codes/4_DP_LSTM.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "import time\n",
    "import math\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import datetime as dt\n",
    "from numpy import newaxis\n",
    "from tensorflow.keras.layers import Dense, Activation, Dropout, LSTM\n",
    "from tensorflow.keras.models import Sequential, load_model\n",
    "from tensorflow.keras.callbacks import EarlyStopping, ModelCheckpoint\n",
    "from sklearn.linear_model import LinearRegression\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "from sklearn.linear_model import Ridge"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from math import pi,sqrt,exp,pow,log\n",
    "from numpy.linalg import det, inv\n",
    "from abc import ABCMeta, abstractmethod\n",
    "from sklearn import cluster"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import statsmodels.api as sm \n",
    "import scipy.stats as scs\n",
    "import scipy.optimize as sco\n",
    "import scipy.interpolate as sci\n",
    "from scipy import stats"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv(\"new_date_union_1.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "wsj_var=np.var(df.mean_compound)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "mu=0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.02763231376845216\n"
     ]
    }
   ],
   "source": [
    "noise=0.1\n",
    "print(wsj_var)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "sigma_wsj=noise*wsj_var"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "n=df.shape[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_noise=pd.DataFrame()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_noise['noise']=df['mean_compound']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(0,n):\n",
    "    df_noise['noise'][i]+=np.random.normal(mu,sigma_wsj)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_noise.to_csv(\"new_price_noise0.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "dfn=pd.read_csv(\"new_price_noise0.csv\",index_col=0)  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_1n=pd.DataFrame()\n",
    "# df_1n['wsj']=dfn['noise']   #with noise\n",
    "df_1n['wsj']=df['mean_compound']  # no noise\n",
    "df_1n['price']=df['c']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "split = (0.85)\n",
    "sequence_length=10;\n",
    "normalise= True\n",
    "batch_size=100;\n",
    "input_dim=2\n",
    "input_timesteps=9\n",
    "neurons=50\n",
    "epochs=5\n",
    "prediction_len=1\n",
    "dense_output=1\n",
    "drop_out=0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "df1=df_1n\n",
    "i_split = int(len(df1) * split)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "cols = ['price','wsj']\n",
    "data_train_1 = df1.get(cols).values[:i_split]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1159, 2)"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_train_1.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "len_train  = len(data_train_1)\n",
    "len_train_windows = None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_windows = []\n",
    "for i in range(len_train - sequence_length):\n",
    "    data_windows.append(data_train_1[i:i+sequence_length])\n",
    "data_windows = np.array(data_windows).astype(float)\n",
    "  \n",
    "window_data=data_windows\n",
    "win_num=window_data.shape[0]\n",
    "col_num=window_data.shape[2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "normalised_data = []\n",
    "record_min=[]\n",
    "record_max=[]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "for win_i in range(0,win_num):\n",
    "    normalised_window = []\n",
    "    for col_i in range(0,1):#col_num):\n",
    "        temp_col=window_data[win_i,:,col_i]\n",
    "        temp_min=min(temp_col)\n",
    "        if col_i==0:\n",
    "            record_min.append(temp_min)#record min\n",
    "        temp_col=temp_col-temp_min\n",
    "        temp_max=max(temp_col)\n",
    "        if col_i==0:\n",
    "            record_max.append(temp_max)#record max\n",
    "        temp_col=temp_col/temp_max\n",
    "        normalised_window.append(temp_col)\n",
    "    for col_i in range(1,col_num):\n",
    "        temp_col=window_data[win_i,:,col_i]\n",
    "        normalised_window.append(temp_col)\n",
    "    normalised_window = np.array(normalised_window).T\n",
    "    normalised_data.append(normalised_window)\n",
    "normalised_data=np.array(normalised_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "x_train1.shape (1149, 9, 2)\n",
      "y_train1.shape (1149, 1)\n"
     ]
    }
   ],
   "source": [
    "data_windows=normalised_data\n",
    "x_train1 = data_windows[:, :-1]\n",
    "y_train1 = data_windows[:, -1,[0]]\n",
    "print('x_train1.shape',x_train1.shape)\n",
    "print('y_train1.shape',y_train1.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.        ],\n",
       "       [0.0729355 ],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.03185703],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.29166667],\n",
       "       [0.23643411],\n",
       "       [0.77529566],\n",
       "       [0.        ],\n",
       "       [0.07216495],\n",
       "       [0.33364574],\n",
       "       [0.42174321],\n",
       "       [0.06654171],\n",
       "       [0.47235239],\n",
       "       [0.62511715],\n",
       "       [0.5267104 ],\n",
       "       [0.23149016],\n",
       "       [0.32983508],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.06089744],\n",
       "       [0.03910256],\n",
       "       [0.30128205],\n",
       "       [0.53205128],\n",
       "       [0.65979381],\n",
       "       [0.65614035],\n",
       "       [0.83557951],\n",
       "       [0.91145833],\n",
       "       [0.80416667],\n",
       "       [0.9243604 ],\n",
       "       [1.        ],\n",
       "       [1.        ],\n",
       "       [1.        ],\n",
       "       [0.97930556],\n",
       "       [0.90178241],\n",
       "       [1.        ],\n",
       "       [1.        ],\n",
       "       [1.        ],\n",
       "       [1.        ],\n",
       "       [0.98196393],\n",
       "       [1.        ],\n",
       "       [1.        ],\n",
       "       [0.95619019],\n",
       "       [1.        ],\n",
       "       [1.        ],\n",
       "       [1.        ],\n",
       "       [1.        ],\n",
       "       [0.96302124]])"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train1[:50]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1149, 9, 2)\n"
     ]
    }
   ],
   "source": [
    "x_train_t=x_train1\n",
    "print(x_train_t.shape)\n",
    "x_train=x_train_t"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1149, 1)\n"
     ]
    }
   ],
   "source": [
    "y_train_t=y_train1\n",
    "print(y_train_t.shape)\n",
    "y_train=y_train_t"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.        ],\n",
       "       [0.0729355 ],\n",
       "       [0.        ],\n",
       "       ...,\n",
       "       [0.35896104],\n",
       "       [0.13766234],\n",
       "       [0.06629834]])"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataframe= pd.read_csv(\"new_date_union_1.csv\")\n",
    "dataframe.columns=['date','wsj','price']\n",
    "cols = ['price','wsj']\n",
    "len_dataframe=dataframe.shape[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "i_split = int(len(dataframe) * split)\n",
    "data_test  = dataframe.get(cols).values[i_split:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "len_test   = len(data_test)\n",
    "len_train_windows = None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "data_test.shape (205, 2)\n"
     ]
    }
   ],
   "source": [
    "print('data_test.shape',data_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_windows = []\n",
    "for i in range(len_test - sequence_length):\n",
    "    data_windows.append(data_test[i:i+sequence_length])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_windows = np.array(data_windows).astype(float)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "y_test_ori.shape (195, 1)\n"
     ]
    }
   ],
   "source": [
    "y_test_ori = data_windows[:, -1, [0]]\n",
    "print('y_test_ori.shape',y_test_ori.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[146.82  ],\n",
       "       [149.55  ],\n",
       "       [153.98  ],\n",
       "       [178.3   ],\n",
       "       [184.39  ],\n",
       "       [206.8   ],\n",
       "       [239.4   ],\n",
       "       [215.85  ],\n",
       "       [196.2   ],\n",
       "       [174.85  ],\n",
       "       [189.35  ],\n",
       "       [195.3   ],\n",
       "       [194.2   ],\n",
       "       [189.25  ],\n",
       "       [194.95  ],\n",
       "       [185.66  ],\n",
       "       [187.8   ],\n",
       "       [185.51  ],\n",
       "       [175.3   ],\n",
       "       [174.19  ],\n",
       "       [169.51  ],\n",
       "       [191.68  ],\n",
       "       [180.74  ],\n",
       "       [181.45  ],\n",
       "       [181.1   ],\n",
