Analysis of COVID-19 behavior along with weather data.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# AI-based Surveying the Impact of Environmental, Climatic, Economic and Demographic Conditions on the Epidemic Outbreak Rate of COVID-19"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$Mehryar{\\,}Majd$, $Ramin{\\,}Mousa$ \\& $Arsalan{\\,}Mousazadeh$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "__Abstract:__ This study aims to have an analytical and predictive approach to the growth rate and spread of the first wave of COVID-19 infection and mainly focuses on its dependency on other global factors in affected countries via Artificial Intelligence-based processing of fatality data/statistics. Fatality, infected, and recovered cases in the form of the time-series data was acquired through the World Health Organization (WHO) & the Johns Hopkins University Center for Systems Science and Engineering (JHU CSSE). Recorded health data subsequently was analyzed by several Deep Neural Network (DNN) models to create a better intuition of epidemic outbreak rate of COVID-19. Monitoring spreading trend changes becomes more feasible via continuous observation of statistical pattern and integrating the other datasets with more global features count as factors influencing the local conditions on historical data conveys pivotal information concerning learning the pattern of epidemic progression. In this study, attention-based Sequence-to-Sequence (Seq2Seq) models empowered by Long-Short Term Memory (LSTM) and Gated Recurrent Unit (GRU) units, were deployed to predict mortality rate. Ultimately these analyses are utilized to monitor infection behaviour on time-series data and were subsequently assessed for their performance to predict the behaviour of the first wave of COVID-19 outbreak rate intensely. Demographic & environmental conditions have likely been influenced significantly to shape the reliable predictive learning algorithms for long-duration risk of COVID-19 pandemic."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Sanitation and Analysis COVID-19 along with __weather data__\n",
    "__Data Cleaning & Preprocessing__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np # linear algebra\n",
    "import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n",
    "import datetime\n",
    "from matplotlib import pyplot as plt\n",
    "import os"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_covid_19 = pd.read_csv(\"novel-corona-virus/covid_19_data.csv\")\n",
    "df_covid_19['Date'] = pd.to_datetime(df_covid_19['ObservationDate'])\n",
    "df_covid_19['Outbreak'] = 'COVID_2019'\n",
    "df_sars_03 = pd.read_csv(\"novel-corona-virus/sars_2003_complete_dataset_clean.csv\")\n",
    "df_sars_03['Date'] = pd.to_datetime(df_sars_03['Date'])\n",
    "df_sars_03['Province/State'] = None\n",
    "df_sars_03['Outbreak'] = 'SARS_2003'\n",
    "df_sars_03.rename({'Cumulative number of case(s)':'Confirmed', 'Number of deaths':'Deaths', 'Number recovered':'Recovered', 'Country':'Country/Region'},axis=1,inplace=True)\n",
    "templ_cols = ['Outbreak', 'Province/State', 'Country/Region', 'Date', 'Confirmed', 'Deaths', 'Recovered']\n",
    "df = pd.concat([df_covid_19[templ_cols], df_sars_03[templ_cols]])\n",
    "df=df_covid_19"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df.reset_index(drop=True)\n",
    "df['Confirmed'] = df['Confirmed'].fillna(0)\n",
    "df['Province/State'] = df['Province/State'].fillna('Others')\n",
    "df = df.sort_values(['Country/Region','Province/State','Date'])\n",
    "t = df.groupby(['Outbreak','Country/Region','Province/State']).agg({'Confirmed':'max'})\n",
    "t = t.loc[t['Confirmed'] >=0]\n",
    "df = pd.merge(df,t,left_on=['Outbreak','Country/Region','Province/State'], right_index=True)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "country_data = pd.read_csv(\"novel-corona-virus/countries of the world.csv\")\n",
