0.1582561433315277\n",
      "Epoch :0.03    Train Loss :0.03889799490571022    Test Loss :0.07269664108753204\n",
      "Epoch :0.04    Train Loss :0.0334240198135376    Test Loss :0.09290338307619095\n",
      "Epoch :0.05    Train Loss :0.02117075026035309    Test Loss :0.15183646976947784\n",
      "Epoch :0.06    Train Loss :0.023604482412338257    Test Loss :0.18468281626701355\n",
      "Epoch :0.07    Train Loss :0.022204110398888588    Test Loss :0.16413189470767975\n",
      "Epoch :0.08    Train Loss :0.01891299895942211    Test Loss :0.12566806375980377\n",
      "Epoch :0.09    Train Loss :0.018770959228277206    Test Loss :0.10409481823444366\n",
      "Epoch :0.1    Train Loss :0.017808157950639725    Test Loss :0.10431643575429916\n",
      "Epoch :0.11    Train Loss :0.016466788947582245    Test Loss :0.11257024109363556\n",
      "Epoch :0.12    Train Loss :0.01581973023712635    Test Loss :0.11222221702337265\n",
      "Epoch :0.13    Train Loss :0.014777449890971184    Test Loss :0.10030911862850189\n",
      "Epoch :0.14    Train Loss :0.013765604235231876    Test Loss :0.08728974312543869\n",
      "Epoch :0.15    Train Loss :0.012768346816301346    Test Loss :0.08044538646936417\n",
      "Epoch :0.16    Train Loss :0.011641453951597214    Test Loss :0.076799675822258\n",
      "Epoch :0.17    Train Loss :0.010536393150687218    Test Loss :0.0699043720960617\n",
      "Epoch :0.18    Train Loss :0.00933805014938116    Test Loss :0.05962339788675308\n",
      "Epoch :0.19    Train Loss :0.00814695656299591    Test Loss :0.05064023286104202\n",
      "Epoch :0.2    Train Loss :0.006977424491196871    Test Loss :0.04342571273446083\n",
      "Epoch :0.21    Train Loss :0.005929416511207819    Test Loss :0.03540901467204094\n",
      "Epoch :0.22    Train Loss :0.005098048131912947    Test Loss :0.028162404894828796\n",
      "Epoch :0.23    Train Loss :0.004601047839969397    Test Loss :0.023445507511496544\n",
      "Epoch :0.24    Train Loss :0.0044433181174099445    Test Loss :0.020302824676036835\n",
      "Epoch :0.25    Train Loss :0.0044457376934587955    Test Loss :0.018978606909513474\n",
      "Epoch :0.26    Train Loss :0.004399802070111036    Test Loss :0.01864570565521717\n",
      "Epoch :0.27    Train Loss :0.004286620300263166    Test Loss :0.01884155534207821\n",
      "Epoch :0.28    Train Loss :0.004181328695267439    Test Loss :0.019309453666210175\n",
      "Epoch :0.29    Train Loss :0.004106777720153332    Test Loss :0.019596334546804428\n",
      "Epoch :0.3    Train Loss :0.004046042449772358    Test Loss :0.019682306796312332\n",
      "Epoch :0.31    Train Loss :0.003984805662184954    Test Loss :0.019412655383348465\n",
      "Epoch :0.32    Train Loss :0.003921032417565584    Test Loss :0.019002767279744148\n",
      "Epoch :0.33    Train Loss :0.003858159761875868    Test Loss :0.018497256562113762\n",
      "Epoch :0.34    Train Loss :0.003798697143793106    Test Loss :0.018028605729341507\n",
      "Epoch :0.35    Train Loss :0.003742682281881571    Test Loss :0.017630966380238533\n",
      "Epoch :0.36    Train Loss :0.0036891340278089046    Test Loss :0.01730974204838276\n",
      "Epoch :0.37    Train Loss :0.0036374686751514673    Test Loss :0.01705753244459629\n",
      "Epoch :0.38    Train Loss :0.003587643150240183    Test Loss :0.016834929585456848\n",
      "Epoch :0.39    Train Loss :0.00353971472941339    Test Loss :0.016630519181489944\n",
      "Epoch :0.4    Train Loss :0.0034935905132442713    Test Loss :0.016416655853390694\n",
      "Epoch :0.41    Train Loss :0.0034490779507905245    Test Loss :0.016197100281715393\n",
      "Epoch :0.42    Train Loss :0.00340600055642426    Test Loss :0.015965383499860764\n",
      "Epoch :0.43    Train Loss :0.0033642235212028027    Test Loss :0.015732595697045326\n",
      "Epoch :0.44    Train Loss :0.0033236457966268063    Test Loss :0.015500010922551155\n",
      "Epoch :0.45    Train Loss :0.003284172620624304    Test Loss :0.015274669043719769\n",
      "Epoch :0.46    Train Loss :0.003245705971494317    Test Loss :0.015056301839649677\n",
      "Epoch :0.47    Train Loss :0.003208165755495429    Test Loss :0.014846155419945717\n",
      "Epoch :0.48    Train Loss :0.0031714763026684523    Test Loss :0.014642161317169666\n",
      "Epoch :0.49    Train Loss :0.003135579638183117    Test Loss :0.014443283900618553\n",
      "Epoch :0.5    Train Loss :0.003100421978160739    Test Loss :0.014247918501496315\n",
      "Epoch :0.51    Train Loss :0.003065959783270955    Test Loss :0.014055361971259117\n",
      "Epoch :0.52    Train Loss :0.0030321532394737005    Test Loss :0.013865184970200062\n",
      "Epoch :0.53    Train Loss :0.0029989704489707947    Test Loss :0.01367739774286747\n",
      "Epoch :0.54    Train Loss :0.0029663837049156427    Test Loss :0.013492186553776264\n",
      "Epoch :0.55    Train Loss :0.0029343722853809595    Test Loss :0.013309699483215809\n",
      "Epoch :0.56    Train Loss :0.002902914537116885    Test Loss :0.013130116276443005\n",
      "Epoch :0.57    Train Loss :0.0028719978872686625    Test Loss :0.012953436002135277\n",
      "Epoch :0.58    Train Loss :0.002841606503352523    Test Loss :0.012779654935002327\n",
      "Epoch :0.59    Train Loss :0.002811733866110444    Test Loss :0.012608666904270649\n",
      "Epoch :0.6    Train Loss :0.0027823701966553926    Test Loss :0.012440393678843975\n",
      "Epoch :0.61    Train Loss :0.00275351176969707    Test Loss :0.012274740263819695\n",
      "Epoch :0.62    Train Loss :0.002725152298808098    Test Loss :0.012111661024391651\n",
      "Epoch :0.63    Train Loss :0.002697288291528821    Test Loss :0.011951101943850517\n",
      "Epoch :0.64    Train Loss :0.0026699204463511705    Test Loss :0.011793045327067375\n",
      "Epoch :0.65    Train Loss :0.0026430441066622734    Test Loss :0.011637477204203606\n",
      "Epoch :0.66    Train Loss :0.002616660436615348    Test Loss :0.011484387330710888\n",
      "Epoch :0.67    Train Loss :0.0025907675735652447    Test Loss :0.011333764530718327\n",
      "Epoch :0.68    Train Loss :0.0025653678458184004    Test Loss :0.011185596697032452\n",
      "Epoch :0.69    Train Loss :0.0025404624175280333    Test Loss :0.011039878241717815\n",