       "       [187.1   ],\n",
       "       [179.34  ],\n",
       "       [180.77  ],\n",
       "       [173.25  ],\n",
       "       [178.35  ],\n",
       "       [177.3   ],\n",
       "       [182.9   ],\n",
       "       [186.28  ],\n",
       "       [178.5   ],\n",
       "       [176.33  ],\n",
       "       [174.85  ],\n",
       "       [179.2   ],\n",
       "       [179.12  ],\n",
       "       [176.55  ],\n",
       "       [174.15  ],\n",
       "       [170.15  ],\n",
       "       [170.7   ],\n",
       "       [165.5   ],\n",
       "       [163.1   ],\n",
       "       [157.95  ],\n",
       "       [162.29  ],\n",
       "       [164.96  ],\n",
       "       [175.6   ],\n",
       "       [172.    ],\n",
       "       [170.07  ],\n",
       "       [178.4   ],\n",
       "       [180.39  ],\n",
       "       [175.79  ],\n",
       "       [175.2   ],\n",
       "       [178.09  ],\n",
       "       [172.43  ],\n",
       "       [169.95  ],\n",
       "       [169.51  ],\n",
       "       [169.6   ],\n",
       "       [167.79  ],\n",
       "       [178.5   ],\n",
       "       [174.03  ],\n",
       "       [172.15  ],\n",
       "       [175.02  ],\n",
       "       [175.85  ],\n",
       "       [172.11  ],\n",
       "       [172.23  ],\n",
       "       [174.91  ],\n",
       "       [168.08  ],\n",
       "       [170.85  ],\n",
       "       [159.94  ],\n",
       "       [161.48  ],\n",
       "       [158.63  ],\n",
       "       [160.    ],\n",
       "       [165.32  ],\n",
       "       [163.05  ],\n",
       "       [167.51  ],\n",
       "       [168.1   ],\n",
       "       [161.82  ],\n",
       "       [156.8   ],\n",
       "       [156.26  ],\n",
       "       [150.25  ],\n",
       "       [147.    ],\n",
       "       [158.1   ],\n",
       "       [166.5   ],\n",
       "       [164.3   ],\n",
       "       [169.1   ],\n",
       "       [166.51  ],\n",
       "       [166.8   ],\n",
       "       [170.94  ],\n",
       "       [160.    ],\n",
       "       [164.79  ],\n",
       "       [169.    ],\n",
       "       [167.69  ],\n",
       "       [167.35  ],\n",
       "       [162.39  ],\n",
       "       [163.5   ],\n",
       "       [164.44  ],\n",
       "       [166.72  ],\n",
       "       [168.15  ],\n",
       "       [167.32  ],\n",
       "       [163.61  ],\n",
       "       [169.    ],\n",
       "       [167.25  ],\n",
       "       [160.1   ],\n",
       "       [153.91  ],\n",
       "       [149.05  ],\n",
       "       [147.15  ],\n",
       "       [144.38  ],\n",
       "       [149.    ],\n",
       "       [154.7   ],\n",
       "       [151.7899],\n",
       "       [156.49  ],\n",
       "       [157.74  ],\n",
       "       [157.75  ],\n",
       "       [183.37  ],\n",
       "       [191.4999],\n",
       "       [181.    ],\n",
       "       [176.74  ],\n",
       "       [187.65  ],\n",
       "       [187.6   ],\n",
       "       [201.9   ],\n",
       "       [214.75  ],\n",
       "       [201.19  ],\n",
       "       [201.85  ],\n",
       "       [199.13  ],\n",
       "       [199.2   ],\n",
       "       [214.22  ],\n",
       "       [219.39  ],\n",
       "       [217.92  ],\n",
       "       [216.5   ],\n",
       "       [211.3   ],\n",
       "       [213.3   ],\n",
       "       [225.65  ],\n",
       "       [238.4   ],\n",
       "       [232.68  ],\n",
       "       [231.95  ],\n",
       "       [238.2   ],\n",
       "       [237.01  ],\n",
       "       [231.8   ],\n",
       "       [236.5   ],\n",
       "       [230.08  ],\n",
       "       [230.11  ],\n",
       "       [229.05  ],\n",
       "       [229.7   ],\n",
       "       [227.24  ],\n",
       "       [222.    ],\n",
       "       [219.9   ],\n",
       "       [220.67  ],\n",
       "       [218.99  ],\n",
       "       [218.95  ],\n",
       "       [219.87  ],\n",
       "       [216.75  ],\n",
       "       [216.4   ],\n",
       "       [216.1   ],\n",
       "       [216.71  ],\n",
       "       [213.9   ],\n",
       "       [214.02  ],\n",
       "       [202.84  ],\n",
       "       [211.    ],\n",
       "       [211.39  ],\n",
       "       [211.55  ],\n",
       "       [209.9   ],\n",
       "       [209.56  ],\n",
       "       [206.61  ],\n",
       "       [208.14  ],\n",
       "       [207.55  ],\n",
       "       [210.65  ],\n",
       "       [204.51  ],\n",
       "       [204.31  ],\n",
       "       [210.8   ],\n",
       "       [212.53  ],\n",
       "       [207.14  ],\n",
       "       [206.    ],\n",
       "       [206.5   ],\n",
       "       [203.4   ],\n",
       "       [202.5   ],\n",
       "       [192.18  ],\n",
       "       [196.25  ],\n",
       "       [194.15  ],\n",
       "       [194.1   ],\n",
       "       [196.7   ],\n",
       "       [201.89  ],\n",
       "       [207.5   ],\n",
       "       [210.66  ],\n",
       "       [207.55  ],\n",
       "       [207.8   ],\n",
       "       [211.81  ],\n",
       "       [214.89  ],\n",
       "       [211.6   ],\n",
       "       [210.68  ],\n",
       "       [210.9999],\n",
       "       [211.09  ],\n",
       "       [216.5001],\n",
       "       [215.75  ]])"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_test_ori"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "window_data=data_windows\n",
    "win_num=window_data.shape[0]\n",
    "col_num=window_data.shape[2]\n",
    "normalised_data = []\n",
    "record_min=[]\n",
    "record_max=[]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "for win_i in range(0,win_num):\n",
    "    normalised_window = []\n",
    "    for col_i in range(0,1):#col_num):\n",
    "        temp_col=window_data[win_i,:,col_i]\n",
    "        temp_min=min(temp_col)\n",
    "        if col_i==0:\n",
    "            record_min.append(temp_min)#record min\n",
    "        temp_col=temp_col-temp_min\n",
    "        temp_max=max(temp_col)\n",
    "        if col_i==0:\n",
    "            record_max.append(temp_max)#record max\n",
    "        temp_col=temp_col/temp_max\n",
    "        normalised_window.append(temp_col)\n",
    "    for col_i in range(1,col_num):\n",
    "        temp_col=window_data[win_i,:,col_i]\n",
    "        normalised_window.append(temp_col)\n",
    "    normalised_window = np.array(normalised_window).T\n",
    "    normalised_data.append(normalised_window)\n",
    "normalised_data=np.array(normalised_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "x_test.shape (195, 9, 2)\n",
      "y_test.shape (195, 1)\n"
     ]
    }
   ],
   "source": [
    "data_windows=normalised_data#get_test_data\n",
    "x_test = data_windows[:, :-1]\n",
    "y_test = data_windows[:, -1, [0]]\n",
    "print('x_test.shape',x_test.shape)\n",
    "print('y_test.shape',y_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.73089358],\n",
       "       [0.71709234],\n",
       "       [0.89115914],\n",
       "       [1.        ],\n",
       "       [1.        ],\n",
       "       [1.        ],\n",
       "       [1.        ],\n",
       "       [0.74562541],\n",
       "       [0.53337654],\n",
       "       [0.30276518],\n",
       "       [0.44296049],\n",
       "       [0.48372746],\n",
       "       [0.29976762],\n",
       "       [0.22308288],\n",
       "       [0.31138652],\n",
       "       [0.16746708],\n",
       "       [0.31585366],\n",
       "       [0.49929742],\n",
       "       [0.02200489],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.87146226],\n",
       "       [0.44143082],\n",
       "       [0.46933962],\n",
       "       [0.52277853],\n",
       "       [0.79341452],\n",
       "       [0.44339197],\n",
       "       [0.50789355],\n",
       "       [0.16869644],\n",
       "       [0.39873703],\n",
       "       [0.21975041],\n",
       "       [0.6967509 ],\n",
       "       [0.94079422],\n",
       "       [0.37906137],\n",
       "       [0.22238267],\n",
       "       [0.12279355],\n",
       "       [0.45663853],\n",
       "       [0.45049885],\n",
       "       [0.14873141],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.03409795],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.20423529],\n",
       "       [0.33112896],\n",
       "       [0.94892473],\n",
       "       [0.79603399],\n",
       "       [0.68668555],\n",
       "       [1.        ],\n",
       "       [1.        ],\n",
       "       [0.79500891],\n",
       "       [0.76871658],\n",
       "       [0.87292818],\n",
       "       [0.48412184],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.00827206],\n",
       "       [0.        ],\n",
       "       [0.85      ],\n",
       "       [0.58263305],\n",
       "       [0.40709617],\n",
       "       [0.67507003],\n",
       "       [0.75256769],\n",
       "       [0.40336134],\n",
       "       [0.41456583],\n",
       "       [0.66479925],\n",
       "       [0.0270775 ],\n",