    "country_data['Country'] = country_data['Country'].str.strip()\n",
    "df.loc[df['Country/Region']=='US','Country/Region'] = 'United States'\n",
    "df.loc[df['Country/Region']=='Mainland China','Country/Region'] = 'China'\n",
    "df.loc[df['Country/Region']=='Viet Nam','Country/Region'] = 'Vietnam'\n",
    "df.loc[df['Country/Region']=='UK','Country/Region'] = 'United Kingdom'\n",
    "df.loc[df['Country/Region']=='South Korea','Country/Region'] = 'Korea, South'\n",
    "df.loc[df['Country/Region']=='Taiwan, China','Country/Region'] = 'Taiwan'\n",
    "df.loc[df['Country/Region']=='Hong Kong SAR, China','Country/Region'] = 'Hong Kong'\n",
    "df = pd.merge(df, country_data, how='left', left_on=['Country/Region'], right_on=['Country'])\n",
    "df['Date'] = pd.to_datetime(df['Date'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_wthear=pd.read_csv('input/final.txt',sep='\\t',error_bad_lines=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "l=[]\n",
    "for index,row in data_wthear.iterrows():\n",
    "    s=row['updated_at'].split(' ')\n",
    "    l.append(s[0])\n",
    "data_wthear['updated_at']=l\n",
    "data_wthear=data_wthear.dropna()\n",
    "data_wthear['updated_at'] = pd.to_datetime(data_wthear['updated_at'])\n",
    "df_new = pd.merge(df, data_wthear, how='inner', left_on=['Country/Region','Date','Province/State'], right_on=['Country/Region','updated_at','Province/State'])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "df=df_new.copy()\n",
    "df.loc[df['Region'].isnull(), 'Region'] = 'Others'\n",
    "df.loc[df['Country'].isnull(), 'Country'] = 'Undefined'\n",
    "df['Month'] = df['Date'].dt.month\n",
    "df['Week'] = df['Date'].dt.week\n",
    "df = df.fillna(0)\n",
    "df=df.drop('Other (%)',axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "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>SNo</th>\n",
       "      <th>ObservationDate</th>\n",
       "      <th>Province/State</th>\n",
       "      <th>Country/Region</th>\n",
       "      <th>Last Update</th>\n",
       "      <th>Confirmed_x</th>\n",
       "      <th>Deaths</th>\n",
       "      <th>Recovered</th>\n",
       "      <th>Date</th>\n",
       "      <th>Outbreak</th>\n",
       "      <th>...</th>\n",
       "      <th>apparentTemperatureMinTime</th>\n",
       "      <th>apparentTemperatureMax</th>\n",
       "      <th>apparentTemperatureMaxTime</th>\n",
       "      <th>created_at</th>\n",
       "      <th>updated_at</th>\n",
       "      <th>City</th>\n",
       "      <th>Long</th>\n",
       "      <th>lat</th>\n",
       "      <th>Month</th>\n",
       "      <th>Week</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>5445</td>\n",
       "      <td>03/14/2020</td>\n",
       "      <td>New South Wales</td>\n",
       "      <td>Australia</td>\n",
       "      <td>2020-03-14T13:33:04</td>\n",
       "      <td>112.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>2020-03-14</td>\n",
       "      <td>COVID_2019</td>\n",
       "      <td>...</td>\n",
       "      <td>1577639580</td>\n",
       "      <td>105.15</td>\n",
       "      <td>1577681460</td>\n",
       "      <td>3/14/2020 11:01</td>\n",
       "      <td>2020-03-14</td>\n",
       "      <td>Sydney</td>\n",
       "      <td>147.0045</td>\n",
       "      <td>-32.23834</td>\n",
       "      <td>3</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>5445</td>\n",
       "      <td>03/14/2020</td>\n",
       "      <td>New South Wales</td>\n",
       "      <td>Australia</td>\n",
       "      <td>2020-03-14T13:33:04</td>\n",
       "      <td>112.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>2020-03-14</td>\n",
       "      <td>COVID_2019</td>\n",
       "      <td>...</td>\n",
       "      <td>1577725260</td>\n",
       "      <td>105.79</td>\n",
       "      <td>1577761380</td>\n",
       "      <td>3/14/2020 11:01</td>\n",