      "Epoch :0.7    Train Loss :0.002516050823032856    Test Loss :0.010896587744355202\n",
      "Epoch :0.71    Train Loss :0.0024921356234699488    Test Loss :0.01075572706758976\n",
      "Epoch :0.72    Train Loss :0.002468719845637679    Test Loss :0.010617291554808617\n",
      "Epoch :0.73    Train Loss :0.002445804188027978    Test Loss :0.010481296107172966\n",
      "Epoch :0.74    Train Loss :0.0024233912117779255    Test Loss :0.01034774910658598\n",
      "Epoch :0.75    Train Loss :0.0024014839436858892    Test Loss :0.01021668966859579\n",
      "Epoch :0.76    Train Loss :0.0023800842463970184    Test Loss :0.010088128037750721\n",
      "Epoch :0.77    Train Loss :0.002359196078032255    Test Loss :0.009962107986211777\n",
      "Epoch :0.78    Train Loss :0.002338821068406105    Test Loss :0.009838671423494816\n",
      "Epoch :0.79    Train Loss :0.0023189601488411427    Test Loss :0.009717855602502823\n",
      "Epoch :0.8    Train Loss :0.0022996151819825172    Test Loss :0.009599702432751656\n",
      "Epoch :0.81    Train Loss :0.002280787331983447    Test Loss :0.009484261274337769\n",
      "Epoch :0.82    Train Loss :0.0022624770645052195    Test Loss :0.0093715600669384\n",
      "Epoch :0.83    Train Loss :0.0022446822840720415    Test Loss :0.009261670522391796\n",
      "Epoch :0.84    Train Loss :0.0022274029906839132    Test Loss :0.0091546094045043\n",
      "Epoch :0.85    Train Loss :0.002210636157542467    Test Loss :0.009050427004694939\n",
      "Epoch :0.86    Train Loss :0.0021943773608654737    Test Loss :0.008949150331318378\n",
      "Epoch :0.87    Train Loss :0.002178624039515853    Test Loss :0.008850811049342155\n",
      "Epoch :0.88    Train Loss :0.002163369208574295    Test Loss :0.008755429647862911\n",
      "Epoch :0.89    Train Loss :0.0021486070472747087    Test Loss :0.008663025684654713\n",
      "Epoch :0.9    Train Loss :0.0021343291737139225    Test Loss :0.008573608472943306\n",
      "Epoch :0.91    Train Loss :0.0021205279044806957    Test Loss :0.008487187325954437\n",
      "Epoch :0.92    Train Loss :0.0021071943920105696    Test Loss :0.008403755724430084\n",
      "Epoch :0.93    Train Loss :0.002094317227602005    Test Loss :0.008323314599692822\n",
      "Epoch :0.94    Train Loss :0.002081885701045394    Test Loss :0.008245846256613731\n",
      "Epoch :0.95    Train Loss :0.0020698895677924156    Test Loss :0.008171333000063896\n",
      "Epoch :0.96    Train Loss :0.0020583157893270254    Test Loss :0.008099754340946674\n",
      "Epoch :0.97    Train Loss :0.002047151792794466    Test Loss :0.00803107675164938\n",
      "Epoch :0.98    Train Loss :0.002036384306848049    Test Loss :0.0079652676358819\n",
      "Epoch :0.99    Train Loss :0.0020260033197700977    Test Loss :0.007902289740741253\n",
      "Epoch :1.0    Train Loss :0.0020159927662461996    Test Loss :0.00784209556877613\n"
     ]
    }
   ],
   "source": [
    "train_loss, test_loss = train_model(gru,\n",
    "               loss_fun,\n",
    "               optimizer,\n",
    "               train_x,\n",
    "               test_x,\n",
    "               train_y,\n",
    "               test_y,\n",
    "               epochs=500)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x262f845bec8>"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(train_loss,label='Train_loss')\n",
    "plt.plot(test_loss,label='Test Loss')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x262f84d7f88>"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# testing the prediction quality if the trained model\n",
    "# 1) Multi Step prediction\n",
    "#plt.plot(train_y.detach().numpy())\n",
    "plt.plot(np.vstack((train_y.detach().numpy(),test_y.detach().numpy())),label='Target')\n",
    "plt.plot(np.vstack((train_y.detach().numpy(),gru(test_x).detach().numpy())),label='Predition')\n",
    "plt.axvline(int(len(x)*0.7))\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x262f8554b88>"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# testing the predction model on multiple time series\n",
    "last_x = train_x[-1].view(10)\n",
    "\n",
    "prediction_val = []\n",
    "\n",
    "while len(prediction_val)<len(test_y):\n",
    "    prediction = gru(last_x.view(1,10,1))\n",
    "    prediction_val.append(prediction[0,0].item())\n",
    "    \n",
    "    \n",
    "    ## replace the predicted value in last x\n",
    "    last_x = torch.cat((last_x[1:],prediction[0]))\n",
    "    \n",
    "# plot the result\n",
    "plt.plot(np.vstack((train_y.detach().numpy(),test_y.detach().numpy())),label='Target')\n",
    "plt.plot(np.vstack((train_y.detach().numpy(),np.asarray(prediction_val).reshape(-1,1))),label='Predition')\n",
    "plt.axvline(int(len(x)*0.7),color='red')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Implement Autoregressive model\n",
    "auto_reg = nn.Linear(10,1)\n",
    "optimizer = torch.optim.Adam(auto_reg.parameters(), lr=0.01)\n",
    "loss_fun = nn.MSELoss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch :0.01    Train Loss :0.014670435339212418    Test Loss :0.10051491111516953\n",
      "Epoch :0.02    Train Loss :0.011478445492684841    Test Loss :0.05624254420399666\n",
      "Epoch :0.03    Train Loss :0.009668531827628613    Test Loss :0.03490261361002922\n",
      "Epoch :0.04    Train Loss :0.007584840524941683    Test Loss :0.03556336835026741\n",
      "Epoch :0.05    Train Loss :0.006658097263425589    Test Loss :0.03195313364267349\n",
      "Epoch :0.06    Train Loss :0.005737870000302792    Test Loss :0.02435385063290596\n",
      "Epoch :0.07    Train Loss :0.005292708054184914    Test Loss :0.022134913131594658\n",
      "Epoch :0.08    Train Loss :0.004909766372293234    Test Loss :0.020633360370993614\n",
      "Epoch :0.09    Train Loss :0.004579460714012384    Test Loss :0.01888328790664673\n",
      "Epoch :0.1    Train Loss :0.004303233232349157    Test Loss :0.017602767795324326\n",
      "Epoch :0.11    Train Loss :0.004018373787403107    Test Loss :0.016383396461606026\n",
      "Epoch :0.12    Train Loss :0.0037649564910680056    Test Loss :0.015334011986851692\n",
      "Epoch :0.13    Train Loss :0.0035474065225571394    Test Loss :0.014438631013035774\n",
      "Epoch :0.14    Train Loss :0.0033571061212569475    Test Loss :0.013807353563606739\n",