       "       [0.26583493],\n",
       "       [0.        ],\n",
       "       [0.09679447],\n",
       "       [0.        ],\n",
       "       [0.07955865],\n",
       "       [0.41093366],\n",
       "       [0.27149877],\n",
       "       [0.54545455],\n",
       "       [0.77495908],\n",
       "       [0.26104746],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.52606635],\n",
       "       [0.92417062],\n",
       "       [0.81990521],\n",
       "       [1.        ],\n",
       "       [0.88280543],\n",
       "       [0.8959276 ],\n",
       "       [1.        ],\n",
       "       [0.54302423],\n",
       "       [0.74310777],\n",
       "       [0.84890966],\n",
       "       [0.70292505],\n",
       "       [0.67184644],\n",
       "       [0.21846435],\n",
       "       [0.31992687],\n",
       "       [0.40585009],\n",
       "       [0.6142596 ],\n",
       "       [0.90555556],\n",
       "       [0.74583964],\n",
       "       [0.18456884],\n",
       "       [1.        ],\n",
       "       [0.73524962],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.18765232],\n",
       "       [0.41917141],\n",
       "       [0.30097076],\n",
       "       [0.52951465],\n",
       "       [0.84987277],\n",
       "       [1.        ],\n",
       "       [1.        ],\n",
       "       [1.        ],\n",
       "       [0.77716634],\n",
       "       [0.65270742],\n",
       "       [0.90304961],\n",
       "       [0.90179048],\n",
       "       [1.        ],\n",
       "       [1.        ],\n",
       "       [0.76210526],\n",
       "       [0.66061563],\n",
       "       [0.58905551],\n",
       "       [0.59089713],\n",
       "       [0.9860563 ],\n",
       "       [1.        ],\n",
       "       [0.95375904],\n",
       "       [0.85735439],\n",
       "       [0.60069102],\n",
       "       [0.6994077 ],\n",
       "       [1.        ],\n",
       "       [1.        ],\n",
       "       [0.85408163],\n",
       "       [0.76199262],\n",
       "       [0.99261993],\n",
       "       [0.94870849],\n",
       "       [0.75645756],\n",
       "       [0.9298893 ],\n",
       "       [0.6685259 ],\n",
       "       [0.34980392],\n",
       "       [0.        ],\n",
       "       [0.07103825],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.04638554],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.08243728],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.10338983],\n",
       "       [0.        ],\n",
       "       [0.01772526],\n",
       "       [0.        ],\n",
       "       [0.47915443],\n",
       "       [0.5020552 ],\n",
       "       [0.62616822],\n",
       "       [0.50901226],\n",
       "       [0.48449892],\n",
       "       [0.27180966],\n",
       "       [0.47406082],\n",
       "       [0.42128801],\n",
       "       [0.89667049],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.89640884],\n",
       "       [1.        ],\n",
       "       [0.34428224],\n",
       "       [0.20559611],\n",
       "       [0.26642336],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.        ],\n",
       "       [0.2       ],\n",
       "       [0.0968059 ],\n",
       "       [0.09434889],\n",
       "       [0.30213904],\n",
       "       [0.67807263],\n",
       "       [1.        ],\n",
       "       [1.        ],\n",
       "       [0.83170996],\n",
       "       [0.8452381 ],\n",
       "       [1.        ],\n",
       "       [1.        ],\n",
       "       [0.84175084],\n",
       "       [0.76855415],\n",
       "       [0.70076154],\n",
       "       [0.48579161],\n",
       "       [1.        ],\n",
       "       [0.91619088]])"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_test"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "LSTM MODEL"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 1149 samples\n",
      "Epoch 1/20\n",
      "1149/1149 [==============================] - 7s 6ms/sample - loss: 0.3347\n",
      "Epoch 2/20\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.1958\n",
      "Epoch 3/20\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.1668\n",
      "Epoch 4/20\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1493\n",
      "Epoch 5/20\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1294\n",
      "Epoch 6/20\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.0900\n",
      "Epoch 7/20\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.0800\n",
      "Epoch 8/20\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.0780\n",
      "Epoch 9/20\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.0773\n",
      "Epoch 10/20\n",
      "1149/1149 [==============================] - 0s 41us/sample - loss: 0.0773\n",
      "Epoch 11/20\n",
      "1149/1149 [==============================] - 0s 41us/sample - loss: 0.0774\n",
      "Epoch 12/20\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.0795\n",
      "Epoch 13/20\n",
      "1149/1149 [==============================] - 0s 44us/sample - loss: 0.0765\n",
      "Epoch 14/20\n",
      "1149/1149 [==============================] - 0s 43us/sample - loss: 0.0776\n",
      "Epoch 15/20\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.0772\n",
      "Epoch 16/20\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.0769\n",
      "Epoch 17/20\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.0767\n",
      "Epoch 18/20\n",
      "1149/1149 [==============================] - 0s 41us/sample - loss: 0.0766\n",
      "Epoch 19/20\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.0773\n",
      "Epoch 20/20\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.0777\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x2145056b588>"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(LSTM(neurons, input_shape=(input_timesteps, input_dim), return_sequences = True))\n",
    "model.add(Dropout(drop_out))\n",
    "model.add(LSTM(neurons,return_sequences = True))\n",
    "model.add(LSTM(neurons,return_sequences =False))\n",
    "model.add(Dropout(drop_out))\n",
    "model.add(Dense(dense_output, activation='linear'))\n",
    "# Compile model\n",
    "model.compile(loss='mean_squared_error',\n",
    "                optimizer='adam')\n",
    "# Fit the model\n",
    "model.fit(x_train,y_train,epochs=20,batch_size=batch_size)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "ulti sequence predict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "data=x_test\n",
    "prediction_seqs = []\n",
    "window_size=sequence_length\n",
    "pre_win_num=int(len(data)/prediction_len)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(0,pre_win_num):\n",
    "    curr_frame = data[i*prediction_len]\n",
    "    predicted = []\n",
    "    for j in range(0,prediction_len):\n",
    "        temp=model.predict(curr_frame[newaxis,:,:])[0]\n",
    "        predicted.append(temp)\n",
    "        curr_frame = curr_frame[1:]\n",
    "        curr_frame = np.insert(curr_frame, [window_size-2], predicted[-1], axis=0)\n",
    "    prediction_seqs.append(predicted)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "e_predicted"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "de_predicted=[]\n",
    "len_pre_win=int(len(data)/prediction_len)\n",
    "len_pre=prediction_len"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[148.21374681689738, 154.39112012684345, 151.9748013883829, 174.80820093333722, 181.7469447493553, 204.06171693086625, 242.98825799226762, 240.71659790039064, 239.39096670985222, 229.2898611009121, 214.953343847394, 213.05001308679581, 221.14295799434186, 212.85405519902707, 207.27504017055034, 208.89299777448178, 195.74879482984542, 181.35077496618032, 181.93645785599946, 176.66291054323315, 171.31249653227627, 173.20097205400467, 176.15453933238982, 178.29545145988465, 179.75162319898604, 181.4682359457016, 183.58431716382503, 184.26221330583095, 183.2664710879326, 181.93745768368245, 181.6176846009493, 179.66481358110906, 179.046701413393, 179.1371929138899, 180.2262724906206, 181.7022611987591, 181.62494350850582, 182.16681972503662, 182.05275916159152, 181.4934972524643, 177.0144154369831, 178.03109902381897, 167.35285500735046, 165.3657696902752, 160.24415856301783, 160.31581290140747, 162.14830761492252, 161.86849349141121, 163.35554299801586, 166.52753286957739, 170.64862097203732, 173.95559441328047, 173.63629164934156, 171.7582923388481, 174.20267821788786, 173.78921155035496, 176.0377876138687, 176.09091255187988, 175.11057666778564, 171.47990458846093, 172.65746896505354, 172.9603243008256, 173.45881485521792, 174.44536762118338, 176.1155015784502, 176.33731069386005, 175.64603655695916, 174.63202129662037, 173.49307466804981, 171.95794961571696, 162.20989259064197, 162.54753595918416, 161.2736352354288, 162.1537868887186, 163.035151463747, 164.51342911243438, 166.55748648643493, 165.13227901935576, 166.57429329156875, 164.05695152282715, 162.58534381866454, 155.42982946038245, 150.64650518298149, 150.8211561501026, 