       "      <td>2020-03-14</td>\n",
       "      <td>Sydney</td>\n",
       "      <td>147.0045</td>\n",
       "      <td>-32.23834</td>\n",
       "      <td>3</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>5445</td>\n",
       "      <td>03/14/2020</td>\n",
       "      <td>New South Wales</td>\n",
       "      <td>Australia</td>\n",
       "      <td>2020-03-14T13:33:04</td>\n",
       "      <td>112.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>2020-03-14</td>\n",
       "      <td>COVID_2019</td>\n",
       "      <td>...</td>\n",
       "      <td>1577821080</td>\n",
       "      <td>99.26</td>\n",
       "      <td>1577854980</td>\n",
       "      <td>3/14/2020 11:01</td>\n",
       "      <td>2020-03-14</td>\n",
       "      <td>Sydney</td>\n",
       "      <td>147.0045</td>\n",
       "      <td>-32.23834</td>\n",
       "      <td>3</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>5445</td>\n",
       "      <td>03/14/2020</td>\n",
       "      <td>New South Wales</td>\n",
       "      <td>Australia</td>\n",
       "      <td>2020-03-14T13:33:04</td>\n",
       "      <td>112.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>2020-03-14</td>\n",
       "      <td>COVID_2019</td>\n",
       "      <td>...</td>\n",
       "      <td>1577898840</td>\n",
       "      <td>102.89</td>\n",
       "      <td>1577941080</td>\n",
       "      <td>3/14/2020 11:01</td>\n",
       "      <td>2020-03-14</td>\n",
       "      <td>Sydney</td>\n",
       "      <td>147.0045</td>\n",
       "      <td>-32.23834</td>\n",
       "      <td>3</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>5445</td>\n",
       "      <td>03/14/2020</td>\n",
       "      <td>New South Wales</td>\n",
       "      <td>Australia</td>\n",
       "      <td>2020-03-14T13:33:04</td>\n",
       "      <td>112.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>2020-03-14</td>\n",
       "      <td>COVID_2019</td>\n",
       "      <td>...</td>\n",
       "      <td>1577984340</td>\n",
       "      <td>110.20</td>\n",
       "      <td>1578027480</td>\n",
       "      <td>3/14/2020 11:02</td>\n",
       "      <td>2020-03-14</td>\n",
       "      <td>Sydney</td>\n",
       "      <td>147.0045</td>\n",
       "      <td>-32.23834</td>\n",
       "      <td>3</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 82 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    SNo ObservationDate   Province/State Country/Region          Last Update  \\\n",
       "0  5445      03/14/2020  New South Wales      Australia  2020-03-14T13:33:04   \n",
       "1  5445      03/14/2020  New South Wales      Australia  2020-03-14T13:33:04   \n",
       "2  5445      03/14/2020  New South Wales      Australia  2020-03-14T13:33:04   \n",
       "3  5445      03/14/2020  New South Wales      Australia  2020-03-14T13:33:04   \n",
       "4  5445      03/14/2020  New South Wales      Australia  2020-03-14T13:33:04   \n",
       "\n",
       "   Confirmed_x  Deaths  Recovered       Date    Outbreak  ...  \\\n",
       "0        112.0     2.0        4.0 2020-03-14  COVID_2019  ...   \n",
       "1        112.0     2.0        4.0 2020-03-14  COVID_2019  ...   \n",
       "2        112.0     2.0        4.0 2020-03-14  COVID_2019  ...   \n",
       "3        112.0     2.0        4.0 2020-03-14  COVID_2019  ...   \n",
       "4        112.0     2.0        4.0 2020-03-14  COVID_2019  ...   \n",
       "\n",
       "   apparentTemperatureMinTime apparentTemperatureMax  \\\n",
       "0                  1577639580                 105.15   \n",
       "1                  1577725260                 105.79   \n",
       "2                  1577821080                  99.26   \n",
       "3                  1577898840                 102.89   \n",
       "4                  1577984340                 110.20   \n",
       "\n",
       "  apparentTemperatureMaxTime       created_at  updated_at    City      Long  \\\n",