      "Epoch :0.15    Train Loss :0.003194790333509445    Test Loss :0.013070610351860523\n",
      "Epoch :0.16    Train Loss :0.003055501962080598    Test Loss :0.012486657127737999\n",
      "Epoch :0.17    Train Loss :0.002932385541498661    Test Loss :0.011994659900665283\n",
      "Epoch :0.18    Train Loss :0.0028222501277923584    Test Loss :0.011407768353819847\n",
      "Epoch :0.19    Train Loss :0.002723650773987174    Test Loss :0.010964534245431423\n",
      "Epoch :0.2    Train Loss :0.002634994452819228    Test Loss :0.010517152957618237\n",
      "Epoch :0.21    Train Loss :0.0025546750985085964    Test Loss :0.010101569816470146\n",
      "Epoch :0.22    Train Loss :0.002481344621628523    Test Loss :0.009760824963450432\n",
      "Epoch :0.23    Train Loss :0.002413895446807146    Test Loss :0.009418973699212074\n",
      "Epoch :0.24    Train Loss :0.0023514525964856148    Test Loss :0.009131825529038906\n",
      "Epoch :0.25    Train Loss :0.002293356228619814    Test Loss :0.008859822526574135\n",
      "Epoch :0.26    Train Loss :0.0022391185630112886    Test Loss :0.00860703457146883\n",
      "Epoch :0.27    Train Loss :0.0021883649751544    Test Loss :0.008378369733691216\n",
      "Epoch :0.28    Train Loss :0.002140794647857547    Test Loss :0.008154691196978092\n",
      "Epoch :0.29    Train Loss :0.0020961633417755365    Test Loss :0.00795168150216341\n",
      "Epoch :0.3    Train Loss :0.0020542754791677    Test Loss :0.00775483250617981\n",
      "Epoch :0.31    Train Loss :0.0020149671472609043    Test Loss :0.007572817616164684\n",
      "Epoch :0.32    Train Loss :0.0019780981820076704    Test Loss :0.007400935981422663\n",
      "Epoch :0.33    Train Loss :0.0019435406429693103    Test Loss :0.007240192033350468\n",
      "Epoch :0.34    Train Loss :0.0019111771835014224    Test Loss :0.007090956903994083\n",
      "Epoch :0.35    Train Loss :0.0018808941822499037    Test Loss :0.006950855255126953\n",
      "Epoch :0.36    Train Loss :0.0018525836057960987    Test Loss :0.0068213241174817085\n",
      "Epoch :0.37    Train Loss :0.001826139516197145    Test Loss :0.006699657998979092\n",
      "Epoch :0.38    Train Loss :0.0018014597008004785    Test Loss :0.006586931645870209\n",
      "Epoch :0.39    Train Loss :0.0017784418305382133    Test Loss :0.00648120790719986\n",
      "Epoch :0.4    Train Loss :0.0017569877672940493    Test Loss :0.006383026484400034\n",
      "Epoch :0.41    Train Loss :0.0017370020505040884    Test Loss :0.0062912278808653355\n",
      "Epoch :0.42    Train Loss :0.0017183920135721564    Test Loss :0.006205949001014233\n",
      "Epoch :0.43    Train Loss :0.001701066386885941    Test Loss :0.006126476917415857\n",
      "Epoch :0.44    Train Loss :0.0016849404200911522    Test Loss :0.00605269568040967\n",
      "Epoch :0.45    Train Loss :0.0016699316911399364    Test Loss :0.005984090268611908\n",
      "Epoch :0.46    Train Loss :0.0016559609211981297    Test Loss :0.005920420400798321\n",
      "Epoch :0.47    Train Loss :0.0016429527895525098    Test Loss :0.005861274432390928\n",
      "Epoch :0.48    Train Loss :0.0016308360500261188    Test Loss :0.005806376691907644\n",
      "Epoch :0.49    Train Loss :0.0016195442294701934    Test Loss :0.005755390506237745\n",
      "Epoch :0.5    Train Loss :0.0016090139979496598    Test Loss :0.005708063952624798\n",
      "Epoch :0.51    Train Loss :0.0015991852851584554    Test Loss :0.005664115305989981\n",
      "Epoch :0.52    Train Loss :0.0015900031430646777    Test Loss :0.005623320583254099\n",
      "Epoch :0.53    Train Loss :0.001581414951942861    Test Loss :0.005585431586951017\n",
      "Epoch :0.54    Train Loss :0.0015733725158497691    Test Loss :0.005550254601985216\n",
      "Epoch :0.55    Train Loss :0.0015658298507332802    Test Loss :0.005517567973583937\n",
      "Epoch :0.56    Train Loss :0.0015587466768920422    Test Loss :0.005487195216119289\n",
      "Epoch :0.57    Train Loss :0.0015520830638706684    Test Loss :0.005458946339786053\n",
      "Epoch :0.58    Train Loss :0.0015458030393347144    Test Loss :0.005432672332972288\n",
      "Epoch :0.59    Train Loss :0.001539874472655356    Test Loss :0.005408198572695255\n",
      "Epoch :0.6    Train Loss :0.0015342665137723088    Test Loss :0.0053854030556976795\n",
      "Epoch :0.61    Train Loss :0.001528952387161553    Test Loss :0.005364134907722473\n",
      "Epoch :0.62    Train Loss :0.0015239050844684243    Test Loss :0.005344290751963854\n",
      "Epoch :0.63    Train Loss :0.001519102486781776    Test Loss :0.005325742065906525\n",
      "Epoch :0.64    Train Loss :0.0015145224751904607    Test Loss :0.005308404564857483\n",
      "Epoch :0.65    Train Loss :0.001510145957581699    Test Loss :0.005292161367833614\n",
      "Epoch :0.66    Train Loss :0.0015059560537338257    Test Loss :0.005276935175061226\n",
      "Epoch :0.67    Train Loss :0.001501936698332429    Test Loss :0.005262645892798901\n",
      "Epoch :0.68    Train Loss :0.0014980725245550275    Test Loss :0.005249216221272945\n",
      "Epoch :0.69    Train Loss :0.0014943513087928295    Test Loss :0.005236577708274126\n",
      "Epoch :0.7    Train Loss :0.0014907611766830087    Test Loss :0.0052246698178350925\n",
      "Epoch :0.71    Train Loss :0.001487291301600635    Test Loss :0.005213433410972357\n",
      "Epoch :0.72    Train Loss :0.0014839322539046407    Test Loss :0.005202813073992729\n",
      "Epoch :0.73    Train Loss :0.0014806760009378195    Test Loss :0.0051927645690739155\n",
      "Epoch :0.74    Train Loss :0.001477514044381678    Test Loss :0.005183245521038771\n",
      "Epoch :0.75    Train Loss :0.0014744396321475506    Test Loss :0.005174209363758564\n",
      "Epoch :0.76    Train Loss :0.0014714474091306329    Test Loss :0.005165619309991598\n",
      "Epoch :0.77    Train Loss :0.001468530623242259    Test Loss :0.005157452542334795\n",
      "Epoch :0.78    Train Loss :0.0014656850835308433    Test Loss :0.005149661563336849\n",
      "Epoch :0.79    Train Loss :0.001462906482629478    Test Loss :0.005142232868820429\n",
      "Epoch :0.8    Train Loss :0.001460190280340612    Test Loss :0.005135130602866411\n",
      "Epoch :0.81    Train Loss :0.0014575335662811995    Test Loss :0.005128334276378155\n",