152.19880856275557, 153.34050232470037, 158.80058152675628, 162.6694991648197, 162.44011368155478, 163.70568256258963, 162.3280522584915, 158.6297786951065, 164.8993089365959, 164.40816561162472, 164.34176222503186, 164.02452358961105, 164.55441034257413, 164.32024114787578, 164.52725501298903, 164.01228642463684, 165.24563472419976, 165.07384726345538, 166.0320721709728, 166.35920957326888, 164.2600160896778, 156.46973590284585, 148.82553239166737, 146.42595082819463, 144.49519991505892, 145.5205586911738, 147.09899833649396, 150.648738180995, 153.94450699806214, 151.1053195786476, 151.81111401438713, 178.74817230880262, 187.70388019067644, 188.5156565555036, 189.51936480175257, 188.12867399265767, 184.2426021676302, 194.78618486166002, 204.06446480870247, 196.80596715211868, 208.0945815849304, 206.0974997740984, 205.46285626411438, 200.5872082066536, 211.0959826606512, 206.1880503362417, 213.59106537818909, 212.26667716026304, 214.8206722640991, 219.01942620038986, 230.14295272111895, 225.5727017402649, 233.42096205949784, 231.6481002032757, 231.86798583865166, 231.89224097132683, 228.24732901453973, 224.65051835477354, 231.51575170606375, 232.25879208147526, 233.00515969991685, 230.6798912191391, 224.48498591780663, 222.5584086945653, 224.14684363007547, 223.95547754108907, 219.66327625989913, 220.68822129130362, 219.30231935828925, 221.52071232795714, 220.97317597568036, 217.39464737176894, 214.61837185658516, 215.10078529536725, 203.00326483574696, 204.82172285959126, 205.74492714881896, 204.18123367123306, 205.14684801220895, 206.3455392894149, 207.0595225942135, 205.80295439720155, 206.85889170229436, 205.44539887458086, 205.1972064781189, 205.34832894444466, 205.98891073584556, 207.90539405703544, 208.56062795042993, 209.22037232518196, 209.05389852285384, 208.15673025250436, 206.9097953081131, 196.23467832237483, 195.27724055528643, 194.51469501398503, 195.80896282225848, 193.16465372920038, 194.2286076593399, 197.68220482230186, 203.2755533838272, 206.76393259048461, 208.79812294006348, 210.89995501816273, 211.9016615438461, 208.16656809568403, 207.2428216576576, 209.3597641134262, 213.47683629512787, 214.36091120389702, 214.2470633645952]\n"
     ]
    }
   ],
   "source": [
    "m=0\n",
    "for i in range(0,len_pre_win):\n",
    "    for j in range(0,len_pre):\n",
    "        de_predicted.append(prediction_seqs[i][j][0]*record_max[m]+record_min[m])\n",
    "        m=m+1\n",
    "print(de_predicted)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.save('single_senti.npy',de_predicted)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In[18]:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "error = []\n",
    "diff=y_test.shape[0]-prediction_len*pre_win_num"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(y_test_ori.shape[0]-diff):\n",
    "    error.append(y_test_ori[i,] - de_predicted[i])\n",
    "    \n",
    "squaredError = []\n",
    "absError = []\n",
    "for val in error:\n",
    "    squaredError.append(val * val) \n",
    "    absError.append(abs(val))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "error_percent=[]\n",
    "for i in range(len(error)):\n",
    "    val=absError[i]/y_test_ori[i,]\n",
    "    val=abs(val)\n",
    "    error_percent.append(val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "mean_error_percent=sum(error_percent) / len(error_percent)\n",
    "accuracy=1-mean_error_percent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "MSE=sum(squaredError) / len(squaredError)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MSE [67.99860002]\n",
      "accuracy [0.97219993]\n",
      "mean_error_percent [0.02780007]\n"
     ]
    }
   ],
   "source": [
    "print(\"MSE\",MSE)\n",
    "print('accuracy',accuracy)\n",
    "print('mean_error_percent',mean_error_percent)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Plot the figure"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "195"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(de_predicted)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x214a2608648>"
      ]
     },
     "execution_count": 54,
     "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": [
    "import matplotlib.pyplot as plt\n",
    "plt.plot(de_predicted, label = 'predicted')\n",
    "plt.plot(y_test_ori, label = 'real')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Train only the stock data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1149, 9, 1)"
      ]
     },
     "execution_count": 101,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train_2 = x_train[:,:,1]\n",
    "x_train_2 = np.expand_dims(x_train_2, axis = 2)\n",
    "x_train_2.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 1149 samples\n",
      "Epoch 1/50\n",
      "1149/1149 [==============================] - 3s 2ms/sample - loss: 0.3489\n",
      "Epoch 2/50\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.1538\n",
      "Epoch 3/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1497\n",
      "Epoch 4/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1404\n",
      "Epoch 5/50\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.1393\n",
      "Epoch 6/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1363\n",
      "Epoch 7/50\n",
      "1149/1149 [==============================] - 0s 38us/sample - loss: 0.1367\n",
      "Epoch 8/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1363\n",
      "Epoch 9/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1374\n",
      "Epoch 10/50\n",
      "1149/1149 [==============================] - 0s 38us/sample - loss: 0.1378\n",
      "Epoch 11/50\n",
      "1149/1149 [==============================] - 0s 38us/sample - loss: 0.1381\n",
      "Epoch 12/50\n",
      "1149/1149 [==============================] - 0s 41us/sample - loss: 0.1379\n",
      "Epoch 13/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1358\n",
      "Epoch 14/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1355\n",
      "Epoch 15/50\n",
      "1149/1149 [==============================] - 0s 38us/sample - loss: 0.1353\n",
      "Epoch 16/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1355\n",
      "Epoch 17/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1356\n",
      "Epoch 18/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1355\n",
      "Epoch 19/50\n",
      "1149/1149 [==============================] - 0s 38us/sample - loss: 0.1357\n",
      "Epoch 20/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1366\n",
      "Epoch 21/50\n",
      "1149/1149 [==============================] - 0s 38us/sample - loss: 0.1343\n",
      "Epoch 22/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1351\n",
      "Epoch 23/50\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.1342\n",
      "Epoch 24/50\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.1347\n",
      "Epoch 25/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1340\n",
      "Epoch 26/50\n",
      "1149/1149 [==============================] - 0s 38us/sample - loss: 0.1344\n",
      "Epoch 27/50\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.1339\n",
      "Epoch 28/50\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.1367\n",
      "Epoch 29/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1357\n",
      "Epoch 30/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1343\n",
      "Epoch 31/50\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.1339\n",
      "Epoch 32/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1339\n",
      "Epoch 33/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1349\n",
      "Epoch 34/50\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.1334\n",
      "Epoch 35/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1337\n",
      "Epoch 36/50\n",
      "1149/1149 [==============================] - 0s 38us/sample - loss: 0.1346\n",
      "Epoch 37/50\n",
      "1149/1149 [==============================] - 0s 38us/sample - loss: 0.1341\n",
      "Epoch 38/50\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.1356\n",
      "Epoch 39/50\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.1330\n",
      "Epoch 40/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1334\n",
      "Epoch 41/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1336\n",
      "Epoch 42/50\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.1335\n",
      "Epoch 43/50\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.1331\n",
      "Epoch 44/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1332\n",
      "Epoch 45/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1330\n",
      "Epoch 46/50\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.1333\n",
      "Epoch 47/50\n",
      "1149/1149 [==============================] - 0s 40us/sample - loss: 0.1354\n",
      "Epoch 48/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1360\n",
      "Epoch 49/50\n",
      "1149/1149 [==============================] - 0s 39us/sample - loss: 0.1346\n",
      "Epoch 50/50\n",
      "1149/1149 [==============================] - 0s 41us/sample - loss: 0.1340\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x1d940af1d48>"
      ]
     },
     "execution_count": 102,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(LSTM(neurons, input_shape=(input_timesteps, 1), return_sequences = True))\n",
    "model.add(Dropout(drop_out))\n",
    "model.add(LSTM(neurons,return_sequences = True))\n",
    "model.add(LSTM(neurons,return_sequences =False))\n",
    "model.add(Dropout(drop_out))\n",
    "model.add(Dense(dense_output, activation='linear'))\n",
    "# Compile model\n",
    "model.compile(loss='mean_squared_error',\n",
    "                optimizer='adam')\n",
    "# Fit the model\n",
    "model.fit(x_train_2,y_train,epochs=50,batch_size=batch_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(195, 9, 1)"
      ]
     },
     "execution_count": 103,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_test_2 = x_test[:,:,1]\n",
    "x_test_2 = np.expand_dims(x_test_2, axis = 2)\n",
    "x_test_2.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [],
   "source": [
    "data=x_test_2\n",
    "prediction_seqs = []\n",
    "window_size=sequence_length\n",
    "pre_win_num=int(len(data)/prediction_len)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(0,pre_win_num):\n",
    "    curr_frame = data[i*prediction_len]\n",
    "    predicted = []\n",
    "    for j in range(0,prediction_len):\n",
    "        temp=model.predict(curr_frame[newaxis,:,:])[0]\n",
    "        predicted.append(temp)\n",
    "        curr_frame = curr_frame[1:]\n",
    "        curr_frame = np.insert(curr_frame, [window_size-2], predicted[-1], axis=0)\n",
    "    prediction_seqs.append(predicted)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "de_predicted=[]\n",
    "len_pre_win=int(len(data)/prediction_len)\n",
    "len_pre=prediction_len"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "m=0\n",
    "for i in range(0,len_pre_win):\n",
    "    for j in range(0,len_pre):\n",
    "        de_predicted.append(prediction_seqs[i][j][0]*record_max[m]+record_min[m])\n",
    "        m=m+1\n",
    "print(de_predicted)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MSE [89.94727498]\n",
      "accuracy [0.9628232]\n",
      "mean_error_percent [0.0371768]\n"
     ]
    }
   ],
   "source": [
    "error = []\n",
    "diff=y_test.shape[0]-prediction_len*pre_win_num\n",
    "\n",
    "for i in range(y_test_ori.shape[0]-diff):\n",
    "    error.append(y_test_ori[i,] - de_predicted[i])\n",
    "    \n",
    "squaredError = []\n",
    "absError = []\n",
    "for val in error:\n",
    "    squaredError.append(val * val) \n",
    "    absError.append(abs(val))\n",
    "    \n",
    "error_percent=[]\n",
    "for i in range(len(error)):\n",
    "    val=absError[i]/y_test_ori[i,]\n",
    "    val=abs(val)\n",
    "    error_percent.append(val)\n",
    "    \n",
    "mean_error_percent=sum(error_percent) / len(error_percent)\n",
    "accuracy=1-mean_error_percent\n",
    "\n",
    "MSE=sum(squaredError) / len(squaredError)\n",
    "\n",
    "print(\"MSE\",MSE)\n",
    "print('accuracy',accuracy)\n",
    "print('mean_error_percent',mean_error_percent)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x1d935986ec8>"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXsAAAD4CAYAAAANbUbJAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAABeoklEQVR4nO2dd3xb1d3/30eWbXnvldiJHWcvssggJIS9V2kLtOWhlJLS0gfohra/h9KntDylk7aUUkoZpYyWTRkh7ISEkEn2dmIn8d5Dssb5/XHulWRbsuWtSOf9evkl6d4r6fha/uh7P+d7vl8hpUSj0Wg0kY1ltAeg0Wg0muFHi71Go9FEAVrsNRqNJgrQYq/RaDRRgBZ7jUajiQKsoz0AgOzsbFlcXDzaw9BoNJqTik2bNtVKKXNCOTYsxL64uJiNGzeO9jA0Go3mpEIIcSTUY7WNo9FoNFGAFnuNRqOJArTYazQaTRSgxV6j0WiigD7FXghRJIR4VwixWwixUwhxW7f93xVCSCFEtt+2O4UQB4QQe4UQ5w/HwDUajUYTOqFk47iA70gpNwshUoBNQoi3pJS7hBBFwLnAUfNgIcR04BpgBjAGWC2EmCyldA/D+DUajUYTAn1G9lLKE1LKzcb9FmA3MNbY/Vvg+4B/6czLgaellA4p5WHgALBwSEet0Wg0mn7RL89eCFEMzAU+FkJcBhyTUm7rdthYoNzvcQW+Lwf/11ophNgohNhYU1PTv1H3l/INULl9eN9Do9FowpiQxV4IkQw8B9yOsnZ+BPxPoEMDbOtRNF9K+ZCUcoGUckFOTkgLwAbO6z+At/93eN9Do9FowpiQxF4IEYsS+iellM8DpUAJsE0IUQYUApuFEPmoSL7I7+mFwPGhHHS/cdmhs21Uh6DRhDX734LmE123uV2jMxbNsBBKNo4A/gbsllL+BkBKuV1KmSulLJZSFqMEfp6UshJ4GbhGCBEvhCgBJgEbhu03CAV3pxJ8jUbTE48HnroWPvqDb9uJT+EXY+Hgu6M3Ls2QEkpkvxS4DjhLCLHV+Lko2MFSyp3As8Au4A3gllHPxHF1gssxqkPQaMKWzhbwOKH+kG/biW0qQHr5VnC0jt7YNENGn6mXUso1BPbh/Y8p7vb4HuCeQY1sKHF3gqtjtEeh0YQn9iZ123DYt63xKCCgqRze/Tlc8PNRGZpm6IiOFbRuHdlrNEHxin2ZsnRAiX3qWJh2Cex5RW07thk2/HVUhqgZPFEi9k5w6sheowmIvVnduuzQWqnuN5VD+jjImwmN5eC0w/oH4I07wKPXR56MRInY68heowmKGdkD1BtWTuNRSC+CrImAVBZPzR7wuKClclSGqRkckS/2UmrPXqPpDX+xbyhTV8LNx1Rkn1Wqttfshdr96n5TeY+X0IQ/kS/2HjcgVUSi84Y1mp50EfvDSuilR4l9piH2B9/xpS83+sT+WGMHje2dIzhYzUCJfLF3+30Q3drK0Wh6YIp9aqGycRqNuobp48CWCsl5sPc13/FNan91i52Lfv8h5/32A7ZXNKEJb6JL7J16YZVG0wN7E8QmQfZEFdmbkXv6OHWbNRHajPpVsUne/Xe/vIsOp5vYGAuf+8tH7DyuBT+ciQKxd/ru61W0Gk1PHE1gS4OMYr/IXqhIH3y+fepY9YXQVM6H+2v4z/YT3Hb2JF645TRSbbF859ltOFw6UydciQKx97NutNhrND2xm2JfAh31cPBtSCkAa5zanzVR3eZMgbQiaCxn7YE6YmMENy2bQG6KjXuvmsWeyhb+8PaB0fs9NL0SBWLvZ+NosddoemJvUt78rM+qaL7iE5+FA35iP1VtbyrnQFUzJdlJxFmVhJw1NY/zpufx7MZypOxR5FYTBkSB2GsbR6PpFTOyTyuEr7wB2VOgcIFvf85UdZs3U0X2znZqqk8wKTely8usmJJLdYuDw7W6wmw4EkpbwpMbPUGr0fSOvQmyJ6v76UXwjfUg/MphZZXCjathzFzY9zoAsvEopXOndXmZxRMyAVh/qJ4JOckjMnRN6ERBZK9tHI2mV+zNKrI3sVi6ij1A0akQY1WRPVBALRNzuwp6SXYSuSnxrD9UN9wj1gyAKBB7beNoNEGR0mfjhILh5X/b+m+W7fpJl6ZAQggWT8hi/aE67duHIVEg9jqy12iC0tkG0h262CdksC/rHDqxkr73WdXyc8fz8PA50HSMxROytG8fpmjPXqOJZszVs6GKvRD8PutH7LA38f68D+HDX8OWJ9S+Hf/mtClfBeCdPdXatw8zoiCy1zaORhMUU+zjU4MeIqXk7d1VNLSpwOlAVSsTc5JhxZ0w7VJY8BXImwW7X6U4O4nZhWm8sOXYSIxe0w+iQOz9bRxdG0ej6UIIkf3W8kZufGwjl/1pDb96cy8HalqZkp8CMbFw9T/gkt/C9MtUfn5LJZ+ZO5adx5vZU9k8Qr+EJhQiX+xd/mKvyxxrNF1wGIJsSw96yAf7ahECOjo9/PHdA5w7LY+vLpvQ9aCplwAS9r7GZXPGYrUIfv7aHr7w1/X87NVd1LXqQGu0iS7PXkf2Gk1XQojsP9hfw+yxaTx8/akcrW9n/viMngflTlPlFva+TuaCr3Dm1Fze2lXF2PQE1h+q48Wtx/ng+ytIjIt8yQlXIj+y7zJBqyN7jaYLfYh9U4eTreWNLJ+cQ05KfGChB5WXP24xVO4A4J4rZvLEjQv58Ptn8odr51Hb6tBlkEeZKBB7/wlaHdlrNF2wN6pbW+AJ2o8O1OL2SJZPzun7tbJKoeU4OFrJTbWxbFIOFotgSWkWAFvKG4dmzJoBEQVib0T2sUnas9doumNvAmsCWOMD7v5gfy3J8VbmFKX3/VpmwbT6Q102ZybFUZyVyJajDYMcrGYwRI/Y21J1ZK/RdMeseBmEjw7WsnhCFrExIUiFKfZ1PcsczylKZ8vRRr2ydhSJArE3bJz4FJ1nr9F0p3tdHD+ON3ZwpK7da8P0SaaRoVN3sMeuueMyqG5xcKJJ/w+OFlEg9p0gYiA2Ua+g1Wi600tdHLOg2ZIJIYp9XJLqZhUgsp87Lh2ALUcbBzJKzRAQHWIfEwdWm47sNZru9CL26w7WkZ4Yy9T8lID7A5JVGlDsp+anEme18Nr2E7g92soZDaJH7GO12Gs0Pegtsj9cx6KSTCwWEXB/QLImQn1PGyfOauGGpcX8Z/sJVj6+kU6XZ6Aj1gyQKBH7WB3ZazSBCCD2NS0O/vnxUcrrO0K3cEyyJkJHA7TX99h154XT+PHF03h7TzWrd1cNZtSaARAlYh+nUsu0Z6/R+AhSy/7Lf9/AD1/YTnpiLGdNzevfa/aSkQPw5dOKyUyK4/UdlTjdHh77qIymDmfAYzVDSxSIvdOI7BN0ZK/R+OOyg8fZpeKllJLDtW1cu7CITT8+l3FZif17TbNfbfmGgLutMRbOm57HO7ur+Nuaw9z18k7+/F5P20cz9PQp9kKIIiHEu0KI3UKInUKI24zt9wkh9gghPhVCvCCESPd7zp1CiANCiL1CiPOHcfx94x/Z6zx7jcZHgFIJLQ4X7Z1uSrKTiOmPV2+SMR4K5sCnzwQ95IKZ+bR1uvnlG3sAePLjI7Q6XP1/L02/CCWydwHfkVJOAxYDtwghpgNvATOllLOBfcCdAMa+a4AZwAXAA0KImOEYfEi4nUroYxP0ClqNxp8AYl9l5MHnpyUM/HXnfAEqP4Vjm+Gdn0HVzi67TyvNJsVmRQK/+MwsWuwunt5wdODvpwmJPsVeSnlCSrnZuN8C7AbGSilXSSnNr+P1QKFx/3LgaSmlQ0p5GDgALBz6oYeId4JWR/YaTRe8Yp/u3WQuespPtQ38dWdeBRYrPH45fHAfvPb9LrvjYgS3nT2Jb58zmWsXjmNhSSaPrSvTq2uHmX559kKIYmAu8HG3XV8BXjfujwXK/fZVGNu6v9ZKIcRGIcTGmpqa/gyjf3htHMOz1x8oTbThcQfeHiCyrzTEviBtEGKflA2TL1C18kvPgiNroGKj2rftabh3HF+dIfjvsycB8Jm5Yymv72BvVcvA31PTJyGLvRAiGXgOuF1K2ey3/Ucoq+dJc1OAp/dQWCnlQ1LKBVLKBTk5IVTUGyhup8+zBx3da6KLfW/C/5UELGEQUOybldjnpgYujBYyl/4evvo2fP4JdeXw3i9UY/KXvqm+BA6s9h565tRcAN7eXT2499T0SkhiL4SIRQn9k1LK5/22Xw9cAnxR+q7BKoAiv6cXAseHZrgDwLRxYg0PUvv2mmji6DpwNMG79/TcF0DsTzTZyUqKI946yGm2pGwoXADxybBwpRL3f9+gJnCT86FsjffQvFQbM8em8u4eLfbDSSjZOAL4G7BbSvkbv+0XAD8ALpNStvs95WXgGiFEvBCiBJgEBM7DGglcDh3Za6IXM999x3NwfEvXfV6x96VeVjXbyRuMXx+IM74PN7wOX3gWbnwLJpwBR9Z2sVTPmprH5qMN3qbmmqEnlMh+KXAdcJYQYqvxcxHwRyAFeMvY9iCAlHIn8CywC3gDuEVKGcQ0HAG8efbGB1jn2muiibqDMP50SMiE1Xd33Wdv8tWNMjjRZB+cXx+ImFgYfxpMPh8SM6H4dGirgdp93kPOmpqLR8LbOrofNvpsCCmlXENgH/61Xp5zDxDgunEU8C+EBnoVrSZ68LiV2C86G6ZeBG/+EA6+oyZNwbd6Vvj+vaua7d4KlcNG8enqtuxDyJkCwOyxaUzKTeb+t/dzyewCbLGjl60dqUTJCto4Hdlroo+mCnA7IGsSnPpVSBsHq38CHqMIWbdSCXanm/q2TgqG2sbpTkYJpIyBLU9C5XYALJ3N/K3odWIb9vPTV3fxoxe28+H+YczSG23cLmgoG9G3jAKx96t6CVrsNdGD6ddnTVRzVivugBPboHy92t5N7Kub1XxW3lDbON0RApZ9G6p3w4Onw9/Oh4fOZNzOP3N39jv88+OjPPnxUZ7eUN73a52sbPsn/P4U+OThEXvL6BF7r40TPBvH6dZlVzURhL/Yg5oYBajZq24dzd0ycdT/xpB79oFYeBN8execdw+0VqnEiYJTOI1tPPCFuSwszqSiMYIz58xVxf/5Dqz704i8ZRSIvWHjeFMvA0f27+6tZu5P3+JQTesIDk6jGUbqDqgiZ8kqj52UMWpxofklYET2Ukou+cOH3PyPTcAgV8/2h8RMOO2bcOsW+PZOmH8DlpbjXJTfTEl2EscjWezrD0HONJh+uWoNOQJEgdibefZG9T5ne8DDdp9optXh4ner94/g4DSaQbD1KdgTNE8CaverzlHmBKzF0rWTlL0J4lOpb+tkx7FmspLjWTYpu/+VLgeLOb6JZ6vbg28zJj2BmhYHDtfoJfINitr98LtZgRezgdqeMxk++3dlr40AkS32Uvp59kZkH8TGMf3KVz49zp7Kkfmm1WgGxdrfwSd/Db6/7qDPwjHpLva2NMob1P/EDy6YyhM3Lhr8gqqBkj5OTSbve5Np1mPE08mJxpN0jm33K9B4FPav6rnP7YLGI5BZCpaYLtlQw0lki73HDUhD7M3IPrDYVzWr/OLkeCt3Pr+dNl1yVRPu2JuCWwBSQstxSCvsuj1rosoCcbQqS9OWRnm9utotyhxEpcuhYuLZcPh9znvvcr5vfebktXLMFcJHPuq5r/EIeFzqi3cEiWyxdxurZf3LJQSxcaqa7UzISeL/rprNtvJGbnj0E+zOk/QSUhMd2JvBEaR4mMuuBCW+