      "Epoch :0.82    Train Loss :0.0014549331972375512    Test Loss :0.0051218243315815926\n",
      "Epoch :0.83    Train Loss :0.0014523863792419434    Test Loss :0.0051155779510736465\n",
      "Epoch :0.84    Train Loss :0.001449889736250043    Test Loss :0.0051095737144351006\n",
      "Epoch :0.85    Train Loss :0.0014474423369392753    Test Loss :0.005103806033730507\n",
      "Epoch :0.86    Train Loss :0.0014450412709265947    Test Loss :0.005098246969282627\n",
      "Epoch :0.87    Train Loss :0.001442685374058783    Test Loss :0.005092887207865715\n",
      "Epoch :0.88    Train Loss :0.0014403729001060128    Test Loss :0.005087715573608875\n",
      "Epoch :0.89    Train Loss :0.001438101870007813    Test Loss :0.005082717165350914\n",
      "Epoch :0.9    Train Loss :0.0014358720509335399    Test Loss :0.005077884066849947\n",
      "Epoch :0.91    Train Loss :0.0014336809981614351    Test Loss :0.005073203705251217\n",
      "Epoch :0.92    Train Loss :0.0014315281296148896    Test Loss :0.005068665370345116\n",
      "Epoch :0.93    Train Loss :0.0014294128632172942    Test Loss :0.005064261611551046\n",
      "Epoch :0.94    Train Loss :0.0014273341512307525    Test Loss :0.0050599887035787106\n",
      "Epoch :0.95    Train Loss :0.0014252905966714025    Test Loss :0.005055831745266914\n",
      "Epoch :0.96    Train Loss :0.0014232821995392442    Test Loss :0.005051787942647934\n",
      "Epoch :0.97    Train Loss :0.0014213076792657375    Test Loss :0.0050478545017540455\n",
      "Epoch :0.98    Train Loss :0.0014193670358508825    Test Loss :0.005044016055762768\n",
      "Epoch :0.99    Train Loss :0.0014174585230648518    Test Loss :0.00504027446731925\n",
      "Epoch :1.0    Train Loss :0.001415582257322967    Test Loss :0.00503662321716547\n"
     ]
    }
   ],
   "source": [
    "train_loss, test_loss = train_model(auto_reg,\n",
    "               loss_fun,\n",
    "               optimizer,\n",
    "               train_x.reshape(-1,10),\n",
    "               test_x.reshape(-1,10),\n",
    "               train_y.reshape(-1,1),\n",
    "               test_y.reshape(-1,1),\n",
    "               epochs=500)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot the loss\n",
    "plt.plot(train_loss,label='Train Loss')\n",
    "plt.plot(test_loss, label='Test Loss')\n",
    "plt.legend();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x262f87b9dc8>"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAD4CAYAAAD8Zh1EAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nOy9d3xc1Zn//z7T+0iaUe/uxk02xpgACUkggVBSCZDNpv4gJEt2k92wabshyyb722yym2y+6eUb0klIIWY3AQIhQABjjBu4yJbV+2ikGUkzmtGU8/3jjmTZGkkjzZVly+f9evHCc++55xwJ/LnPPOcpQkqJQqFQKM5/DEu9AYVCoVDogxJ0hUKhWCYoQVcoFIplghJ0hUKhWCYoQVcoFIplgmmpFvb7/bKurm6pllcoFGfQHIgAsKLYucQ7UczGiy++OCClLM52b8kEva6ujr179y7V8gqF4gxu+fZzAPziA5ct8U4UsyGEaJvpnnK5KBQKxTJBCbpCoVAsE5SgKxQKxTJhyXzo2UgkEnR2dhKLxZZ6K+clNpuNqqoqzGbzUm9FoVAsAeeUoHd2duJ2u6mrq0MIsdTbOa+QUhIMBuns7KS+vn6pt6NQKJaAOV0uQoj/K4ToF0K8PMN9IYT4qhCiSQhxSAixbaGbicVi+Hw+JeYLQAiBz+dT324UiguYXHzo9wHXznL/OmB15p87gG/msyEl5gtH/e4UigubOQVdSvkUMDjLkDcCP5Iau4ECIUS5XhtUKBSK84Gx8RS/eKGdZCq9ZHvQw4deCXRM+dyZudZz5kAhxB1oVjw1NTU6LK0vwWCQ1772tQD09vZiNBopLtYSsvbs2YPFYtF9zX379tHf38+11872JUihUJzr/OlYPx//9Uu4rGau37w0Nq0eYYvZvudn7ZohpfyOlHK7lHL7hFCeS/h8Pg4cOMCBAwe48847+ehHPzr5ORcxT6VS815z3759PPzwwwvZrkKhOIcYjI4D8MCLHXOMXDz0EPROoHrK5yqgW4d5zyluvPFGLr74YjZs2MD3vvc9AJLJJAUFBfzTP/0TO3bsYM+ePezatYu1a9dy5ZVX8uEPf5g3velNAIyOjvKe97yHHTt2sHXrVh566CHGxsa49957+elPf0pDQwO/+tWvlvJHVCgUeRDOCPpTxwP0DS9NcIIeLpddwF1CiPuBS4GwlHKau2W+/MtDhznSPZz35qZyUYWHe27csKBnf/jDH1JUVEQ0GmX79u289a1vxe12Ew6H2bZtG5/73OeIRqOsWbOGZ555hpqaGt7+9rdPPn/vvfdy7bXXct999zE0NMSll17KoUOH+MxnPsPLL7/MV77yFb1+TIVCsQSExxIYBKQl/GZfFx+8auVZ30MuYYs/B54D1gohOoUQ7xdC3CmEuDMz5PdAM9AEfBf40KLtdgn58pe/zJYtW7jsssvo7Ozk5MmTAFgsFt785jcDcOTIEdauXUttbS1CCG677bbJ5x999FE+//nP09DQwKtf/WpisRjt7e1L8rMoFAr9CUUTlLhtXFJXyAMvdrAU/ZrntNCllLfNcV8Cf6PbjjIs1JJeDB577DGeeuopdu/ejd1u54orrpiM97bb7ZPhgrP9B5RS8uCDD7Jy5elv7aeeemrxNq5QKM4aobEEBQ4zb7u4io//+iX2d4TYVlN4VvegarnkQDgcpqioCLvdzuHDh3nhhReyjtuwYQONjY10dGhv51/84heT917/+tfz1a9+dfLz/v37AXC73YyMjCzuD6BQKBadcDSB127m+s0VADx9fOCs70EJeg5cf/31RKNRtmzZwr333sull16adZzD4eBrX/saV199NVdeeSUVFRV4vV4A7rnnHqLRKJs2bWLDhg189rOfBeA1r3kNBw8eZOvWrepQVKE4jwmNjVPgMOOymnBZTYTHEmd9D+dULZdziQnBBa3o1SOPPJJ1XCgUOu3z1VdfTWNjI1JKPvCBD7B9+3YAnE4n3/3ud6c9X1xcrBp9KBTLgPCYZqEDeGwmhmNnX9CVha4z3/zmN2loaOCiiy5ibGyM22+/fam3pFAozgKhaIICh5av4rGbGVYW+vnP3Xffzd13373U21AoFGeRWCJFPJmetNDdykJXKBSK85NQVBPvAseEy8XM8FjyrO9DCbpCoVDkycQBaIH9lMtlJK4sdIVCoTjvCGXS/k87FFUWukKhUJx/hMbOcLnYzYzEEqTTZzdbVAn6GRiNRhoaGti4cSM333wz0Wh0wXP9+c9/5oYbbgBg165d/Pu//zsADz74IEeOHJkc95nPfIbHHnssv40rFIolI5zxoU89FE1LiIyfXStdCfoZ2O12Dhw4wMsvv4zFYuFb3/rWafellKTT8y9gf9NNN/GJT3wCmC7o9957L1dffXV+G1coFEtGaExzuUxY6MViBCMphmNK0M8ZrrzySpqammhtbWX9+vV86EMfYtu2bXR0dPDoo49y2WWXsW3bNm6++WZGR0cBePjhh1m3bh1XXHEFv/nNbybnuu+++7jrrrt49tln2bVrF3fffTcNDQ2cPHmS97znPZNZoo8//jhbt25l06ZNvO997yMejwNQV1fHPffcw7Zt29i0aRPHjh07+78QhUKRlfBYAqNB4LKaIBbmhiffwHuNDzNylkMXz9049D98Anpf0nfOsk1w3b/nNDSZTPKHP/xhspNQY2MjP/jBD/jGN77BwMAAn/vc53jsscdwOp184Qtf4L/+67/4x3/8R26//Xb+9Kc/sWrVKm655ZZp877iFa/gpptu4oYbbuBtb3vbafdisRjvec97ePzxx1mzZg3vete7+OY3v8lHPvIRAPx+P/v27eMb3/gGX/rSlybrsisUiqUllKnjIoSAE3/EnIyw03DkrB+MKgv9DMbGxmhoaGD79u3U1NTw/ve/H4Da2lp27twJwO7duzly5AiXX345DQ0N/PCHP6StrY1jx45RX1/P6tWrEULwzne+c15rNzY2Ul9fz5o1awB497vffVo1xre85S0AXHzxxbS2turw0yoUCj0IjSUoyPjPOfY/AGwxNDOciX45W5y7FnqOlrTeTPjQz8TpdE7+WUrJNddcw89//vPTxhw4cGCylO5CmKt+stVqBbSD22Ty7IdEKRSK7ISjCbwOMyRicOKPpM0uihNhEqEOoOys7UNZ6Atg586dPPPMMzQ1NQEQjUY5fvw469ato6WlZbL5xZmCP8FMJXPXrVtHa2vr5Lw//vGPedWrXrVIP4VCodCL8ISF3vIUjI8Su+SDAFj7Dp7VfShBXwDFxcXcd9993HbbbWzevJmdO3dy7NgxbDYb3/nOd7j++uu54oorqK2tzfr8rbfeyhe/+EW2bt06Kf6gVXX8wQ9+wM0338ymTZswGAzceeedWedQKBTnDlrpXAscewgsbkyX30VCGvEM6nwOOAdiKdokAWzfvl2eWTb26NGjrF+/fkn2s1xQv0PFQrnl288B8IsPXLbEO9Gfxt4RRuNJLq5dnA5Cmz77CG/bWs49x98CdVfCzT/g8D0NOLw+6v/+cV3XEkK8KKXcnu2estAVCsWyJpZI8b77XuBTv1kcazmVlozGxnn10AMQCcB6LZmw0biastFjsIC8lYWiBF2hUCxrfvhsK12hMYYWKeJktPckv7TcyytbvworXwNr3wBAs2UN9vQoDDYvyrrZOOcEfalcQMsB9btTKE5nKDLO157QggwWqz658cl/4yLRxotbPw/v/A2Y7QB0OTKuz+59i7JuNs4pQbfZbASDQSVMC0BKSTAYxGazLfVWFIpzhq890UQknuTGLRXEEmniyZTua4hgE/vSqxle+3aYErYcdq0kjhW6zp6gn1Nx6FVVVXR2dhIIBJZ6K+clNpuNqqqqpd6GQnFOkE5LfrK7jTc1VNJQU8BDB7sZiSWxuoy6rmMeaadDbmXtRGJRBpfdxnHDCjb1TM9rWSzOKUE3m83U19cv9TYUCsUyIDyWIJ5Ms7HSi8emie3wWAK/y6rfIrFhLPEh2mQplzpOF3S3zUSrLGFT6OQMD+vPOeVyUSgUCr0YzByCFjkteOya7ap79cOhVgDaZcmp1P8MHruZzmQBcrT3rEW6KEFXKBTLkqGIJuiFTsukhT7RKk6/RVoBTdC9Zwq6zUxPugCRTkI0qO+6M6AEXaFQLEuCGUH3OS2TYju8SII+ZKnAZDxdTj12E30yk8g00q3vujOgBF2hUCxLTrPQJwRd79DFoVYiBjcGR8G0Wx6bmT5ZpH0Y6dV33Rk4pw5FFQqFQi8mfegOy+Q13euTD7XSZyyn2D39oNVtm2KhD58dC10JukKhWJYMjo5jNxuxW4xIKTEbxaJY6O2UU5wlcsZjNxPAi0QgzpKFrlwuCoViWTIYHafIqVnnQgg8NrO+PvR0CkLtNCf8lHiyCLrNTBITcWvRueVDF0JcK4RoFEI0CSE+keV+jRDiCSHEfiHEISHEG/TfqkKhUOTOUOSUoINmMesatjjcBekExxN+il3TM7QnQiUj1uKz5kOfU9CFEEbg68B1wEXAbUKIi84Y9k/AL6WUW4FbgW/ovVGFQqGYD4ORcQqnCrrNpK+FPiVkcSYLHWDYVAwjPfqtOwu5WOg7gCYpZbOUchy4H3jjGWMk4Mn82Qucne8XCoVCMQOD0XGKpmRvahb6Ygh6aVYfutVkwGI0EDL5YPjcEfRKoGPK587Mtal8FninEKIT+D3w4WwTCSHuEELsFULsVfVaFArFYjI4Ok6RMyO04S7WiQ59E4uGWkkLEz2yKKuFLoTAYzcRFD6IDkBy8RtG5yLo2boen1kO8TbgPillFfAG4MdCiGlzSym/I6XcLqXcXlxcPP/dKhQKRQ7EEiki4ymKnBkL/aG/5ZPtt/PmyAOgVzXXwRYi9nJSGLOGLUImFp1M6OLo4vvRcxH0TqB6yucqprtU3g/8EkBK+RxgA/x6bFChUCjmSyiqWeKFTotWR6VjD3Gjkw+nfwIPvEcfUR9qZdBSATBjwS+3XUv/B87KwWgugv4CsFoIUS+EsKAdeu46Y0w78FoAIcR6NEFXPhWFQrEkBCNxQEv7J3gC4sM8u+rv+XbyejjyIASb8l8k1EafoZQipwWzMbuUemwmOpITgr74fvQ5BV1KmQTuAh4BjqJFsxwWQtwrhLgpM+wfgNuFEAeBnwPvkapLhUKhmIGXu8K847u7icR1ztzMMBTJWOgOC3RqzegjxQ08m96oDRgbym+BdAqig/SmvZTM4G4BKPPYODri0D6chYPRnDJFpZS/RzvsnHrtM1P+fAS4XN+tKRSK5cpXHz/BsyeDtAYjbKjw6j7/RNq/z2WBI3vB6kH6VhOSmUbR0cH8FhgLAZK+hIPiopkFvarQwQOjVqTDjDgXLHSFQqHQk47BKI8d7QNgRO/65BkGRzWXy6SFXrkNj8PKEG5tQL4W+pj2QuiM22c8EAWoLrIDgqSz9NxwuSgUCoWe/Oi5VtIZh+yiCXo0gRBQYE5C32Go3I7HZiYkndqAsTwt9Ex98/aYjRL3zH18q4s0d0vUenaSi5SgKxSKs0YknuT+FzrYXKW5WXSvT55heGSUArsZY+8hkCmo2o7XbmIEBxJD/hZ6RtADKdfsFnqhJugho/+ciXJRKBQKXfjt/i5GYkk+/JrVAIzoXf0QoHMv/3ToGm62PAtd2oHohIUuMRA3e/L3oWeeD+Ge9VC0xG3FYjLQT+FZORRVgq5QKM4af24MUO93ctVaLbFwUVwuvYcwkeTjsf+GvT+AghpwFU82uYiZPLpZ6IPSPauFbjAIqgrsdCa9MD4C8ZH81p0DJegKheKs0Ts8Rq3PgdlowG426l+fHCDUQRIjbda1MHgSKrcDYDMbsZgMRIye/H3oY4OkDBaiWGe10AGqihycjGcieRa50YUSdIVCcdboDccp92qHiG6baXEs9HAHffj48cr/hLXXw5bbJm95bGZGhFsXCz1m8gJiVgsdoLrQzssRz+TeFhPVsUihUJwVxpNpBkbjlHoWV9BluJPOtA+7xwfX/uy0ex67iTAuiHblt0h0iFGjF7vZiMs6u4xWFzl4YqxAy58Pd+a37hwoC12hUJwV+oZjAJMWuu7lbDPIUDud0n9ac4sJPDYzQ9Kli4UeFpr/XIhs9QtPUV3ooI9CpDAoQVcoFMuDCUE/ZaHr3EEIIJVAjPTQKf1aUtEZeOxmgmmndkCZTznbaJBg2jWn/xy05KIURmK2UiXoCoViedATnrDQ7cCEy0VnC324GyHTdEs/Ra5sFrqJgWSmtkostPB1xgYJpJxz+s/hVCx62FIGocX1oStBVygUZ4UJC70s43JZQQ+rIweg8WEYj+izSObQsUv6tUqLZ+Cxm+mbEPSFxqKn08ixITridioK7HMOL3CYcVlN9BuKF/1QVAm6QqFgbDxFeCxBKr14RVJ7wjHsZiMemwme/CL/cPwdfDt9D/z8Ftj3I30Wybg0uqR/8sUxFY/NTM94RoQX6kePhRAyzVDaxTt31s45XAhBVaGdjnSRFraYTi1s3RxQgq5QXMAMBgPs+trHuORfH+ZY78ikFb0Y9IZjlHttiBOPwhOf53jxNdw2/mmk0aJfWnzGpREwFON3TneHeO1mAqn86rm0dWprXLSynnq/M6dnqoscnIwXQDoBo/0LWjcXlKArFBcwJ5/8GTcNfJcPr+zHKCCeTC/aWr3DMRqcQfj17VC2iRe2fI7n0huQtoL8o04mCLczbCzEV+DBYJgefeKxmwjh0j4scM2fPnEAgKsvXp/zM1WFdg5HMo0uFtHtogRdobiASQ6cBOD/Wz2KxWQkkVpEQQ/HeFf0h1qX4lt+gsOpCWvSqqOghzroE8WToZFn4nNaCcmMoC/Ah36sd5iTbW0AeHylOT9XXeigJZnpLaoEXaFQLAbW4VYAjH0vYTEJxhdJ0NNpSf9wlLVj+2HdjVBYi9uq1VZJmL06WugddKR9Mx5WlnltjGInLUwLWvNI9zCFYlT7YC/K+bkSj5Vu6cvscfFCF5WgKxQXMJ6xjLj0HMJsNJBILs6h6EAkzkrZgT05DHVXAFrYIkDM7M10AMoTKZHhTk6OF1E5g6BrlrvQKi4uwIfeOhChSGQKbDl8OT/nc1oZxUHS7FaCrlAo9Eem05Qlu5EIGDiOxQiJVJrFaAfcG46x03BE+1CndaucqH44ZtShtgpAZACRjNEp/TNa6H6XFaNBEDUurOJiSzBKjW0MDCawunN+zp+JiY/YK5SgKxQK/RkMdOMSY/QWbgOZwiHjSGB4TP/6KpqgH2XcXa2Vs+WUhR4xePJL8pkg3A5At/TN6EM3GgQlbivDuBfkQ28LRqi0xjTrfI6U/6n4XFrETdhSuqjJRUrQFYoLlEDbUQBGVtwAgE2OAdA/on/oYm84yqWGo6RqTvWSd9s0C31EuGB8NL9UfJgUyi7pn9HlApoffUg65+3mkVLSMhCh1BSZl7sFoMBuxiBg0FSiDkUVCoX+jPScAMC+9jVg82JORwHoH4nrvlaq5zCFYhTrqldNXnNbTQjBqTDCfK30jFB2Sj/lswm6x0Z/0jFvH/pgZJyRWJIiMTqvA1HQGl0UOS304Nd+zkVqdKEEXaG4QEkOnCQlBaW166BsM6akJuiBRRD0gv7nATDUXzF5zWAQuCymKY2b8/Sjh9qJGRxg885a0rbMa6M3YZ95vcQYPPBeePZrpwlva1ArT+BOh8ExP0EH7WC0Mz0R6ZJn+d4ZUIKuUFygmMOt9BlKsFhtULYZQyIKyEVxuVSG99FnLJv0n0/gtpm06oeQv6B3PE+rZTUVBY5Zh5V7bfQnnZCIQiLLzzpwHA7/Bh79NHx5Axz7XwBaBrQXnjWxQEF3WWhJZJ5bpINRJegKxQWKJ9pO0FKpfSjfDDKNQ4zrb6FLydrYIZqdDdP3YDcTmCiWlY+gRweh5xB72Dir/xy08r3h2bJFY2Ht36/7HFg98ML3AO1A1GSQGGND8/ahg3YweiI+kS3aPu/nc0EJukJxgVKS7CbqyljMZZsBKDJE9fehRwJ4GWHYu27aLbfNRCCRZ7EsgNa/AJI/xdfNWQGx3GvXmlxAdj/6xGFp/augajsMtgDQMhBhbYFEyNS8fegAPqeF41EnvPX7sOrqeT+fC0rQFYoLkHCwjwJGkUUrtAv+NWB2UJnu4Y7mv4Pel3RbK9bbCECiYOW0e26bmd5EJsQwH0FveQppdvLMWC3lBdlDFico89hmr+cycThrL4DCeu2wNZWkNRhhvTcT0rkQC91pIRxLE1//5mmuJ71Qgq5QXID0tR0DwFayWrtgNEFFAwFzJdWJZnjo73RbK9qtrSX8q6bd89hM9MYsIAx5C3q0fAcJTHO6XEo8c9RzmXC52AqgsA7SSWS4g9aBKGvdmW8vC/Kha7Hog5E8wzNnQQm6QnEBMtytWc2FVWtPXRQGIrYSHpE7YahVt7USgRPEpQmbv27aPbfNzHA8pYnnQgV9pBcGGun1XQowp8vFZjYi7JlCWdnWHAtpLxiLC4rqAQh3NzEaT7LO3KeNmfhmMw98mWzR4KgSdIVCoSOJwEnSUlBad7pf22w00JoohGhQC9/TAcNgE22ylCL3dKHV2tAlkfbChQt6y1MANDm2AsyYJToVkydTKTFbHfZYGGxeMBg0lwsQ6tJegLWpNjBaJ6/Ph4n0/+BSW+hCiGuFEI1CiCYhxCdmGPN2IcQRIcRhIcTP9N2mQqHQE1OolYAowmY/vUGDxWigR2bcCcPduqxlDTfTLCvwu6Y3nPDYzSTTknQ+FnrLk2Ar4Ei6BoM41YR6NvwFHoKiCEJt02/GQto3BgBPBRjMxPqbteeizVC8RnNRzZOiTMON4Kj+cf4TzCnoQggj8HXgOuAi4DYhxEVnjFkNfBK4XEq5AfjIIuxVoVDohDU+QNg0/WDPbDTQg45lXlNJnJEOWmQZRVl6fE7Uc0la8hD01meg7gpODIxRWWjHbJzbTi3z2uiQxTCURdDHQsRMbtJpCQYjFNbCUAsmg8A+1AglGxa0zXPF5bIDaJJSNkspx4H7gTeeMeZ24OtSyiEAKeXi9VhSKBR540wEiVqyCLpJnKrbPaxDNmOoDaNM0mGoxGExTrs9Uc9l3Lyw6odICcPdyKIV7G4e5JK63A4ryzw2WlJ+ZBYLfTg8wAu9aR46lPmGUliHdaSdbcUCMdINJbl3KpqK22rCYjQwEFlCCx2oBKZWk+nMXJvKGmCNEOIZIcRuIcS12SYSQtwhhNgrhNgbCAQWtmOFYhkj02me/9p7OfTnXy/qOt7UEOPW6YJuMRronXC56JGeHtQ6Ig3aahBZqhN6Mhb6mGmBNdHHI5CKM5ByMRgZZ+eK3MIJJy304S5IJSavSykZHhwgjJPnTga1a4X1+Ma7udqvfabkomxTzokQAp/LsuQWerYakWcWTDYBq4GrgNuA7wkhCqY9JOV3pJTbpZTbi4uL57tXhWLZMzTQw6UDv6H+ib+hq/nwoqyRTqUokMOkHNP/DpqMBsaFhaipUJ+qgEGtANioqy7r7QkLPWp0a4eR6dT85s8kBh0f0ea5bF6CXoKQ6dNcS880BbEmh4kKJy+0anMPWCrwiCiXmbTwy4Va6KC5XZY6bLETqJ7yuQo487SkE/idlDIhpWwBGtEEXqFQzIPeZi2hx0mMsZ/+NbGxiO5rhIK9mEQa4SqZdk+gFZEaMhfr43IJNjEiXJhc/qy3Jyz0UYMbkKdiwHMgmUpr0TjAoUEt/ry6aPY6LhOUe210yswLLeN2kVLylceO4xFR6qurOBmIMBgZpzGuvSRWDz2llQLwVuW8xzMpclqX9lAUeAFYLYSoF0JYgFuBXWeMeRB4NYAQwo/mgmnWc6MKxYXAaJdWo3zPurtZlTrJgfv+Xvc1wgOaPWb2Zm9yXOK20o9fH5fLwAnaqcDnyh55MtG1aNbaKlnoGIyy8bOPsO+YJjPP98FlK3PP3iz32jWXC0wejD57MshLbX1YSVBWqv1uXmwb4sVhzdlg6z+oWefzaGxxJn6nhYGldLlIKZPAXcAjwFHgl1LKw0KIe4UQN2WGPQIEhRBHgCeAu6WUwcXatEKxXEn3NxKTZi65+ePsd15BTf+fdF8jEtQE3VZQnvV+sdtKpyzSxUKXwZMcT5VNRniciTcj6EPpiQJdufnRT/SPEEuk2fWc9o2mI2bL2X8O4LSaGLeXk8IIIa1Q1l+aBvAbtW9EZSVlmI2CvW2D/Ll/itWfh7sFNJdLMBJflDZ/oPm+50RK+Xvg92dc+8yUP0vg7zP/KBSKBWIfbqbbWMkKk4mYdyXFo8+RSiYxmuYf9zwTsZCWTOPyVWS9X+ax0dJZCOlhiA2DzbOwheKjiJFumlJX4MsSsgha1qbfZaE9molRz9FCn6wIGQ2CGQalm50r5peOX+FzExz0U5JxubQFI6zxpiEKZlchmyq9PH18gMb+BKNOH65EcMEhixP4XFZiiTTR8RTOWWq2LxSVKapQnEP4Yu2EHLUAGAprMYsUgZ5WXddIDmvp6wXFZwaraZR6bTTFvdqHfKz0YBMAzbI8awz6BDVFDk6OapZ6roLeP6wJ+o5SSEuBp9BPVWFu/vOp606NRW8diLLKnSm+ZfNySV0RR3qGSaUlSa/23yRfC33i97BYB6NK0BWKc4R4LEp5upfxAq2Ilb24DoCh7pO6riNHA4xLE56C7C6KMo+N7rQOoYvHHwHgUHpF1izRCWqKHBwLZ6zVXC300Theu5lXVxsZFk4uX1M27+3V+hycTGix6FJK2oIR6l0Tgl7IxbWFk2PtJZlKkXkK+kT6/8AiHYwqQVcozhF6Wo5gFBJTqVYwy1uuiUikT9/4AtNYgCHhRRiy//Uv81rpmUwuWmC2aDoF+35EsPRyuiie3UL3OTk+nEk6mofLpdhtxZ4M4Sos5Z+un7/QVhc5aE8XI0b7GBgKExlPUWXPWM72gklBrylyYN30Jtj4NnBmj9bJFd9k+r+y0BWKZc1QuxZ37q3WEldKqjRBTwxmSU/PA0tsgGFj4Yz3yzx2+ihEIhZuoZ/8Ewx30lj5VoAZD0VBE8yENJK2uOcn6C4rRAcxufw4LPP3R9cWOSZDF/vatXA83B4AACAASURBVHj5MmumJZ3Ni89lZUuVlytW+2H9DfC27897jTOZTP9fpGxR/b3yCoViQcR7tIp+FSs3AWB3ugnixbhQK3kGnIlBIlnS/ico89pIYiJq9eNcqA/9xfvA4eeQ6xVA86Rlmo2aTOx43OzFPg+Xy5aqAggNLjguvMbnmAxdDHU3AR6KjZkKkzbtDOEXH7gMo2HhYYpn4ndZueOVK1hd6tZtzqkoC12hOEcwDp2gDx9O96kk66CpFFtUn6qHE3hSQ8StM7sOCh1mLCYDIVPJwgp0jfRC4x9g618xEJU4LEbsWeq4TFDr0wQ9apynhe62apmiC+geBFDqttFn1OLNY4FmjAaBR0S1OuhG7ZDWZjbmVOwrV2xmI596w3q21cz8DSkflKArFOcI3kgrAWv1addGbWUUxnt0WyOdSlEow6QcMwu6EIJSj5V+g39hUS4H7weZgm3vJhgZn9V/DlDssmI1GQjhgcgcdf2Ge4h2HCQ6nqLEZdHCFhfQPQjAYBDYCspJCDOE2qgssGOMhyet8/MRJegKxTmATKcpT3QQcZ/eCWfcWUlxOoBMp3VZZ3gogFmksqb9T6XMY6MrXaT50OebBNP7kta6zbeSYGR8svXaTBgMguoiBx2UwmBz9vWGe+B/Pwb/vRnbD1+HmSRljjQkYwsWdIBqn4teUYo90ql9UxibUgv9PEQJukJxDhDs7cAtxrRmzVMQhbXYRIJgvw5p+EA4oM0z2bFnBko9Nq1zUXJs/mVth7vBo/m1g6PxGZOKplJb5KAxUaLVcsnW5/PX74cXfwBlmzAkY9SIPsotGX/3Al0uoPnvT6TKKB9vpc7nPNWt6DxFCbpCcQ7Q1/IyAI6K01vCWTN9OAd1ikUfGcyk/RdmT/ufoMxj48RYJkN0vn704U7waklLgzm4XEALITwYyVjag1nCNAONsPWv4bovArBC9FBsHNXu2Rduodf4nOxP1lEnu1nlTWvdiuzKQlcoFHkQDbQAUFR1uoXuKdNcMCO9+gj6RNq/uyh72v8EZV4bbcmJVnRZvh0k49m7/aTTmnvEU4GUkuDo+KwhixPUFDloTGSKZQ2e8bOmEpqv3F0GPu33US968IkR7X6eFvpLcgUGIdkg2jIWuhJ0hUKRB8mQZgX7ympPu+6v0rJGE0F9YtGTYS3t3ztD2v8EZV7bqc5F2Sz0578F/2cbtD13+vVIP6QT4KlkNJ5kPJXOzeXic9AuS5HCMNkU49ScA4AEVwnYC4mYCllp6MWV1knQ01rD57rxxowPXblcFIpljV6HkjNhGOlhCDc2h+u0654CH8M4EXo0mwDkaD8JacRTOHuDmTKPjQG8pIUpu4U+1ArpJDzwbi1McYKJRCRv1WQ25Gwx6BPUFDlIYCJqr5juchnVXkK4NL9/n6WaNaY+DJnmFvkcitYUOQjipUv6KRw8COMjyuWiUCxndn/rQxz591ct6hrWsT4GjdlDCQPGEqwRfWLRjVEt7d9gnDkuHLRD0TQGxmwl2bNFR/vB4Yf4CDzwXkgl6ByKEh3IfJPwVBLMFKAqysHlMtGYImipnO5yGc2EMjq1yJxOUUEd3ZnDU5GXi8RuMVLstnLCuApj29PaReVyUSiWLwXB/WwYP0Tr0b2LtoYr3s+oJbvVPGwtx6tTLLo1Hpw17X+CUo/WkCJkLsluoY/2QdlGuP6/oP1ZaPw9f/W959n11B7tvreKp08EEAJWl7imP38GNrORUo+VDlEOwTNCFzMW+pERbU9N6TIKZQiGWjRr2phfwvtF5R4GvRtORfMol4tCsXwpSmguhZ6nf7R4a6QCxOzZKwaOOysoTvXr4vZxJAaJmud2UVhMBnxOCwGDP7sPfbRPs5g3vBmEkWTXQdqCUYb720kbbSQtXn7xQgdXri7OuaxtTZGDpmQJxE8PXUyPaIL+yUe1fx8dz8TQdzyfV4TLBF//q21c9/rrTl1QLheFYnkSj0XxS81yq+v+PenUPJsY57hGEcOkXdlDCWVBDS4xxnBofk3Ann/gS7Tcu5nQgPZCkuk03mSQuDX3RsrdaZ8WVz71ZSIljAa0Q0qzDXyriHVpnYMqRJAhUzFPHB+gJxzjHTtqct5vVaGDl2MZt9MUt8tQfyfD0sHB3jhN/aMcHMsI+lBrXgeiE7isJuw1F5+6oFwuCsXyJNDZjEFIXrY2UE6Axr2P6b5GsEdrgWYqyB55YvFpkS8DnSfmNa+xYzf16Tbavv/XpFMpdt/3cUoYJF21I6fnyzw2WhMFWtRKJHDqRnxESzjKHFJSugFj4AgA9ZYQJ2Jevvd0MyVuK69dP3tG6lSqCu3sj2QEekqkSzjQSUBqbpCf7G6jNV1MekK6dBB0bZ4iKMhEGCmXi0KxPAn1aMKS3PFBxqSF4T0/03+NvlYAbL7qrPfdpVrs9XDv/Oqi22N9xKSZLWN7eOlL13JZ+3d4oeA6drz1ozk9X+q1cWKyc9EUt8vEIeUUQbdHOnERZaU1RGe6iOdbBrnlkup5FbaqKrTTli7WQhenRLokwr2MmH1sry3kVy92Mo6ZMWemwmIeES7TqNiq/Vu5XBSK5Um0X0v4KVm5lSOeK1gbfIzxeEzfNYJaSKK7JLt7wlep1UWPD7TOa15vIsBhzxXsc72KLWN7eNnawJYP3jdjY4szKfPYOD6WEfSpkS6TYYQZ67tU67N5sa0L61g/0l2BEHDLJdlfUDNRWaCFLsadpyJdxpNprLEBjO5Srt9czmhc6yiULMjUvNFT0OuvBKtHF7/8UqEEXaGYhdRQG0lpoKSyHtbfRAGjtB7eresaySHN+i0qr896v9BfTlRaYR6x6DKdxp8eIOGsYM0dP2T3mo9RfeevsVhtOc9RUWCf0rkom6CfstABrrMfQcg0V1y8ha/c0jDvHp9VhXYAwrbqSZfLwc4QPkK4/ZVct7EckSlNLvxawpVuLheAi98Hf3dQOxc4T1GCrlDMgmmkkwHhw2S2ULJmOwChtkP6LjLcTUTacHuyhxMKg4GAsQTLaO4FusKD/dhEAjwVuDyF7HzHP+MtnF/7tDqfg0HcpIzW0yNdznS5eKsZxcEr0vsAKK9exRsbZs9EzUZ5gSakveYKGGwBKdl9TCtaVlZRS5nXxiW1mvVsK9Pa9OlqTRsM+lr8S4ASdIViFpzRbgYtWjhhRd16YtJMuveIrmtYor0Ejb5ZXSFhSxmeWO7JRcFuzQdtKVpYNx+AOr8TEIxaTo9FTw73ksbI+35xglRakkxLjqWrqYlnDm298xdzAKtJi0VvleVa6OJoH8eamoBTxcTuvGoFb2qowJLpu5pvj8/lhhJ0hWIWihJ9ROyamBhNJjpNNThDjbqu4Yz1M2yePRV/zFmBLzVH84cpjAS0yBmnP/ewwTPxOS24rSaCxuJJH/qhzhCP7nmJfunhT8eDHOoM0ROOcTQ9xV/uWZigA1QW2NmX0FxPsdY99Pdk3EyZbwOvWVfKV27dCnVXaElNq65e8FrLESXoCsUMJMbjFMsBUu5TYjXkWk1ZvEXXdQqSAcZs2ZOKJkh7qilkhOhoOKc545mD1sIZ/PK5IISgzu/UinRlLPR//NUhvKlBbIWaP/vpEwO0BaM0yszvyOzMK+yvqtDBM9FKMJgIHHuWokwOAGc25DAY4ZL3g2nuOjEXEkrQFYoZCHS1YBQSY9GpCoip4nUUMzSZrJMvqWQSnxwi6Zpd0M2ZWPRAR1NO86bDXSSlAV/p/CJNzqTW56A5XgAjPURjMY73jbDKEaWguIpNlV6ePhGgNRjhaDrzTcBbyeTJ5QKoLLTTGk4jSzdA516KReYF5pq9IYdCQwm6QjEDQ5mmEvbiuslrjspNAHSf2K/LGoP9nZhEGsMcfmdXiWZpT8TFz4VptIcBUYTRlF+dk3q/k8YxD8g0TSdPkpbgTQ+Cq4QrV/vZ1x7icPcwLcbMSy8PdwtokS7JtCRa3IB/+DDrnREQBuUrzxEl6ArFDET6tYPFwspVk9dKV2vJJyPt+kS6DPW2AmAtnP3wsiizh1iOsej2WC8h0+x++Vyo8znpklrkR1fbCQRprLEguEp55epiUmnJQwe7KSz0Qdkm7Z88mAh17HNvwi6jvMJ0VKvqaJi9OqRCI7/Xt0KxjEkNtpGWguJMYg9ASUU9wzigX59Il9GA5ut2zZBUNIG/rIZxaSQ9lFssuicRYMCxau6Bc1Dnd07Gog92N7PCUYNIp8BVytaaQpwWI6PxJJf6iuCvHgNDfpJSWaDFoj8ZqWEFUBd9GUo35vtjXDAoC12hmAEtBr3wtGQcYTDQZa7DMzy/uiozMT6YObwsq5t1nMFopN9QjGV07v6eMp3Gnxpg3Dl739BcqPc7aZVlJA02PIG97CjJFCdzlWAxGbhspSb2NT6HlpCTZynbieSiHx03EZYOxESnIkVOKEFXKGbAEe1i0Dz9sHLYs4bKROuCytkmxuOnXwh3Mi6NFPrnFt+QpQzX2Nyx6MOhIA4RB8/sfUNzodBhxmpzcMy1g0tiz7K1YEy7kWk28co1mluntmh+WaEzYTMb8bsstAzGeFms1i6qA9GcyUnQhRDXCiEahRBNQohPzDLubUIIKYTYrt8WFYqloTDRx6g9iyiWrMdDhEDP/Pp8dpw4iPx8OS98+e2EBwO88ODX2dbzC5ota+fsIAQQtVdQlOzLek+m0xzf9yTpVIrBHs33b57DL58LQgjq/U7uH2mgTAyxI5FpYJER2avXl1LhtbG9Tr8My8qMHz3gybhalIWeM3MKuhDCCHwduA64CLhNCHFRlnFu4G+B5/XepEJxtkkmxilOD5BwTY/acFVvBqD3xL55zdl39FksIsUl4UcQX93CJQc+xQnrBso+8Nucnk95qilmiHgsOu3e3gf/D2t23cSen/8rw/2ZpKLi2mnjFkKtz8musU0kpJGa9t9pFzMiW1Fg59lPvpaNlfqVnK3K+NFF1SWZtZSFniu5WOg7gCYpZbOUchy4H3hjlnH/CvwHoG8pOoViCejvbMYsUhh9K6bdq1yzDYBo5/wiXZL9J0hKA403/JYucy27i29mzcf+SIF/9hj0CUxF2sFpoPP0Mrq97SdYd/D/B6C26SfEAq0AFMzhl8+VOr+TYZw8LzZiSIyCyQ5Wty5zZ2PCj16y4Sqo2AY1OxdtreVGLoJeCUw9Wu/MXJtECLEVqJZS/s9sEwkh7hBC7BVC7A0EArMNVSiWlMFOLb3fUbZy2r0CfxlBvBgGjs9rTnPoJL2GUtZufw3rP/0cO//me5gtuWc6Ooq1WPSJ+HjQXC2Bn96BkTS7V32EcgL4G39KSoq8k4omqPdrLpCj3qu0C66SvJKH5mJrTSGlHiubVlXDHU9A5bZFW2u5kYugZ/svN9nBVQhhAL4M/MNcE0kpvyOl3C6l3F5cnH+MrEKxWER6tYxMf/W6rPd7LTV4RudXAqBgrI2gbeEiW1ihvVzGAqcs9H1/+L9siu/jpQ0fY/stn6aXYlamWgiKwnm9LGajzucEIFL/OkAsugvk2o1lPP+pq3FZVVT1fMlF0DuBqf8XVgFTj9rdwEbgz0KIVmAnsEsdjCoWi3gsyqEnfrUo/T0nSA+2Mi6NFFdkr4Uy6l5JRbI950iXdCpFebKbMc90F06ulFavIi7NpPtPFQdLNj9FGCeXvPUfMJkttK64DYCQSb/MyjWlbtaVubls83qtKXTtZbrNrdCXXAT9BWC1EKJeCGEBbgV2TdyUUoallH4pZZ2Usg7YDdwkpdy7KDtWXPC89OiP2Pzk+3nhV19atDWsI230GUpnTJ2X/jV4iBDsnzsuHKC/uwWHiCP8qxe8J6PJRIepBkf4VAy8Z/gEXeb6ySiZ9dffxZi0MDpHsa/54LSaePgjr+TSFT64+Qdwzb26za3QlzkFXUqZBO4CHgGOAr+UUh4WQtwrhLhpsTeoUJxJovcwAJuO/CddzUcXZQ3PWCdD1pnrkjgrtZC63qYDOc0XaNH27KxYm9e+hlwrKYtpLheZTlOZaGXEu2byvtdXyslrvo/vhn/Jax3F+UlOcehSyt9LKddIKVdKKT+fufYZKeWuLGOvUta5YjGxhZrop4gUBkL3374orpeSVC9jrpn93aUrtdDFSFduJQCiPce05+rzq3WS8q+nhEHCwT76u1vwEIWS9aeN2XjFTdSuUweJFyIqU1Rx3uGLtdHp3MDRhk+xYfwlXvyfb+s6f3gwgIcIsmDmOO7i8lpGpR0CuTW7kAMniEgb/rKFN5wAsFdpL4SuE/sn4+DdNZvzmlOxfFCCrjivGI/HqEj1EC9YySVvvIshPMjWv+i6Rn+b5saxlsxc3Eqr6VKDc6R5xjFTcYy00GOqnLXNXC6UrspUe2w7yFjnSwBUZipAKhRK0BXnFT3NhzGJNObSdZqoWldQMKJPoawJhnu0+bwVsx9gDrtWUBZvzWlOf7yDkKMuz51BaeUKRqQd+o9gHDhGP0V4fSqTUqGhBF1xXjHY/jIA3hrtUHLUu5aqRJuufvTxTJx3Sc2aWcelilbjJ0R4aCDr/QN//BnH9j5ObCxCWbqfRMHCQxYnEAYDXZZ63MNNFI420WvLf07F8kEJuuK8ItajuUMqVmq+ZGPZBhwiTneLftEuxlArQby4PIWzjrNVaCWNerJEuuz++b/R8MwHqXnoNl565D4MQmIunf0FkSthz2qqEi1UJduJehceBqlYfihBV5xXmAdP0Isfp7sAgIJ6zX8cOPmibms4op0MmOYuZ1u8QjuMHOk4fNr13T/7HDsbv8AB+06GhYdt+z+t7bV6Wk27hVGsVXu0iQSGsg36zKlYFihBV5xXeCOt9NtORZ9UrdlKWgpiXS/rtkbReDcjjrlLz5bVrCUuzaT6Tn076G0/wc7jX2S/43I2fHQXsVvuJ4rWIKOsXh/xnaj2CFBYt0WXORXLAyXoivOGdCpFZbKDqPuU39judNNlKMca1MflMh6PUZoOkPTMXXrWaDLRZazEPnyqWFZfk9Y82n7VRzFbrNSt307X9T/mufq75nTh5Er56gYA0lJQuVoJuuIUqvqNQjdkOs2er70bgWTH3/5E9/n7u5opE3FE8enZlgHHKoqjTbqs0dNyhFohMfiy13A5kyFnPeWjp5KLxnq1uPTS+lN9MNftuAZ2XKPL/gCKSioZoICYwUaVS7865IrzHyXoCt3Y88AXuXRwF8M4kel03jHXZxJoeYkywFV1uusi7ltP5ejTREfDOOYpcOPxGEef2cX4od9QGd5HdbofBHiq1s/9MDBeuIay4T8zFhnB7nQjBpsJ46RgkUMJm0qvRZps5N+TSLGcUIKu0IUTB55m65H/YAQ7HhGhv6eNksrcrNxcmUizL11xevq8tXIThnZJ5/EDrNn2qnnNeeI/r2HL+CGGcdDkuoQ235ux12xly8Wvyel5S9l6DO2SrqZDrNpyOY6RFvpMlXh1fpmdyc4P6psdq1geKB+6Im9SyST2393OoCig6RVfAObfni0XRPA4YZz4Sk4vmlW6WqtbEmrZP6/5RoeH2DB+iOf9b8H6iZNs+9guLnvvF2h47a05f7soqtNcK6FMfHxxvJNhhz6t3xSK+aIEXZE3Jw48SZXsoXPb3dRf/HoAopm0dD0pDB+l07JymtiW164jKq2k+w7P8GR2Oo9rLx3LmtditS2sa33Fig2kpCDRd4xYdJQyAiQKp3c5UijOBkrQFXkTOvC/pKRg9eVvpsBfRoBCjAPHdF0jMR6nNtHCSOH00D+D0UiHuQ53aH5rhlu0hKDS1QvvxWK1Oeg2lGMdaqKnVYu0Mc9SA0ahWEyUoCvyxtf7FMct6ydrivRa63Wvr9J+7EWsIoGp5uKs90Pe9VSPN+XcQQiAvsNEpI2ymvyyLQfsdfjGWgh1aD7+ghwPVBUKvVGCrsiLgd4OVidPEKo8dRgZKVhDVbKdVDKp2zrBE3sAKF2bvQO8qGjAQ5Su5tzqkwO4ho/TMaXbz0KJFayiItVNrFtbW68EIoVivihBV+RFy+7fAVC89cbJa8bSi7CLcXpa9XO7yO79DOOgsj57+rxv9aUA9DY+l9t86TRV482EPfnXQjGVrscsUhR2PUmAQt0SiBSK+aIEXZEXhpOPMUABKzedahzsrdMyGQPNubVny4Wi8GHarWtmtKZr1l1MXJpJduQW6RLoacNLBErzt6YLMpUf1ySOEbCoyHDF0qEEXbFgkolxVo/sobngFadFnlSt0QQ91qVPpMt4PEZtooXRLAeiE5gtVtrM9bgHc1uz57jWJdFdk3/qfMUqrbaKQUhGnSpkUbF0KEFf5oSHBtjz26/O77AwR47vfRwPEYxrXnfadYfLS5coxTKYW3u2uWg/theLSGKe4UB0gqGCjdTET8xYG32gt519j/wYgLGOQwBUrl14hMsETncBvfgBSBepkEXF0qEEfZlz7Kd3s+PgP9P88m7d544++11GpJ11V7552r1++0p8EX3qqwxmDkTL1r1i1nGGigbcYoyu5uyVFzt/dAfbnruLfQ/fh2ngCH348BYV67LHflsdAFadap4rFAtBCfoyprejia2BXQAE86gX3tV8mOFQ8LRr/V0tbBn+M4fL3jhZm3wqscI1VKa6iceiC153Atm9nzBOKurWzjrOt0aLgOlrfH7avdaje2mIPse4NFG3+5+pGHmJXrt+1nTUo81VVK1CFhVLhxL0ZUzbg/8KSOLSTLpnYf7spoPPUPmjV+D5ygr6P1vP7p/9KwAn//BVjKSpfv1Hsj5nqWzALFK0HXlhXuvtf/Qn7P3Pt5z2IvBlDkTnSsevWbuVmDST7JhediDw8BcZkxZOvO5HuGSECtlHtGDdvPY2G64tN3LYspnyFSpkUbF0KEFfpvS0NbJ14CH2+2+gzVyPK7wwf/bQSe3w8Lnyd9JvrWHn8S/x3Pc/xtrOX3HQeRmVK7JbpNVbrgJg8NhT81ov9fJv2T7yOPu//7cA7N31TValTjJadumcz5rMFtrMK/EMne5y6e1ooiH0Rw6V3MSGy69nX/2dAJgrN2WbZkFsvPxGNnzqaSxWm25zKhTzRVVbXKa0/+5z+IDaN/4z7Q/+C6sH/7ygkrbJwHHGpZFL3vdlhDCw96u3clnHdwHovuzOGZ8rqaynW5Rg6dozr/VcY92kpWBn4AGe+7aVi7t/zmHrFi5+x7/k9Hyo4CI2Bn5POpWaDHFs/Z8v4UdSff0/ArD9rz7Li39czabX3DKvvSkU5zrKQl+GdLc2si34v+z330hZzWpkyQYKGSHQ0zbvuWzhFnqMFZjMFowmE1v/9n6e972Jg/YdbLj8xlmf7XZvoSZyaF4RNr5ED/u8V3PctIbLen5Cj7Gcqjt/nbPla6i6GKeI0daouV1kOs2Kvkc45Lp80gdvMlu4+A3vXXBBLoXiXEUJ+jKk83f3IhHUvfkzALjrtEbKPY3z82cDFMXaGbTVTH42mkxc+uEfsuXjf5zT2k9VXYqfEN2tubWHi0VHKWaIROFKPO/6GXsKr8f817+eVyRK+carAOg//CQAPe0nKGGQRM0VOc+hUJyvKEFfZnS3HGPr4B/YX3wTpVVa5EXVuksAiHbML3MzmRinPNVNzLuwRhUlG16p7enQn3Ma39ehhTmaimopq1nNjr/7GRX18zu4rFxxEQMUYOh8PrP2nwDwX3TVvOZRKM5HlKAvMzp/9y+kMVCfsc4BPAU+zZ89kHvhKoC+jhNYRApT8cJiq2vXbWcYB7I9t/oq4W5N0F1lCw8nFAYD7c5NVA4fBCDVtpsRaaduff4JRArFuY4S9GXEeDxGw9AjHPBfP639W599Ff7I/EraDrRqDSPclQuLrTYYjbTYNlASyu2bwVigGQBfVX7JOeMVO6iQfQS6WykZ2k+LfQNGkzr/Vyx/chJ0IcS1QohGIUSTEOITWe7/vRDiiBDikBDicSGEKmixBHQ1HcQiUhhXTPcXx3wXUZXqIhYdzXm+iQ72JXmUgx0r20FduoNwsG/OsXKwlbg04y+rmXPsbBSt11w9Lc/+htpUO5FSZZ0rLgzmFHQhhBH4OnAdcBFwmxDizBqm+4HtUsrNwK+A/9B7o4q5CWaqG/rqG6bds1Vtxigk7cdyzxgVwSbCOCn0ly94T+412sulZf+f5hxrGe2g31Ccd33y+o2XMSYtlB35HgYhJ/egUCx3crHQdwBNUspmKeU4cD/wxqkDpJRPSCknUvt2A6qG6BKQ6DlMQhqpXDk9YaZ4pVbYaiJRKBdcIy30mqrnHbs+lRVbriQlBbHWuSNs3GPdDFkrFrzWBGaLlWbrOmrSXSSlgRVbXpn3nArF+UAuf1MrgY4pnzsz12bi/cAfst0QQtwhhNgrhNgbCARy36UiJ+yh43QZK7PGbFeuuIheirE2P5rzfMXjHQznWQ7W7nTTYazGHsxeMOu09VK9jDn1sQWGSzQ3S7N5FQ6XV5c5FYpznVwEXWS5JrMOFOKdwHbgi9nuSym/I6XcLqXcXlysT5U7xSlKxpoJOrM3KBYGA62lV3NR9AXCQwNzzhUZCVHCIKnC/BseB9zrqBg7ftq1gd52dn/zTgY+W8uR3Q8zEh6kgFGkNz//+QTOlZcDMFi0VZf5FIrzgVwEvROonvK5Cug+c5AQ4mrg08BNUsq4PttbXoSHBgj2dRIa6NV97shIiArZx7hv5oqERZfeikWkOP7kL+acrydTgtZSln852FTpZooZYqBby1R96anf4vzmxVzSez8uGSH2l2/Q364Jvtm/Iu/1AFZuv5rDlk0U7XyHLvMpFOcDuQj6C8BqIUS9EMIC3ArsmjpACLEV+DaamPfrv83zn4NPPID3v1fi++YGCr62lt0//zdd5+/MpLrbKjbOOGZ1wyvpoRhL4+/mnC/UqWV3FupQDtZTr7k/uo5pyT7J57/HqHDS/ddPc6DkTWwceYbBE9o9d5k+gu50F7DhU39hzbar0wN5BAAAESFJREFUdJlPoTgfmFPQpZRJ4C7gEeAo8Esp5WEhxL1CiJsyw74IuIAHhBAHhBC7ZpjugiV+8DcM4+D59Z/ipLGequM/mrGzzkIIt2kdeEpWzexiEAYDbWWvY310L+HB088wWo/u5bnvfoS2ezcyck8Zm/Z8krQUlOvQwb76Iq1SYrR9H+PxGGtG99JcdCXVqzZRfOX7sIgkZS99C4DiatUgQqFYKDmFL0gpfy+lXCOlXCml/Hzm2meklLsyf75aSlkqpWzI/HPT7DNeWMh0mhWhZznh2sGlt3ycoS0foEr2cOTZ/1nQfM9//f3s3fWt066l+44wJi2U186eKl+04xYsIkXjk/dPXutqPkzF/a9nR+d9jJh9HC69kf2lb2Hv5nuwOVwL2uNU3N4iOkQF1sDLHH/hUZwihmXd6wFYufkVnDSuoFp2E5E2Cnylea+nUFyoqEzRs8DJl57DT4j0qmsA2HjNuwjhYvz57897rmPPP8qlgV+xct/nGB0emrzuDB+n01w7Zwz36oYr6RYlWKe4XTqefxCLSNJ56+Ns/OST7PzQd9n5oe+w460fnff+ZqLftZayaCOjLz/MuDSy5rIbJu8FVr1NG2MsyytEUqG40FF/e84CA/v/F4D6nVr4vs3u5FjpDWwa+QsDvR2zPTqN2JP/RVRaKWSEl357KpioPN5CyDV3RIowGGgveS3rx/ZNtpWztz9Jh6igdv3sTZjzIVG8iQrZT13vIzTaNp/Wtm7t1e9jXBoJ2fKPQVcoLmSUoJ8FvF1PcMK4Cn/ZqWCh/9fenUdXWZ8JHP8+WchGIBuBQBKSQEgMhDXEBBc2rYgUWq0jTqt2cMrowcp0nMO4zHhm6oxnbHsE7bF1HFrF6mjHpVNstYCIw2DZCQE0LAESSMISIIRA4CYhv/njfROzXXITbnK353POPbn39