W7PwrElq+4lt6rEtjfIGQ+wzRti+CcRpt8LZd+FKKmCMqOVYYwe/enMvC+9ZzZyfrmJjWc92h2GH2wlHjYyno+t7Fl+sP6xuM7XYDx1uY+m1WfUSeonsHeSl2LhoVgG//vwpbDhcr2tsa8IXtxOcbSqjJhAOI9HArwsV4LN1NvxF3ebNpLy+g8ykOJLi+1xjOfykjYVl38aSWUy6aKOioYN/fHyEzKQ4LEJw/zuB2x2GTFvt8C+sPL5F/W1Kz4K2ajUZ64/RkL09eRztnS5cI5QFGOFib/S2jIlVk1PWhICRvZSSmhYHuUYWwpVzC5k/PoO/rT2M26NLImvCEDOiD2bjdBr745K7bjetg10vKeEft5jy+naKMsMgqvfDkphJlqWd9/bV0Nju5MbTS7jx9BI+2FfD7hMDnFNztMCfFqoKnMPJ4Q/U7bLvqtuj67rurztIZ0wiM361hen/8ya3PbN1eMdjEOFi7xfZg7JyOnuKfWO7k063hzy/sq43LSuhvL6DN3dWjsRINZr+YW9Ut862wPXqvZF9N7FPzFR1cgDmXgdCUN7QTlFGGPj1/iRkkGFpY1t5IwCLJ2TxpYXjKI1r4DfPf8DDHx7ihS0V7DjWFPprbvkHtNd5V+0OG2VrIHcGjFsCCRk9xP7Y4Z3sc+Zy/vQC7rxwKpefMmZ4x2MQBtdtw4gp9mbFy9jEgDZOVYu6rPOv9nfu9HzGZyXyxLojXDSrYNiHqtH0C/+I3tGsRMUfR5DIHlREf2wTnHItbo/keGNH+H3GE9JJlep3KE2VFO78C+LjB3nbUklNdRrL/vM77Kj/668tn8D3zp+CNaaX2NXtgvUPqPv1w9jgXErVjnHmZ5SbULIcdr4Ei2+BnCk4qvYiq/fQljSdP35hbu9jHmIiXOz9bBxQkX0AG6fKXCbuF9nHWATzxmWw6UjDsA9To+k3dr+I1h5A7DvNyL7bBC3AqTeqCpQpeVQ2duB0y/CYnPUnIZ14aScOJ/fFP4p4+10oPRvmfomcD3/FlosrOT71v/j72sP85YNDuDyS/3fJdPVcKVXTlonnQIwhcXv/o1Ih82ZB9U5wdYI1bvDjbKlS1TtLlqnHDWWqf8CYOerx+b+Ao2fBP64CYSG+uYJCAfbSeSMq9BCNNk6gyN7szpPSdeVgUnyM7nqvCU/8J2YDZeSY2wKJ/SnXwHLlJ4dV2qU/xpdXGm1M9ByGqZfAdc/D2f8PihaTsPEBSjPj+dkVs7hizhie3nCUFrsR3B1+H566Gva/6Xu9I+sgNgkWfx2kR6U/DgXv/RyeuMJ3pWVmOeXPVrdpY+Hap1SQmT2R96b9hKscd5F27vcDvtxwEmVinxgwsq8O0ootOT5Wi70mPPGP7ANl5JiRfSAbx4+j9WGUdumPIfZjbHaSnfWQ6tfG+vRvQVM5PH8TVO/m+tOKaet088KWY2q/OUHqnwXTVA7pRZCt+t4GXdnaX8rWqFTWcqMt94ltqtl67nTfMWPnwQ/K4L9e4hnncqoz5pCTljQ0798PIlzsA9k4gSJ7BxmJsT1WDqbYrHS6PCfvkm1N5OLv2QfKyAk2QduNI3VtWASMSQ+zyN6WDsBL103A0tkMKXm+fZPPV/n4+96Eh89hTq6FWWPTeGLdEZXGaC5oavRLnW4qVwvMzNz2ofDtWyp9q5HLPlS3lZ9CzlRflV0TY5XslqONzC3qZrmNEJEt9mYLwi6RfWAbp7uFA5AUp8S/zaHFXhNmdInse7FxeonspZSs2lnFnKJ04qxhJgXmHIRZoTPZT+yFgPP+V9kjna2II+v4yunF7K9u5bLfrMJToQq6dRH7xnJIK1LZSPFpXaL+VocL2X3hUygcWWuMNVN9wUipIvuCUwIefqKpg8pmO/OGuzlMEMLsLzzEeCN7f88+wARti6OHhQOQbFNXBLp0gibs6OLZB0g/7GxVHrUleJ2bnceb2V/dymfmFQY9ZtTwiv0edesv9iZFi1Xp8kPvccXsfJ6+QDLDtQuLdCFtadBg+PKdbdBRr2wcISBrAtQdpM3h4o7nPmXmXW/y2QfX8Ul/V+eWrYW4FJh/PRzfqoqftdX4/PpubDnaCMDccTqyH3q8nn3v2Tg1QZosJ8erf5QWuxZ7TZhhb/Llywe0cVr6tHCe33yMuBgLl8wOs7RL6D2yN4m1wbjFcPh9xNrfsfi9L/Jz7sclLezPOgt3wxEeW3uYlqoydXxaEQAys5T2yn2c99sPeGZjOVfNK+R4Ywdfevjj/uXtH1kL4xbBhBUg3bDqx2p7kMh+w+F64qwWphWkBtw/3ESJ2BuRfVxSQLGvbeskK7lnGlZyvBHZd2qx14QZ9iZIyVeTgYFsnM7WXi0cu9PNy9uOcfa0XNIThyAFcaiJTwVhgZrd6nEgsQcltNW7YO3vIWsSsc5WDsRN5fljacS42vn9K+v59l9fAeClMgvfeXYbD+8SxLcdJzNe8uzXlvDrz5/CK/99OplJcdz8j000tnf2Pb72enXVMf40KFyorjD2vwlFi3xpl35IKXl7TxVLS7NGzTKLkjx7c1FVzwlah8tNp8tDqmHZ+JNsU6enVUf2mnDDqEVPfErgbJwgkb2UEpdH8o0nN1PX1sl1S8aPwGAHgMWiumh1NCjRT8oOfFyJ0X3L0QxffhXikmmqaOPAMy9CHDz52Xx2bzkIx+Dej9rpTKpmSu50YiolL55ZS0yxujrKTo7nj1+Yy1V/XsdLW49z/WnFvY/PnJjNnQFxiXDjKiX4RgP17uyraqW8voObzxjZ4mf+RLbYm12pvDZOoor23S7vYgvTokmx9TwVXhtHe/aacMPRDMn5SvCDZeN0K4J2oLqVS/+wBrdH0un2cM+VMzmtNIiIhgMJGUrsk3KCzz0UnAJJucrOMeyTRVlQnOiEJ3/NNFsD0yZI5PEYXv3hVWSmJCI8Z8LfXyPmje9D9Q7Y+iT810vMGzeTuBhLaGWVvZUrJ/jG0Qtv7VJlV86ZFuQKZQSIbBvHFHuzvLG3NaHvj9m72OsJWk2YYm8CW6r6CWjjtPSwcbaWN9LhdHPV/LH84dq5fHFRmEb1JqZvb7ZVDIQlBla+B1c+2GVzXpGRT994FJrKEaljyUpNQgihAr0rHlTZeh/dr+rl7PkPQghyU+OpbA6hKmb9IUBARuBzWFbbxi3/3Mwx44vjrV1VnFKUHnBucKSIjsjebDbu363KWFlorrpLie9p4yQZkb22cTRhh91oFh6fGsTGae1h45TVthFjEdx92czwS7UMhFfs83s/Lm1sz222VPX8xqNG2mW3jKPsifCl59Rq2lU/8ubm56favCvqe6XhsHpNa88sPpfbw23PbPUWcbt5eSnbKpr43vmBLZ6RIrLF3tk9su/Zh7a1l8g+Kc7w7HVkrwknpPTz7FOhqaLnMY6ekX1ZXRtj0xNODqEH78KqoJOzfZE+TpVFaKqA8Ut67i9eatwugw1/BaedvDQbu46HUEK5/hBkFAfc9eD7B9lW3sjCkkz+8+kJPj5Uz5g0G19YOG5gv8cQcZL81QeIqwNETNfUS+gySdvsFfuekb3FIkiOt2qx14QXznaV6mdLM2ycIHn23SP7ujaKs0d+mf6ACcXG6Y308Spibyr3pl0GpPh01dXu2EbyUmxUNtn7XmRVfxgyS3psbu908eD7h7hgRj6P3nAquSnxNHV08scvziMjaXSzniI/so/1WwYeILL32jgBInswiqFpG0cTTpgTsjYzG6ebZ+92KQvTb4JWSklZbTvzR2lBz4AwxT6lDxsnGGf8QJ2D41tUimYwxi1RGT9la8hP+xwdTjctDlfADD1Anf/2Wt/krB+vba+k1eHiK6eXkBhn5dEbFtJsdzIvDM57ZIu9q8Pn14NP+Dv9xT64jQOoyF7n2WvCCbNUgmnj2JuVtWPUXwnUpaqurZNWh4vxWVEU2efPhCv+FML7pKtVr/tXkbfgiwBUNdmDi32DkYmT0TOyf3ZjOSXZSZxarMY+fczoLKAKRGTbOE57N7FXkb3HT+xNiyY5SP/N5Hirjuw14YU5IWtLV9G9dHddLOhXBK3V4WL1rirKatsAKDmpbJx0ddvXBO1QMO86OLaJBQf+ANB7Ro5ZV6dbZH+4to0Nh+v53IJClfUTZkS22LvsXavPGWL/i5c3e0W+xe4kMS4maCOBZJtVp15qwgszsjdtHOhq5fgVQXt563G++vhGnliv6sScVJ79uMWq01Pe9L6PHSwLboQFX2Hszr9whmWbt6FRQLw59r7IXkrJT1/ZSZzVwmfDsdYQ0SD2Vp9nf6RFTbrUNTTy4xe2I6Wkxe4KGtUDeoJWE354xT5NVXAEtTBo+7/VfW+XqlQqm1QywktbjxNjERSGW6/Z3sicANe/on7P4UYIOO8eAKaLI72nX9YfgsTsLo1hHvuojHf31vCji6aRO4q59L0R2WLv7OgS2f99QxUAZ5Wm8OLW47yzp5oWuyuoXw+QFG/VhdA04YVp48SnqOge4O2fwiu3qXZ73i5VydS0+uq8FGYkEDvCrfBOKuISwZpAfmwrlU29if3hLhaOxyP51ap9LJ+cw3+Fa/kJQhB7IUSREOJdIcRuIcROIcRtxvZMIcRbQoj9xm2G33PuFEIcEELsFUKcP5y/QK+4fJ79jmNNvLBDlTC9YEqqsa2ZZrszYNqlSUq8VRdC04QX/i0HU4yKlYWnqoj+6LouXapqWhyMy0wkJd7KhJPJwhktEjPJj+3oPbKv2+/reAXUtjpodbg4Z1puWHr1JqFk47iA70gpNwshUoBNQoi3gC8Db0sp7xVC3AHcAfxACDEduAaYAYwBVgshJkspR74DiLMDknPZW9nCfz2ygdTkFOgEq9tBdnIclc0dtDr6juxb7aq5QTj/ITVRhKMFEKpefcFs+NYuZXX8sgQOvKU6JYER2TcxPiuR3149h7SEyE6+GxISMsnpbAsu9vZmaK2CrIneTRVGSYRwt8j6jOyllCeklJuN+y3AbmAscDnwmHHYY8AVxv3LgaellA4p5WHgALBwiMcdGkZkf9fLO7AIwRMrTwdLLDjbyUu1caLJ3qeNk2yz4vJIHC7PCA5co+kFR4uK6i3Gv2/aWLWAatwS2L/aLxsnldoWBzkp8cwfn8HE3ADNxzVdScwkQ7QGz8Yx2xn6i32DKfZh1se3G/0y8IQQxcBc4GMgT0p5AtQXAmAmw44Fyv2eVmFs6/5aK4UQG4UQG2tqagYw9BBwdkBsAnWtnZxanKEyEYym4wVpaqVci90ZsC6OSUq8LpmgCTMcrV0mB71MOk/Vf6/4BAAZl0SNIfaaEEnMJJ1mqlsc1LYGyMipCyT2Ku11bLj18e1GyGIvhEgGngNul1L2VjwikNfRY+2xlPIhKeUCKeWCnJycUIfRP1wOsMZjd7mxxRolUo1uVflpNiqb+47sk0yx15O0mnDB0Rxc7AF2/Bts6TR3Wuh0e8hJ1mIfMgmZpMoWpFQZTD2o3Q+ILhO0xxo6yEiM9WpFuBLS6IQQsSihf1JK+byxuUoIUSClPCGEKACqje0VgH8hikIgwFkbAVwdYE3A7vRgizW+14wGJvkZNhrbzVIJwSP7ZB3Za8IN08bpTs5kuO5F1bQndxo1rcqK0JF9P0jMIsbRxJyxKfx7UwU3nt5tlWzdAdXL1i/Lr6KhI+wtHAgtG0cAfwN2Syl/47frZeB64/71wEt+268RQsQLIUqAScCGoRtyP3CqRVV2p5t4qxnZJyqxT/NdciX35tlrsdeEG8HEHqD0TJh8HqQXUd2ibAgt9v0gMROkh2tmp7H7RDM7j3crMld3oIuFA8rGCffJWQjNxlkKXAecJYTYavxcBNwLnCuE2A+cazxGSrkTeBbYBbwB3DIqmTgej6pkZ03A4fT4bJw45dnn+y186GuCFrSNowkjAoh9dYudA9WtXbbVGGKfq8U+dIwm7heVxhEXY+GpDUd9+6RUnr2f2EspOdbYEfZ+PYRg40gp1xDYhwc4O8hz7gHuGcS4Bo/RuMRjtdHpDmDjpPnEPjWEyF7n2mvChgBi/7+v7ubVT4+zcvkEvnPuFOKsFq/Y5ySH54rOsCRRiX2qbOEz88by7MYKbj1rkloV21qlisxl+XLs69o6sTs9ERPZn5wYYu+yqBrSvgnaRO8ErUmvnr3xRdDc4RymgWo0/cTR0qO/7NH6dhJjY/jL+4e483lVCqSm1UFcjIVUnV8fOobY017Pf8/2YHO38vAaoxZO1Q51m33ypV1CJJc4NhqUOIUSdZvVL7LvbCc53kpKvJUWR++1cdIS1BdBkxZ7TTjg8ajosltkX9Vk56JZBRSkJ3D/2/uZMy7dm3apFwP2A8PGoa2asW/exF9zl3LD+hS+saKU9P1vqRX5RYu9h3vTLnVkP4oYkX2n6BbZWxO8+8zoPrthC7RW93wNIN4aQ2JcjDdzR6MZVbxFznxi7/aoKD4v1cbtZ09ixZQc/veVXWwrbyRb+/X9w4zsyz8GRzOntr0PnW38e2M57H0dSs5Q835AR6ebdQfrAC32o4sp9qjI3Gfj2LxRf36aDYGHMa98AZ78rCoiBeoy+bXve2uQpCfE0qgje0044Ff3xqSu1YHbI8lLs2GxCP7vqtnEWS0crGkjJ3l0W+GddMSngsUKh94HwOJq5+acnaxdv1b1s51yAaCu9M/81Xs8+fFRTilKD97oJIyIXLE3mo07hIpsfBO0iT6xT7WRSyPC2Q4ntsF7P1fHHHofNvwFytYCkJYYpyN7TXjgXwTNwFzan2dE8XmpNm4/R00i6rTLfiKEsnKayiEmHtLHc038h0xtUlrAJFXX8YN9NVQ22/nd1XN44eunjeKAQydyxd6lBN2OimzivTaOTe2Tkplj05ifauTR5kyFNb+Dlkpf27H2WkBF9k0dnWg0o45X7H0TtGY5Xv+kg+tPK+aiWfmcOWWALf2iGdPKyZsOc75Ibu0GfhD7NEfiSlUdIuD9fTWkJcRy6SljsFhOjjmRyBV7I7K3mzaO1c/GkR5wO7lu8Xj+cIHxh138dUBC9S5fJ5o2Q+wTY3VkrwkP/GvZG1QZKZb+a0diYyw88MX5nDdjBFr6RRrmJG3+bKULF97HJ+NX8q2W63h/Xw1SSt7fV8OySdnEnCRCD5Es9kZk75DmBK3xq5qdq5ztWCyCmKYjgIBSY8lAzT5oKFP321SBtvRE7dlrwoQANk5Vk50YiyBL18AZGszIvmC2ag6zaCWzr7uXusw53P3KTraUN1LT4uCMycNU02uYiFyxNyL7Dtl9gtYQe2MCl8YjkDoG0gpVTfDafX42jpppT0uIo6ndiZQ96rlpNCNLEM8+Jzn+pIoywxpT7PNP8W6Kt8Zw16XTOVTTxjUPrQfQYh82GJF9ezCxNyZpaSiDjGI1MZM9Gap3Q6OxRNovsu90e+hwjnzVB42mC4Ei+2Y7eak6qh8ykvNVRk63RudnTc3jya8uYnxmIksmZIVtr9lgRLDYKx+z3dPdxjH+QGZk33AE0o2+kdlTVC1wj1Eaoc03QQsE9u1r9sE9Bb461xrNYOlsU/1k2+t77gsq9ieX8IQ1i26GL78GcT3bOC6dmM1b3z6Df960aBQGNjgiV+yNyL3No1bH2vyrXgI425XV03ICMkyxnwQeQ9DTxnltnPTEXsS+/pB6rdp9w/N7aKKP8g2w6VEoW9Nzn6NZzTvFxCKlREpJZZO9SyaOZpAkZcG43sX8ZFyVHLnlEozIvc0dYFEVKKFvKgeksnFA2TgmhQtg72sgJWkJ6uqgMVD6pWEX0dEwxL+AJmoxV3N3BInsjaj+6//YTKvDRbPdpSN7TZ9EdmRvsdLhVt/A8dZu2TiuDmXhgJ+NY4h9TBzkz1JfGJ1t3si+KVBkb0wEa7HXDBmtleo2oI3ja0m4pbyBNQeU1ajFXtMXkSv2LrvqUuVyE2e1+BY++E/Qmlk3ZmSfUawakqePh+Q8ta2txmfjBEq/1JG9ZqjpM7JPxuX2UNPi8JbWLc4K/6qLmtElcm0cZwfE2lTjEqvfd5pX7O3QfFzNupvCHmOF3KlK7JOy1bb2OtJzxwFBPHtjIliLvWbIaDEj+wCfKaO8cW1rJx4JX19RytLSbMZrsdf0QeSKvRnZO/2ajYNfNk6H0bg5FSx+XwbX/FPVxGiqUI/barDFWoizWgJ79k4d2WuGmNYqdRsssk8v8tbDyU+1UZzdM2tEo+lOZIu90X+2i9h7s3HsgXt5po/zPR+grRbRUEaGLSawZ+/Snr1miDHFPqBn3wzxKVQ2qSBDe/WaUIlcz95pB6sNu9OvJSH4ZeO0d5ns6oFp45Svhz/M50vWtwPbOGZkH+gfU6MZCN7I3i+A2PkC3DdRZZDFJQcsfqbR9Ebkir2rQ4m9q7uN41cuwYiSAhKXpK4CPn0WpJvz3O8HSb3sJbJ/6RY4+O7gfg9NdOHsALtRidXfxjn6sdo+6/NwyrVUNquWg5mJul69JjQiV+ydfjaO1U/sLRblyTs7Ats4/iRmg7sTRAxTnLuJbT0W4H1Mz76x6/aOBtjyD5WrHwr2JvDocgxRj5mJk5ynPkNmPaaOBkjJh8/8BYpOparZTm5q/ElTXlcz+kSu2Ls6wJpAh9NDfGy3X9PsVtWX2CdlqdszfgDAwrb3A7yPEdk7msDt8m03c/hbTvQ9VmcH/P4U+OThvo/VRDamhZMzVZXtMEsad9RDQob3sBNNHV1KGms0fRG5Ym9E9o7uE7Rg9KHtUC3e/Nq79SApB2KT4LRvciJpKmc413CwprXb+3T47puX3+Ark2ym0fXG8S0qcqvZ0/exmsjGFPtcowiXORfU0dBF7KuaHeRpv17TDyJX7I3Ivkc2Dqhc+2DZOP4s+666bI5LInnaucywlPGNJz6hzeEXwZuRPXT17RvNyL6q77GWf2wcG8IXgyayMT8DuVPVbUdPsffWw9GRvaYfRLDYOwzPvtuiKlBi72hRGTl+7d16MG4RTLsUgJSsMcTgoaqmmrN+/R4Pvn9Q1bd3BhF708ZprfT5rn5IKVl/qI5Ol0dNvoEWe43y7IXFV7rDXFjV0eDtoNRsd9HhdFOgI3tNP4hMsZdSiXlccs9sHFALq4z+ssT3YuP4YzQ0+NvnJjAhO5l7X9/D1vJGdQVhSwPA0VLLH97ez6GaVl9k7+4MmKnz5MdHueah9Tz7yVEd2Wt8tFaqxIAkozFGRz14PF0iezPtUufYa/pDZIp9e52K2tMKDRune2Sf6Mt66M3G8SdRTdbOz4E/f2keMRbB27urVWSfqpoQP7xqI79+ax9XPvARLZUH8VhUTZ3uk7Q7jzfx01d3AbB352b1D52cr/xanZET3bRWQ0qerw9qe72apJUen9g36xx7Tf+JTLE3omqZPs5YVNXds7d5u1CFLPbef7460hPjWDA+g9W7q8DVgSe5AIC62ip+dNE0cpKsxLVUsMVVDMCDr36k7Bpgx7Emrn/kEzISY7lwZr4vqp9+GUi3t4a+Jkppq1GRfUI6IFQgYF4ZGmJfZS6o0pG9ph9EptgbfnlnchFAYBvH2a7uhxzZG2JvTJidOz2PPZUtuBwdfFIXi0cKrpqWxE3LJ/DKDROJF04KZq4A4MCh/fx97WG2ljdyzUPriYsR/OPGRVw+ZwxT3PtxxaVC8enq9UNJ1dRELk67WtBniVH2YLuf2BufweoWJfY5KboVoSZ0IlPsjcjenlQI+NWyNzErXwLE9VPsjcj77GmqUmZrWwt76tx0xqYwI0NF7wnG4qsxs84AYEmOk/vf3s9XH/uEzKQ4nv/GUiblpbCkNJtiSxU1cUWQMka9fi/ZO1uONnD3Kzu547lP2X2iObRxa04uXHawGiKemBkwsq9pcZBqs/YMYjSaXojMQmiNRyEhA3uMKnoWMPXSJNTIPj5VlUM28p5LspOYXZiGrbaT06YWEl93QH0R7HoZ6g6o5+RMBVs65xZJ7qyRxLg8PL3yVK/XmpYQyyRrDbs6p5KTlIsVkC0n+M4zW+l0e7hhaTETc1JwS8nqXVX8+MUdWCxqvu5wbRvPfG3JYM6SJhxxOXyVWRMyu0b2hthXtzhOumbXmtGnT7EXQjwCXAJUSylnGtvmAA8CNsAFfENKucHYdydwI+AGbpVSvjk8Q+8Fo4m43akmOwMuqjIJVeyFMP75fJ76MzctxvaLTiaNyYH2DNj5Iux4zvectCJIySfVVctjN5xKZnI8E3P9sn9cneR4avi0/TSeeukYDwG79+/j+a3ZxFktvPppV0tn8YRM/vKlBTy/pYK7X9nF+kN1LJ6QFdr4NScHbofqlAYqsm+t8hN708ZxkKstHE0/CSWyfxT4I/C437ZfAndLKV8XQlxkPF4hhJgOXAPMAMYAq4UQk6WUI5ti0ngE8mZgdypbpWc2jl9UFKrYg8rI8StOlWAxFldZ49U+6Ybl34fONnU/1qbqmTQfZ8l710DJcsj/H79xHsWChwVz5vP7TQ00xqeya+8+Tim8kMdvXMR7e6upaXFgEYLCjARWTMklzmrh2oXj+NO7B/nNqn389foFpCXE9vcM+ajZq3K6T8IGyhGJf2SfnAfHNvuJfTqgPPt54zICP1+jCUKfYi+l/EAIUdx9M2CuRkoDjhv3LweellI6gMNCiAPAQmDd0Aw3BDweZeNMudAX2Vu72zh+XX16K5fQncTMrqWMzZaEsQmw4k5Y8BWYelHX5yTnw6H3fMf5Y7RFXLrwVJ6eP5m6JzJIc9bx08tnkpYQy+VzxgYchi02htvOmcT/e3EHi36+ml9/bg4Xzy4I/fcwKVsDj14MX/gXTD6v/8/XDD3+nn3udNjyhPpCjkuBmFiklFQ368he038GOkF7O3CfEKIc+BVwp7F9LFDud1yFsa0HQoiVQoiNQoiNNTU1AxxGAFqr1EKmXm0cI3KKTVStCEOlu9ibq2etNiic31PoQUX2Jt0XTdUfUreZE1hYkklxcSnLC9ycUpTe51CuWzyel7+5lIzEOF7cGqAaZyhselTdntg6sOdrhhaPR312zc9n/kx1W/ah169vcbhwuDzkpmjPXtM/Bir2Xwe+JaUsAr4F/M3YHsgL6FkrAJBSPiSlXCClXJCTkzPAYQTAXLmaPh67K5iNY0TY/bFwoIdn3yWyD4bZ+apgTgCxP6wKrRmrJWNSC4jvqO56zAe/ggdOgyc/7yuuZjC7MJ2p+SmcaOqg37TXq8lkgOpd/X++ZuhxG/0SrIZnn2eIfWsVJBqTs82q53Fuqo7sNf1joGJ/PfC8cf9fKKsGVCRf5HdcIT6LZ2Qwa9Jk9BLZD1TsTc/erHVjNhu39hJlnXIt3LgaZlypVkI6WsHVqapl1h+CzAk+vzzFXEXr8T1/6z/Vex5YDZsf7/HyBekJHG+099jeJ9v/pSYDMydA9QCqbe57E+w6/XNIcfldKYK6kjRTcr2ZODrHXjMwBir2x4EzjPtnAfuN+y8D1wgh4oUQJcAkYMPghthPmo6q27QiP7Hv9mua2Tj98etB/fP51xh3hhDZxyVC0amQYnjqLZXw4tfhL8uhZjdklviOzShRE7tVO9Tjznb1hTD/yzD+NNjznx4vPzY9gfq2Tjo6+zkHvutlyJ8F0y9XqaJuo+Vi7X5Y90Dvz22rhX9+HjY+0r/31PSON3jwE3LTyvHLsQe0jaPpN32KvRDiKdQE6xQhRIUQ4kbgJuDXQohtwM+BlQBSyp3As8Au4A3glhHPxOlsA0ssxCV6BTAhrpsvb2bjDCSyB59v3z0S6w3Tu285Dsc3Q+0+NZHsL/ZTLlK5/NufVY9rdgMS8mbA1EtUvfu6g11edky6eu9+WznNx9Q6gJxp4HH6XnfN7+DNO3vvqdtmFJEzv5Q0Q0Ogz1NeYLHXkb2mv/Qp9lLKa6WUBVLKWClloZTyb1LKNVLK+VLKU6SUi6SUm/yOv0dKWSqlnCKlfH14hx8Al8MbaTfbVbSaausu9qaN00t540D4F6eC0CJ7k1TjcrzhiPLezcvzzFLfMUlZMPFc2P6cKohWtVNtz53um/ztFt0XpKn37reV01qtUvvMuuk1u5U9dcjomVt/OPhzzfTT6t39e09N73g9e3+xn6Fu/XLs462Wnp9pjaYPIu8T4+zwXga32F0IAUndI3vTxgm1vLFJt/o4A4rsj65TFQzP/h/l1U+9uOtxsz8P+15XaZFVu1TGUEaJ6p2bP1uJ/dJbvYePTTfEvj+RvaMVnG2QnGvk2FuUcOfOUBE/qLTQwvnUtTp4ZmM5B6vbsMVa+MaZExlr5n3X7FX2T8wg8vw1PszPU4xfE/H8WerW9OyN3rNCr4vQ9JPIE3u/RSktdhfJ8daeTZkHbeMYGTn9iezjU1Su9OEP1eOcyTB2fs/jplyojtv8uJqszZ2uhB6UlfPeL4yoPBdQNc2FgOON/RB7s/Vdcp4ae0aJEnszqgeoP8ymI/V84a8f43B5KEizUd/WyQf7a3h+0QlywGf/mFcHoMRfxPjGrAmdQBP+2ZPhvHtg5mcAc/Ws9us1/Sfy/iP9FqU0252k2gJEneaiqn6nXhqrFgfi2YOK7s0J5KxJgY+JTYBTv6LKLlRs9F3Gg3EVIGGvzx2Ls1rISY7vXewbjqhm5mYWkVne2WyQkTsNjnwEnz4DGcVqIVjDYR5ZU0ZiXAyrv72cdXeezbNfW0Jju5OHVm3yvXb1Tt/91hr481J46RshnQ5NN7yfJz8/Xgg47Zscdabz6NrDHKlr1wuqNAMiAsXeF9k3d7hICeRtWgcY2dvSVdRau0897k9kD5BqZOQk54Gtl/mCpbersbk6uop93gxIH9/Dtx+TnsCJpl48+21PwX++A03Gejf/yB5g2beVqBzbBBNWQGYJrtqDvLWriivmjmVirjpPpxSl88I3lnLxxHhcxOCSFurLtqnXsDfDP66E2r2+FcMmrTVd00k1gQmSylvT4uALD6/nJ6/s4lhjh3eeRqPpDxFo49j9bJwgkb05MWtOuIaKxQJzroXNj6n8+X5H9obYB4vqTRIz4bT/hnfv8Xm2YHj8l6go3a9Z+ph0G3sqW4K/nmk7VWxUi7zMLl2m2I+dDzevgff/DxauhLX307nnLTrdHj47v7DLS03MTYYsD+6aTMra4zix9WM22faztPYZFlRux1F6AfEH36D2RBkn3BnUV5ax9LVzeX3Cj3g39gxqWh3ctGwCyycP4UK6SMEr9j7P3uX2cNPjG6ltdfDoDadid3qYP17XxdH0n4iO7FvsQSL7pCy47gU1Gdpfzv8FpBXCCyt9i4pCjexNsc+e2PexS2+Hzz8B47qVMZ56sVoMdWC1d9OYtARONNpVA/RAmLZTxUZ121qlrlAS/b7sUvLhkt9C7jRkRjGJjhrm5MczY0xaz9fraCAmMZPkotmUuMv47ep9tOxcxUFPAdfuWgzAnX94jEv/uIY3XngMq8dB+d7NfHy4ng2H63luc0Xfv380EiB42HC4nq3ljfz0spmsmJLLBTPzddqlZkBEoNj7ZeM4nKQGqwhZepbqCNRfbKlw1v+o9MnjW5RohpqNEmpkDyq6m35Zz2qURYvUYrAjvtpyBekJdDjdNLY7A7+WmT1zzE/sk3JUN6QAbGhOB+DmU4I0x+hogMRM8qYsYqysZPetxayI30f6zPM5/+xz8BDDrVNbeOi6+dwxUVlHX58bz9o7zmJJaRb7qlr7/v2jkQCLql7fUYkt1sIlpwyg0J1G40cEin0Inv1gMa2VE1tDj+rBl36ZHYLYByPGqhbanNjm3TQ+U004bzzSEPg5ZqroiW0qW8Yvm6c7zXYnD2xT/vp5+e1BXq9BWWCzrwYRQ8JrtyFcHWTNPp+vnTMLS950ZomDnDclg7TjawEQzapqxuS8FA7WtOL2BLkKiWbcXT17j0fy5s5KVkzOJbF7+rBG008iUOxVNo6UklbHMIl9Vqla6dpWE7pfD+pq4rT/huJlg3v/glOgcrt30nP55BzGZSbym7f24Qkkou31quCay65WvbZW+fz6bjyzoZxtbcoTtjSWBX7/9nqVmZQ6Ri32qvhEnQ+zj+7Yueqq58halc+fmAVNyrqZlJtMp8vDkbq2QZ2CiMSM7GNUZL/5aAPVLQ4unJXfy5M0mtCIQLFXkX17pxu3R5ISaIJ2sMTEQpbhu/dH7G2pcN7PVL2cwVBwihLRelXiIM5q4TvnTWb3iWZe3R6gYXlHI0wwShlVbFTZMUEi+0O1rViTMlXmUbBqmB0N3iqMnPpVdVt4qi/DaMw8dcxr31fCNeMz0HwcPB4m56lJZW3lBKBb6uWbOyuJi7Fw1tTAfyuNpj9EoNjbIdbmVyphmFZ35kxRt7GjsMCl4BR162flXDp7DFPyUvjrB4e6Hut2gaNJPSelAA68bUT2gQXkWKOdsRmJKgVz/1s9UyadHWpexFxzUHIGTLsM5t/gO6b0TFUOIiYOVvxA2VZuB7TXetsy7q/qJXsoWumWermnsoVpBSnDE7Booo4IFHsV2bfYVcvAYbFxQBUQg679bEeKnCkqYvZrOmKxCC6bM4btx5qobvbLuffvXzr787DvDbXyNYiNc6yhnTHpCSrFs7XKN6lrYmb2mGmrQsDVT8ApV/uOySiG7+yGb3wEy74DqUb/mqYKkuKtFGYksK96FCJ7t1OlrIYrLocx4a8+s8caOxiboXPqNUND5Im9URunxYjsh0/sRzGyj4mFvOlQvgHe/Tkc/gDAe7n/3l6/zl9esc+Aedfj7SUTILKXUnK80a7q7Uw6V/nwe17tepD/64VKmk/sefhcvpK4ZnQi+/d+AQ+tGPn3DRW/NSLqb9HhrX2k0QyWyBJ7j1tFrVYbzUZkHzT1crDkmpH9KNUpKTgFyj9WC6HW/xmAqfkpFKTZeHevX7crMxMnMUNNLE9YoR4HiOwb2510ON0qsk9IVxPJu1/1lVno8nr9WJCWaizM2vcGVGxgbsxBDtW08cB7B/joQG3orzNYytaq/gDhuprX5fAuqKpr68Tu9Ki/hUYzBERWPpdfnnJzR5DyxkNFZqm65O5P6uVQMv0KXypl3QEAhBCsmJLLK9uO0+nyEGe19LRdlnxTXRFk9VzYdcyor+MVmOmXwavfgn9dD1MuVk1bBrL6OClb2U47VHOzsfF2nB4Pv3xjLwBnTM7h8wuKWDElh6T4Yfp7eTwqE0l61O+RkD487zMY/CJ7s9aRjuw1Q0WEib2ZzZDg59kPU2RvjfNNeo4GpWeqn7fugnV/UhOxMVbOnJLDUxuO8tauKi6eXdDTdpl0LtxZEXBBlSn2haZPPPe/VE7+2t/DrpfUtuwpXV8vFIRQVo7RYD03po3dP70Al0fy1MdHeeC9A7y/r4ax6Qk8vXIxRZmDzFYC9SW46yWVCWSxqJLNncY8QUeDT+ybT6hS1/2tk2RSe0BdMQ1FyWF3pzcT51iDIfbas9cMEZFl4/hF9sM+QQuq5MIF9w7f64dC9iRlXRnVNM+YksP0glR+8Nyn7KtqCWy7BFk5awqMN7KPscKKO+D2HXDLBlW6oXZvz9cLBXOSFqCjEVtsDMnxVm5aPoFPfnQOj31lIa0OF9c8tJ5/bSznYE1r8PIPobD3dXjuRtj7mnpcud3v/f0Wnz16Mfzzmq5WVajU7IM/zoetTw58nP74RfbHdGSvGWIiTOx9tUVa7E6sFkFC92bjQ0lC+uBz5geLaccYbQXjrTE8fP0CbLExrHx8I87WWmU3hdCV63hjBwmxMWQkdrsaSspSE9KLblaPrbb+21dphm+fUdxVbAFrjIUzJufw5FcX4XR7+N6/P+XsX7/P3P99i7c/2QEH3+nfe4Gvi9Y+oxx0ILFvr1drFY6sUSWl+2L3q3B8q++x2Zbxoz8M7MuiOy6Ht3HJscYOkuJiSBuuOSdN1BGhYh9Ps91Jis0a+R19TLGv3e/dNCY9gT9+YS5lde1s21emLJcQzsPxpg7GpNuCn7Opl6gIvb/VQgEK5qjnTjq/h9ibzBybxvo7z+atby3n/66axSLLXua9cSk8caV3BW7ImFcg+1Ypv75yuy9N1nx/U6zjU2HVj32F4gJRvQeevU5dCVQY9fyNRW3U7IGDb/dvfIHwj+wbVNplxH9+NSNGhIq9zah4GQVRUWIW2NK8k7Qmiydkcd3i8VRXHcceG6ByZQCONXT0nv0RY4WLfwPLv9P/cS76Gty2TaV8Ott8lls3LBbBpLwUrp43ht/Je7G6jJ4BDWX9e7+afSpKbqtWpRsqt8N4o4KoKfaVhthf9TA42+Hhs+GNHwZ+vbfvVgXoErPgyaugrQ7qDqmCcsn5sO6B/o0vEC4/z76xj7+FRtNPIkzsDQGJtQUvbxxpCKGqaHYTe4A7LpxKVkwbJ5yhWU3HGu2+ydlgTLnAVyKhv+OMifVN7AaJ7r3UHyTB3coj7gvV48ajfb+Hx200SnGrBjOzPq/6666+C1qO+2oS+Uf2yXkw+Xz41k5Vu2j3yz1f9+h65f0vvQ0+85B6/pE1KrLPngLTLlX1gQZLN89e+/WaoSTCxL6rZz9spRLCjayJXs/en6R4K2Pj7ZR3xAcukGbw4pZjXP6ntdS2Ohgz3F2QQhV7oy7P+3KuehyK2G//F/xupmrq7nbAuEUwfimUfQiFC1XDmbgUv8j+U8ibiZSSE3Yr7qIl0FTOz57/hOc2VdDeqSb52fiIunpa/A2OJU7FhZWXXnuV1uN7qYkbC+lFKp1zsKtzXQ6wxtPmcNHY7tSZOJohJcLE3j/PPkoie1Bi31wBnT1LEmeKNqqciew43hTwqa9vP8G3nt2Kw+nmi4vGceW8sQGPGzJCFfuqXSAsxI6ZRZ0lCxrL+37t2v3qC/+de9TjnKnwuUfhvzfDV99SbSETMtR7u51Qs5eOzGl87YlNLPnFO3zrbWUZfbxpA9/51zZOu/cdHly9A+fOVzhWcC7EJfLHD46y0zOemfbNJLsbeXiXhZcOGl+kzQGK0PUHo2KrzrHXDAeRJfZmT1gjso8Kzx589fEDVKlMcDXRRDLv7Knusn3HsSa+969t3Pr0FuYavWXvuXIWhRnDnF1kpmyai738cdrh6S+qxWLVuyCzlEWTx1LmyqTy6N6+a+C3Gb/j0Y/UbfZktaArq9R3TEK6EvvafeDu5IE9Cby3t4ZvrChlysx5ADzzmSyeWbmYuUXpbHnnX8S62/nB3kk8+P5Bntt0jI6c2ZS6lG02ftJMntxtXAE0HxvgSTFwd4LVRlWzClryUkdpdbYmIokssTciexkTT0O7k/TuKYSRSslylV7ZrRE59iaEqwNbWo5X7FvsTm57eguX/GENr20/wecWFPHIl08lIW4YU1T96S2yr96lavGs/7O6nzed/1oyHnvSWDpry/jfV4OUXDZp9ftCS84LvEo2MVO9tzE5+2ZdDl9fUcr3L5jKLVedB8JCYvMhFk3I4u83LOTX0w/iSsjCPe507n19Dy6Ph9I5Z3hf7toLVjBnpmoK31wdgtXUG0ZkX9emPsfZybr9oGboiDCxV559VQd0ON2UZA+g7eDJSGImlCxTk4v++d5GgbTkiafxaUUT+6taePD9g7y87Ti3nFnKuh+ezc+vnEV6YlyQFx4GehN7Y4Utu1+F+sOQO53s5HhOmz+PsZZ6Xtx8FKe7l7o2rVWQP1t98ZmF6gK9f3s9VO/CY4nlkCefGWOMNQjWeLUOoHafelx3kOSy1VhnXM5vrp1PVlIcl88ZS84UX19gkTmBq888FYB9+/f040QEwKjYWtvaCUCOFnvNEBJZprYR2R9udANQmpM8mqMZWaZdBv/5tlpMlDddbdv3JsSnsvzsS0nY/CH3vbmX9YfquGBGPt87f+rojDMuWVXTDCT2ZkZRpzHRmat+D5FeRAxuEuw1fHyontMnZQd+7dYadZUz48qAtX8An2dfu4/mxPG42q1MK/BbcJY9WXn/HY3wz6vV4rGlt1OQlsB731uhFukJqX4PWxrEJVI6JpEmkUpVxaHA7xkqLjvExFHb6sBqEaQmRNa/p2Z0icjI/mC98lBLc6Mksge14AnhSx2UUjUfKT2TzNQkvrR4HKt2VdFsd7Fy+YTRG6cQhuAG8OzrDqqmJ4lZ6rEh9qSPA2BCbC1v7qwM/LpS+pqyLPu2KuIWCFPsq3dTEVNEslFf30v2JCX2L35D1dO5+gnIGA+oOkvWGIsqN1FyBoyd732aO7kAW0cVO4NMhPeJlF7Pvq7VQVZynF5QpRlSIkzsVWS/r76TFJs1ui6DU/JUa0CztEDldmitVCtWgZuWTSDOamFhSSZzx/WjiNlwkJAZxMY5CDmTYeZVKmrOLFHb05XYnl3g4K1dVYHTSDsaem3K4nvvDJBuaDjMblcBU/NTuopq9mSVtrn3P3D2Xb6+ut353KNw1d+8D5Nzx5Mv6nlnd3Xg4/vCL5OsrrVT+/WaISfCxL4DRAz7axxMzE2Ovsgob4bPb96/St1OPAeA3FQbT920iN9ePWd0xuaPGV37I6WycTJL4Zy74Wsf+gq2GXV1Fme2Udls55I/rOGNHd3SHM3J2SDtFru8t8GG1pyuFg74qnoWL1PloINhjfPWngeIyyhkjKWBAzUD7MDlt0akttVBlhZ7zRATYWKvJrgO1rRGl19vkj1JiWh7vVrRmTNNRfwG88dnhkfudkIGtDd0nUxurwd7k/La4xK91gmgfPOkXKbaGvjxxdNo73Tx4xd3dK2K2Raq2Pvq+mzvLGBqQbfSxmPnw4o74TN/VaWRQyVlDBk0c7Q6gD0VCt7IPo7a1k6yk0Zw0lwTFUSY2NvxWG1UtziiU+z9i6JV7YT8maM7nmCYnv0jF6jmKOArKuafE+9PZgmWhjK+umwCK5eXUtvaSVmd3yIyb2Qf2MaRUvLTV3bx5WdUwTgPFg7L/J6RvVnWObWffQpSxwDQVlPe62rloLh9acN1bQ6yU3Rkrxla+hR7IcQjQohqIcSObtv/WwixVwixUwjxS7/tdwohDhj7zh+OQQfFZcdlURFRaU4UTc6amGJ/bCM0lStbJxxJzFQLkMrXw84XVVVKMxMnWBZN1iSvRbWwRFkxnxz2i6Jbq9RtUg4ANS0O7n5lJ+/uqaa62c59b+7lkbWHGZuvRLmCXFKSk5maP8CmJd0xxD7DVUulf8P3UDEiewdx2J0esnRkrxliQsntehT4I/C4uUEIcSZwOTBbSukQQuQa26cD1wAzgDHAaiHEZCmle6gHHhCXAwdqIdXE3CiM7NPHgyUWdhkZOblhKvbmYieLVUX4ldtUJo6I8Wbe9CB7Emz9B3Q0UpqTRkZiLJ+U1fP5U4vU/tZq9bsnZCCl5EcvbGfVrir+vrbM+xKfX1DIz87Pg19D0eQ5bLjmHCyWIZrXMZqz5Is6Dta09r9ipeHZN7vUPIX27DVDTZ9iL6X8QAhR3G3z14F7pZQO4xgzBeFy4Glj+2EhxAFgIbBu6IbcCy47DhmLEAxNa7uTjRirymApX68eh2tkb/rm59wNq36kMoiObVQLmmKCrHrOnqxu6w4gChewoDiTT8p8kb27pQp3Qg5r99VwqKaNVbuq+O55k5mYm0x1i4MJ2cksKc1CeJxgsSJypyOGSujBG9mPEfUcrG5l2aSc/j3fiOybOtXFdnayjuw1Q8tAV21MBpYJIe4B7MB3pZSfAGOB9X7HVRjbeiCEWAmsBBg3Lkg011+cdhzEkhxvJTYmsqYjQsa0O2zpXgEKO6ZfrsoQn3ojfPo0fPyQShM9+67gzzHr/9Tug8IFLCzO5K1dVVS32EmOt7J9xx5sLhs3/F2VGp45NpWbzyhVefH+WOJUO8m8IZ7PiE9GJmRQTB07atr6/3xT7J3qC0inXmqGmoGKvRXIABYDpwLPCiEmAIFCpYCzVVLKh4CHABYsWDAEPd0Alx07cdFT2jgQ5gRn3syhaYI9HCRlw6KV6v6EM+Gj+5V9s/gbwZ+TUaxsH6Mj14Ji5dt/sK8Wl9vDLFc9Kbnjee7SJdidHuYUpfcUepOS5UP4y/gQ6eOZ6KrnpYGkXxo2Tr3DtHF0ZK8ZWgYq9hXA81Llvm0QQniAbGN7kd9xhcDxwQ2xH7gcOKSK7KMWMwIOVwunO1MuVGJ/7v9CbC9VHmNiIaPEO0k7uzCdqfkp/G71PlJssfwjppnMomLGjR9Ay8ShIn0chXVbOTggsVeRfb1DfUFnJenIXjO0DNTreBE4C0AIMRmIA2qBl4FrhBDxQogSYBKwYQjGGRouOx2e2OipYx+ILFPsp4/uOEJl/Gnw7d0w44q+j82e7M3aibEI/ufS6VQ0dLDnRCMZsgmR3E+ffKhJH0eWq4qqZjsdnf3ISdj7Omx7CoA6uyDVZiXOGqU2pGbY6FMVhRBPASuAbCFEBXAX8AjwiJGO2Qlcb0T5O4UQzwK7ABdwy4hl4gC4HHTI1OgW+6KFKkqe8ZnRHknohDq3kD1JrQx2uyDGymml2VwwI59PDxzBgsdXU2e0SB+P1eMgm2Ya2jtJiAshI6d8Azx1DSAgOY/DnalkJ/dS2VOjGSChZONcG2TXl4Icfw9wz2AGNWBcdto81uhpWhIISwwsvXW0RzE8ZE9S9W+qdsCYOQD89uo5NB5PVAnCCaNc88dIGy0S1dS3dfadful2wiu3Q2oh3PIxxCdT8dA6sqMwa1gz/ETWtaLLTpvbSnI0R/aRTPHpqofsY5fCntcASIiLoSDW6FAWJmJfKGpoaO/s+/itT0L1TrjolxCvFL6mxaEnZzXDQsSJfYvbGt02TiSTUQxfX6Nsn9V+aZpmUbWwEfta6ttCEPujH0NyPky92LuppsVBri6VoBkGIkrspctBhyc2ulMvI52MYph0LjQcUWUWwCf2tvTRGpUiPhlPQpaK7EMR+5o9kOtrImN3umm2u8jVvWc1w0DkhMBSIlxqUVV6NKdeRgPp41XhsNYqVbAsXCJ7QGSMo7Cthk3tzsAHtFSBvVFlFtXshXnXeXfVtKj0y0jqw+B0OqmoqMBuH0C9II0Xm81GYWEhsbEDD2QjRxXdKpJyyChPvYwGMorVbeMRJfb2RvU4UIPxEUakj2Pc8U9YHSyyX30XHHgbvroanG1deuXWtBpiH0E2TkVFBSkpKRQXF0dff4khQkpJXV0dFRUVlJSUDPh1IsfGcapJOgex0Z2NEw0YnatoOKJuOxohNkk1DB9t0scxlhoa2oJEsjV7Ve397f9Sj3OmeXdVN0ee2NvtdrKysrTQDwIhBFlZWYO+OoocsW84DMAxmR3dK2ijAbMyZqMp9g1hYeEAkDmBOJyI5iALx43PKZ8YLQ0DRPaRNkGrhX7wDMU5jByxr9oFwD5ZqG2cSCfWprJYGsJQ7I22hqmth3ru62j0zS+0HIekXFXb36Cm2Y5F6PLGmuEhcsS+ehduSzxlMl9n40QDGeOhoUzd72gIC78e8JZizraX9dxnRvXmlYlfVA8qss9MiidmKEsva4aU9957j0suuQSAl19+mXvvvTfosY2NjTzwwAP9fo+f/OQn/OpXvxrwGIMROWJftZOGpAl4sOhFVdFA+vjwtHGSsumwplLgPNq1Ry5AvSH2865Xt7nTuuyubnZElF9/MuF297+qy2WXXcYdd9wRdP9AxX64iBxVrN5NdeICAG3jRAMZ42HHv1XJgXCK7IWgMWkCJZ3HaO90k+Q/f2RG9nOvgy1PQMkZXZ5a0xrZC6rufmUnu443D+lrTh+Tyl2X9l7htaysjAsuuIBFixaxZcsWJk+ezOOPP8706dP5yle+wqpVq/jmN79JZmYmd911Fw6Hg9LSUv7+97+TnJzMG2+8we233052djbz5s3zvu6jjz7Kxo0b+eMf/0hVVRU333wzhw4p++7Pf/4z999/PwcPHmTOnDmce+653Hfffdx33308++yzOBwOrrzySu6++24A7rnnHh5//HGKiorIyclh/vz5Q3qeIFIi+/Z6aK3kRHwJtlhL9DYuiSYyikF6VK/dcIrsgY7UUkrF8Z6raOsPqx65KXlw2zaebJ7FBb/7gE6XWhymI/vhY+/evaxcuZJPP/2U1NRUb8Rts9lYs2YN55xzDj/72c9YvXo1mzdvZsGCBfzmN7/Bbrdz00038corr/Dhhx9SWVkZ8PVvvfVWzjjjDLZt28bmzZuZMWMG9957L6WlpWzdupX77ruPVatWsX//fjZs2MDWrVvZtGkTH3zwAZs2beLpp59my5YtPP/883zyySfDcg4iIwSu2gnAEet4nXYZLZjpl9V71BqLMBJ7V+YkssufY2d9lWoTadJQpmryA9XNdn7+n920dbr5pKyeJROyqI3wyL6vCHw4KSoqYunSpQB86Utf4v777wfg6quvBmD9+vXs2rXLe0xnZydLlixhz549lJSUMGnSJO9zH3rooR6v/8477/D446pNd0xMDGlpaTQ0NHQ5ZtWqVaxatYq5c+cC0Nrayv79+2lpaeHKK68kMVG1Ur3sssuG+tcHIkXsq1UmzgExnhSddhkdZE5Qt2a/3TASe5GrJl47K3fDRD+xrz8MxUpM7n1jD063JC7GwurdVUwrSMXlkTqyHya6py6aj5OSkgC1cOncc8/lqaee6nLc1q1bhyx1VErJnXfeyde+9rUu23/3u9+NSHpqZPgdldshIYMKZ5r266OF1DGqNPDe19XjMBJ7W4HROKZmr2+jywHNxyCjhKYOJ89vPsZ1S8azdGIWb++uprpFLZjRYj88HD16lHXr1gHw1FNPcfrpp3fZv3jxYtauXcuBA6o5Tnt7O/v27WPq1KkcPnyYgwcPep8biLPPPps///nPgJrsbW5uJiUlhZaWFu8x559/Po888gitraqT2bFjx6iurmb58uW88MILdHR00NLSwiuvvDK0v7zByS/2UsLh92HcElodLm3jRAtCwPgl3jaF4ST2qXkltMt44ur2+DY2HAEkZJZQ1ayE/ZSidM6elsfR+nY+3FcLQG6KLoI2HEybNo3HHnuM2bNnU19fz9e//vUu+3Nycnj00Ue59tprmT17NosXL2bPnj3YbDYeeughLr74Yk4//XTGjx8f8PV///vf8+677zJr1izmz5/Pzp07ycrKYunSpcycOZPvfe97nHfeeXzhC19gyZIlzJo1i89+9rO0tLQwb948rr76aubMmcNVV13FsmXLhuUciB7pYaPAggUL5MaNGwf25Jp98KdT4eLfcO6HEynNSebB64Z+JlsThnzyN/jPt9X9m9dA/qzRHY+BxyNZ/ZNzOD3uIIl37FVlHHa/Cs98Eb76NmvtxXzx4Y95euVixmclsuQX7wDq+2vtD87qu+nJScTu3buZNm1a3wcOI2VlZVxyySXs2LFjVMcxWAKdSyHEJinlglCef/JH9gfeUreTzqXF7tI2TjQx/jTf/TCK7C0WwZsJF5PoamD9fx7lgfcOcGTPJrUzZ0oXy6YgLYFvnjmRm5aV8NqtyyJK6DXhxcmvjAdW05E2kRf2QovdqW2caCJ7ihL5MEu9BLjhuhs4/shfEJse4ZedhUxN3sj4tHEQn0JNSzXg8+e/e/6U3l5KM0iKi4tP+qh+KDi5I/vOdihby/ueU/jhC9tp63Tr1bPRhMUC45aAJRZiE0d7NF2YWZhB7oqbWWTZww8XxVLQWUZnpiqlUNPiwBZr0ZljmhHl5Bb741vA7eA1+wwS42IASEvQkX1UcdqtcOYPleEdZlhnXgHAhXHbmCCOUxGrJveqW9TiKV0NUjOSnNyhRfFSmm/Zxeu/3si3L5hEbko8K6bkjPaoNCPJ+CXqJxzJLIGsiRQe/CdCuNhqz2cCZp9ZnXWjGVlO7sge2N1iw4mVqQUpXDW/UJeH1YQXE89B1Kt6Ke/WZwFGZK8/p5oR5qQX+71VatHC1PyUUR6JRhOA0rO9d1fXptPQ1qki+1Qt9icLxcXF1NbWjvYwBs1JL/Z7KltItVnJT9WXxZowpHgpxMRjTy6iAxtrDtTS1OHUkf0oIaXE4/GM9jBGhZPbswf2VrYwNT9VT3ZpwpO4JJj9OazWROLXWfjPpyeAKC2L8PodqrTJUJI/Cy4M3kAE1KKqCy+8kDPPPJN169ZxxRVX8Oqrr/YoM3zFFVdQXl6O3W7ntttuY+XKlUM71lHmpBZ7KSX7Klu4Yu7Y0R6KRhOcy/+EFZh7bB3v7FU59trGGVn27t3L3//+d6644gr+/e9/s2HDBqSUXHbZZXzwwQcsX76cRx55hMzMTDo6Ojj11FO56qqryMrKGu2hDxkntdgfa+ygxeFiivbrNScBi0qyWH+oHoCc5Ci0HfuIwIeT8ePHs3jxYr773e8GLDO8fPly7r//fl544QUAysvL2b9/vxb7cKGx3cmUvBSmj0kd7aFoNH2yaEImvK3u68h+ZPEvZRyozPB7773H6tWrWbduHYmJiaxYsQK73T4aQx02TuoJ2plj03jzW8uZNy68lsprNIGYNy6DuBgLQkBmUtxoDycqCVZmuKmpiYyMDBITE9mzZw/r168f5ZEOPSd1ZK/RnEzYYmM4pSiNQzVtunXmKHHeeeexe/dulixRC/GSk5P5xz/+wQUXXMCDDz7I7NmzmTJlCosXLx7lkQ49fZY4FkI8AlwCVEspZ3bb913gPiBHSllrbLsTuBFwA7dKKd/saxCDKnGs0ZxErD1Qy9H6dq5dOG60hzIihEOJ40hhsCWOQ4nsHwX+CDze7U2KgHOBo37bpgPXADOAMcBqIcRkKaU7lMFoNJHO0onZLB3tQWiikj6vJaWUHwD1AXb9Fvg+4H9pcDnwtJTSIaU8DBwAFg7FQDUajUYzcAZkHAohLgOOSSm3dds1Fij3e1xhbAv0GiuFEBuFEBtramoGMgyNRnMSEA7d8E52huIc9lvshRCJwI+A/wm0O8C2gKOUUj4kpVwgpVyQk6MrVWo0kYjNZqOurk4L/iCQUlJXV4fNNri1GQPJxikFSoBtRomCQmCzEGIhKpIv8ju2EDg+qBFqNJqTlsLCQioqKtBX74PDZrNRWFg4qNfot9hLKbcDueZjIUQZsEBKWSuEeBn4pxDiN6gJ2knAhkGNUKPRnLTExsZSUlIy2sPQEIKNI4R4ClgHTBFCVAghbgx2rJRyJ/AssAt4A7hFZ+JoNBrN6NNnZC+lvLaP/cXdHt8D3DO4YWk0Go1mKNHL+DQajSYK6HMF7YgMQoga4MggXiIbCNdWMnpsA0OPbeCE8/j02AZGsLGNl1KGlM4YFmI/WIQQG0NdMjzS6LENDD22gRPO49NjGxhDMTZt42g0Gk0UoMVeo9FoooBIEfuHRnsAvaDHNjD02AZOOI9Pj21gDHpsEeHZazQajaZ3IiWy12g0Gk0vaLHXaDSaKOCkFnshxAVCiL1CiANCiDtGeSxFQoh3hRC7hRA7hRC3Gdt/IoQ4JoTYavxcNErjKxNCbDfGsNHYlimEeEsIsd+4HZVmvkKIKX7nZ6sQolkIcftonTshxCNCiGohxA6/bUHPlRDiTuMzuFcIcf4ojO0+IcQeIcSnQogXhBDpxvZiIUSH3/l7cBTGFvRvOJLnrZfxPeM3tjIhxFZj+0ifu2D6MXSfOynlSfkDxAAHgQlAHLANmD6K4ykA5hn3U4B9wHTgJ8B3w+B8lQHZ3bb9ErjDuH8H8H9hMM4YoBIYP1rnDlgOzAN29HWujL/xNiAeVQ32IBAzwmM7D7Aa9//Pb2zF/seN0nkL+Dcc6fMWbHzd9v8a+J9ROnfB9GPIPncnc2S/EDggpTwkpewEnkZ1yhoVpJQnpJSbjfstwG6CNG4JIy4HHjPuPwZcMXpD8XI2cFBKOZgV1YNCBu7OFuxcjWh3tkBjk1KuklK6jIfrUaXFR5wg5y0YI97VrrfxCVWv/fPAU8M5hmD0oh9D9rk7mcU+5K5YI40QohiYC3xsbPqmcYn9yGhZJagmMquEEJuEECuNbXlSyhOgPmz4la4eRa6h6z9cOJw7CH6uwu1z+BXgdb/HJUKILUKI94UQy0ZpTIH+huF23pYBVVLK/X7bRuXcddOPIfvcncxiH3JXrJFECJEMPAfcLqVsBv6MavgyBziBulQcDZZKKecBFwK3CCGWj9I4giKEiAMuA/5lbAqXc9cbYfM5FEL8CHABTxqbTgDjpJRzgW+jek2kjvCwgv0Nw+a8GVxL1yBjVM5dAP0IemiAbb2ev5NZ7MOuK5YQIhb1h3pSSvk8gJSySkrpllJ6gL8ySg3YpZTHjdtq4AVjHFVCiAJj7AVA9WiMzY8Lgc1SyioIn3NnEOxchcXnUAhxPXAJ8EVpmLrGJX6dcX8TytedPJLj6uVvGBbnDUAIYQU+AzxjbhuNcxdIPxjCz93JLPafAJOEECVGRHgN8PJoDcbw/P4G7JZS/sZve4HfYVcCO7o/dwTGliSESDHvoyb0dqDO1/XGYdcDL4302LrRJboKh3PnR7Bz9TJwjRAiXghRwih0ZxNCXAD8ALhMStnutz1HCBFj3J9gjO3QCI8t2N9w1M+bH+cAe6SUFeaGkT53wfSDofzcjdRs8zDNYF+EmrU+CPxolMdyOuoy6lNgq/FzEfAEsN3Y/jJQMApjm4Caud8G7DTPFZAFvA3sN24zR/H8JQJ1QJrftlE5d6gvnBOAExVB3djbuQJ+ZHwG9wIXjsLYDqD8W/Nz96Bx7FXG33sbsBm4dBTGFvRvOJLnLdj4jO2PAjd3O3akz10w/Riyz50ul6DRaDRRwMls42g0Go0mRLTYazQaTRSgxV6j0WiiAC32Go1GEwVosddoNJooQIu9RqPRRAFa7DUajSYK+P+Q2WFnqr/ydwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "plt.plot(de_predicted, label = 'predicted')\n",
    "plt.plot(y_test_ori, label = 'real')\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": "3.7.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}