y73ub9z73Pe/N73/T3DZz9KqFzl0JpXr7Fley11MYtTH6Q4Ip8bjq6irvYcNdUnSOA8V4e4dgIzZvLd1tUumz7AcaWezPrdVMUX9Phz9URUmnVidBhnuJQ6s92y2CFJ7Jnyb0TM+Ls+jUEpf6cJvY/Vnj3FmIYSziZNb9eeOmYiXw3IJanc9fPHLXUxs+c/QcTt/0gMFzm4chEVK+8HICrZtZJqY/JmW1PM7v+I0p0biBQHYVm3u/6heqHlxChAUt6CTsvz5j9Kdn7fxqCUv9OE7ibOrlgp3fIRwWKImTC307ILqbNJaz7G6cqj3e6/8khJa13M2CFJjJk8naLIaUyp28AQRwWbkxcx9ubOibIrQcHBHI6fTnbdVuqKf0+jCWb0jXe6tG1vxSQM4wRDqJKhpGaO79P3UipQaUJ3g82rnqbqX8fS2ND+fqr6i7VEFa2khmgyJ83otF3iRCvJl2279tUuzVevcv7dxTgYQNqCp1vbRy1+i8P3rGHoswcp/OvlPZrUKnL8t4gUB5NOfcihsBwGDop1edveqsxbxsmCf9ITn0r1Ef1luUFMxeckmxPs/eyd1rZLdecpe/kuMhv3czjv2S7n407PmcoZYgg+cu2ZCLe/9xPGNuzhq/FPtlYhAqtwxajcgl4lyKyCuVwgkjBppHZ4/8xGmDdvMZPveKBf3kupQBTQCb356lUaGxw0Njh6VMi4rcYGB2kNVuGI0KJVgFW6rezn88ly7KNo6k/Im7e4y20lKIijg/PJqNvudMjmeOleckuWsyd8KlO//XivYuzKgLBwDg6y5jtJ6GI4SCnlewI2oTuu1HPquSxCn08k9PlE9r0wu1dJvbxkBxHSwJGgNHIdu6g88iXb33mOsQ3F7Jz4nNNk3kJGzSaWOg7v/XOnZft3rCfirXk0SQhDv/cfbh+qiL3979ky5F5Gjb/JrftVSnlGwCb0g9vWkUQ12+K+yfbB3yDXsYt9m7qf46Sjswe+AKDhjp/SZII4/cEyphz+BbsG3srUBUu63T4t/y4AzhS3n3F458evk/7RfTgkjHP3rW431OIuo3ILKFiy8roLSiilvEPAJvRLX36Cw4QybtErjH90FSdJIHTjCz0+Sg+q3MlZBpM19Tb2RhUy6dImLkokaQ++6tIRdcKwFA4HZzCocmNr29WmJtK3PUt5aDpRSzZqgWOllEsCNqEnVf8fByLGEzlwMGHhkZTnPEJ2Uwl7N/6uR/tJrNvH8cgcJCiI0GmP0GyEsoLniEu8Vg2Q9k4n3swYx1etd24e2rWBOC5wYdJiYhKG9SgepVTgCsiEXnV0PyObK6hPndXaNmnBD62KPptcP0qvrTnDyOYKLidaNTzH3TyfuqWHmDzn+z2KJ37qdwiVq5SsewOAmt2raTTBZE77do/2o5QKbAGZ0I9vs8bKR+R/fSPOgLBwyrMfJqvpAOX23OLdObbHGiYZmFHY2jY4rueVmDIn3kJZUCoxB/4bgKSTn3MwbByDYxN6vC+lVOAKyIQeXraeCkkiZXT7W+WT861Zf0/t3eDSfi4e2UqzEUaOv77ruCUoiJMZd5PVtJ/iDe+R1nyMupG3Xdc+lVKBJ+AS+pXLlxhTX0RlQudL9Yan3cBp4gg+vtmlfUWeLuJYcDKDYuKvO67Rtz1Mkwki6X+XAZBSeM9171MpFVgCLqEf2raGCGkgPGdOp2USFMTx6Imk1hVdcxzdNDezedXT5NZv41RcvlviShiWyr6oG0nkHOVBKYzIuP5KQUqpwBJwCf3i/g1WgYX8zgkdoCmlkETOUVV2oMvlV+ovsnPFvRQefYVdg2YxYdHLbovNTLBmTawaOr2bNZVSqrOAS+jx1Vs4PCCbiKjoLpcPzbWufKks/rTTsjNV5Rx7cSZ5Fz5lS9oSpvzofbeUaGsxbuZCNo/4K9Lnuq9SkFIqcPhlQj9+qJijPx7P7nX/1a69rvYcGY2lnB96o5MtIXXMJGqIhvL2t+IfO7ib5tdmkNxYTtG0Vyj4/vNuvxU/dEAYhT9YwbCU7isPKaVUR36Z0KvWrCC9uZycTY9RtPat1vYjOz8lRJqJzp7pdNug4GDKIsczvLb9pYsn1r7EQHOJk/d+xKRvfK/PYldKqd7yu4Ref7GWnOpPKIq8iaOhmYz74nGKP7Ou77588HMaTAijJjlP6ACOETeSbE5SXVXW2pZQs5vD4TlkjHN+dK+UUp7kdwl935rXiZbLhE9fyvAffsLx4BSGb1xGXe054s9so/Qa4+ct4nOshF++ay1gDdWkNR3l4tCpfR6/Ukr1lt8l9JiStykLSiF76u1ED46jce5y4s15St54jIzGUmqHdV8MOSN3GucYBAf+BEDZ7s8JFsPA0f1TCEIppXrDrxJ6afEXjGk6yMnM+1tPWGblzWJH/Dzya/5IsBiis2d0u5/gkBBKY28h+8KfcVyp59KhTTSZINIn3trHn0AppXrPrxL6+fUvctkM4IY72heVGPOXP6OGaBwmlNGTZznZur2w3PkMlMvs3/xHBlbvpCwkvV/qbiqlVG/5TUIvLd5E3oVP2T1iYacJsmIShlEx82V2j11GeESUS/vLKvwmF00EDcUfkHGlhDNxk/sibKWUcpvOlYt9yLnTlcTED0NEuPLxM9QQzdi/+Ocu182dfneP9h0eEcXOQQVMrFlLqFwlNL2w+42UUsqDfDahH967hZHv30lZcCqnkmZS6NjNlqxlFLhhoqxW2fMI3W7NvJgywbWhGqWU8hSfHXI5t34FDYQSahoorHydCkli8t1PuPU9sm65hwYTQpUkkjgi3a37Vkopd/PJI/SzpyqYULOOoiHzyXvkP9m1/h3iUm9gQFi4W99n4KBYNic/gETGMtyte1ZKKfdzKaGLyBzgJSAYWGmM+fcOy8OAN4EpwFngPmNMmXtD/drBj39OoTQx7PalBIeEMPmOB/rqrSj8wYo+27dSSrlTt0MuIhIMvALcCeQA94tITofVHgZqjDGjgeXAC+4OtEVjg4PR5b9lT3geI7Mm9tXbKKWUz3FlDD0fKDXGHDHGNADvAgs6rLMAWGU/fx+YLSLivjC/VrzuTYZQg8n/m77YvVJK+SxXEvoI4Hib1xV2W5frGGOagFqg0+UmIrJYRHaIyI7q6upeBRwSHk1R5DRyp2uJNqWUasuVMfSujrRNL9bBGPMa8BpAXl5ep+WumDh7Icxe2JtNlVLKr7lyhF4BpLR5nQxUOVtHREKAwcA5dwSolFLKNa4k9O1Apoiki8gAYCGwusM6q4GH7OffAT4zxvTqCFwppVTvdDvkYoxpEpHHgDVYly3+2hjzpYj8GNhhjFkN/Ar4jYiUYh2Z65iIUkr1M5euQzfGfAx83KHt2TbPrwD3ujc0pZRSPeGzt/4rpZRqTxO6Ukr5CU3oSinlJzShK6WUnxBPXV0oItVAeS83TwDOuDGc/uKrcYPvxq5x9y+Nu++NNMYM6WqBxxL69RCRHcaYPE/H0VO+Gjf4buwad//SuD1Lh1yUUspPaEJXSik/4asJ/TVPB9BLvho3+G7sGnf/0rg9yCfH0JVSSnXmq0foSimlOtCErpRSfsLnErqIzBGRAyJSKiJPejoeZ0QkRUQ2iEiJiHwpIkvt9jgRWScih+y/sZ6OtSsiEiwiRSLyB/t1uohsteP+rT2VslcRkRgReV9E9tv9XugL/S0iP7K/I/tE5B0RCffW/haRX4vIaRHZ16atyz4Wy8v2b3WPiEz2srh/an9X9ojI70Qkps2yp+y4D4jIHZ6Juud8KqG7WLDaWzQBTxhjbgAKgCV2rE8C640xmcB6+7U3WgqUtHn9ArDcjrsGqzC4t3kJ+JMxJhuYgBW/V/e3iIwAHgfyjDHjsKaoXoj39vcbwJwObc76+E4g034sBn7ZTzF25Q06x70OGGeMGQ8cBJ4CsH+nC4Gx9ja/sHOP1/OphI5rBau9gjHmhDFml/28Diu5jKB9Qe1VwLc8E6FzIpIM3AWstF8LMAurADh4YdwiMgi4FWtufowxDcaY8/hAf2NNYx1hV/uKBE7gpf1tjNlI52pkzvp4AfCmsWwBYkQkqX8iba+ruI0xa+0ayABbsKqxgRX3u8YYhzHmKFCKlXu8nq8ldFcKVnsdEUkDJgFbgaHGmBNgJX0g0XORObUCWAY026/jgfNtvvze2O8ZQDXwuj1UtFJEovDy/jbGVAI/A45hJfJaYCfe399tOetjX/q9LgI+sZ/7Utzt+FpCd6kYtTcRkYHAB8DfGmMueDqe7ojIPOC0MWZn2+YuVvW2fg8BJgO/NMZMAi7hZcMrXbHHmxcA6cBwIAprqKIjb+tvV/jC9wYReQZriPTtlqYuVvO6uLviawndlYLVXkNEQrGS+dvGmA/t5lMt/3baf097Kj4nbgLmi0gZ1pDWLKwj9hh7SAC8s98rgApjzFb79ftYCd7b+/s24KgxptoY0wh8CEzD+/u7LWd97PW/VxF5CJgHfLdNHWSvj9sZX0vorhSs9gr2uPOvgBJjzIttFrUtqP0Q8Pv+ju1ajDFPGWOSjTFpWP37mTHmu8AGrALg4J1xnwSOi0iW3TQb+Aov72+soZYCEYm0vzMtcXt1f3fgrI9XAw/aV7sUALUtQzPeQETmAP8AzDfG1LdZtBpYKCJhIpKOdVJ3mydi7DFjjE89gLlYZ6QPA894Op5rxHkz1r9pe4Dd9mMu1nj0euCQ/TfO07Fe4zPMAP5gP8/A+lKXAu8BYZ6Or4t4JwI77D7/HyDWF/ob+BdgP7AP+A0Q5q39DbyDNdbfiHUk+7CzPsYaunjF/q3uxbqSx5viLsUaK2/5fb7aZv1n7LgPAHd6ut9dfeit/0op5Sd8bchFKaWUE5rQlVLKT2hCV0opP6EJXSml/IQmdKWU8hOa0JVSyk9oQldKKT/x/yYi/dp8KFUXAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# testing the prediction quality if the trained model\n",
    "# 1) Multi Step prediction\n",
    "#plt.plot(train_y.detach().numpy())\n",
    "plt.plot(np.vstack((train_y.detach().numpy(),test_y.detach().numpy())),label='Target')\n",
    "plt.plot(np.vstack((train_y.detach().numpy(),auto_reg(test_x.reshape(-1,10)).detach().numpy())),label='Predition')\n",
    "plt.axvline(int(len(x)*0.7))\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x262f8831c88>"
      ]
     },
     "execution_count": 66,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# testing the predction model on multiple time series\n",
    "last_x = train_x[-1].view(10)\n",
    "\n",
    "prediction_val = []\n",
    "\n",
    "while len(prediction_val)<len(test_y):\n",
    "    prediction = auto_reg(last_x.view(1,10))\n",
    "    prediction_val.append(prediction[0,0].item())\n",
    "    \n",
    "    \n",
    "    ## replace the predicted value in last x\n",
    "    last_x = torch.cat((last_x[1:],prediction[0]))\n",
    "    \n",
    "# plot the result\n",
    "plt.plot(np.vstack((train_y.detach().numpy(),test_y.detach().numpy())),label='Target')\n",
    "plt.plot(np.vstack((train_y.detach().numpy(),np.asarray(prediction_val).reshape(-1,1))),label='Predition')\n",
    "plt.axvline(int(len(x)*0.7),color='red')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}