       "0                 1577681460  3/14/2020 11:01  2020-03-14  Sydney  147.0045   \n",
       "1                 1577761380  3/14/2020 11:01  2020-03-14  Sydney  147.0045   \n",
       "2                 1577854980  3/14/2020 11:01  2020-03-14  Sydney  147.0045   \n",
       "3                 1577941080  3/14/2020 11:01  2020-03-14  Sydney  147.0045   \n",
       "4                 1578027480  3/14/2020 11:02  2020-03-14  Sydney  147.0045   \n",
       "\n",
       "        lat  Month  Week  \n",
       "0 -32.23834      3    11  \n",
       "1 -32.23834      3    11  \n",
       "2 -32.23834      3    11  \n",
       "3 -32.23834      3    11  \n",
       "4 -32.23834      3    11  \n",
       "\n",
       "[5 rows x 82 columns]"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import locale\n",
    "from fastnumbers import fast_real\n",
    "Literacy=[]\n",
    "Pop,Pho,Agric=[],[],[]\n",
    "Coa,Clim,Bir,Dea=[],[],[],[]\n",
    "Net,Inf,Arab,Crop,Ind,Serv=[],[],[],[],[],[]\n",
    "for index ,row in df.iterrows():\n",
    "    a=str(row['Literacy (%)']).replace(',','.')\n",
    "    Literacy.append(fast_real(a))\n",
    "    #----------\n",
    "    a=str(row['Pop. Density (per sq. mi.)']).replace(',','.')\n",
    "    Pop.append(fast_real(a))\n",
    "    #----------\n",
    "    a=str(row['Coastline (coast/area ratio)']).replace(',','.')\n",
    "    Coa.append(fast_real(a))\n",
    "    #-----------\n",
    "    a=str(row['Net migration']).replace(',','.')\n",
    "    Net.append(fast_real(a))\n",
    "    #-----------\n",
    "    a=str(row['Infant mortality (per 1000 births)']).replace(',','.')\n",
    "    Inf.append(fast_real(a))\n",
    "    #\n",
    "    #\n",
    "    a=str(row['Phones (per 1000)']).replace(',','.')\n",
    "    Pho.append(fast_real(a))\n",
    "    #\n",
    "    a=str(row['Arable (%)']).replace(',','.')\n",
    "    Arab.append(fast_real(a))\n",
    "    #\n",
    "    a=str(row['Crops (%)']).replace(',','.')\n",
    "    Crop.append(fast_real(a))\n",
    "    #\n",
    "    a=str(row['Climate']).replace(',','.')\n",
    "    Clim.append(fast_real(a))\n",
    "    #\n",
    "    a=str(row['Birthrate']).replace(',','.')\n",
    "    Bir.append(fast_real(a))\n",
    "    #\n",
    "    a=str(row['Deathrate']).replace(',','.')\n",
    "    Dea.append(fast_real(a))\n",
    "    #\n",
    "    a=str(row['Agriculture']).replace(',','.')\n",
    "    Agric.append(fast_real(a))\n",
    "    #\n",
    "    a=str(row['Industry']).replace(',','.')\n",
    "    Ind.append(fast_real(a))\n",
    "    #\n",
    "    a=str(row['Service']).replace(',','.')\n",
    "    Serv.append(fast_real(a))\n",
    "df['Literacy (%)']=Literacy\n",
    "df['Pop. Density (per sq. mi.)']=Pop\n",
    "df['Coastline (coast/area ratio)']=Coa\n",
    "df['Net migration']=Net\n",
    "df['Infant mortality (per 1000 births)']=Inf\n",
    "df['Phones (per 1000)']=Pho\n",
    "df['Arable (%)']=Arab\n",
    "df['Crops (%)']=Crop\n",
    "df['Climate']=Clim\n",
    "df['Birthrate']=Bir\n",
    "df['Deathrate']=Dea\n",
    "df['Agriculture']=Agric\n",
    "df['Industry']=Ind\n",
    "df['Service']=Serv\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['Country/Region'] = df['Country/Region'].astype('category')\n",
    "df['Outbreak'] = df['Outbreak'].astype('category')\n",
    "df['precipType'] = df['precipType'].astype('category')\n",
    "df['Country/Region'] = df['Country/Region'].cat.codes\n",
    "df['Outbreak'] = df['Outbreak'].cat.codes\n",
    "df['precipType'] = df['precipType'].cat.codes\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Outbreak</th>\n",
       "      <th>Country/Region</th>\n",
       "      <th>Population</th>\n",
       "      <th>Area (sq. mi.)</th>\n",
       "      <th>Pop. Density (per sq. mi.)</th>\n",
       "      <th>Coastline (coast/area ratio)</th>\n",
       "      <th>Net migration</th>\n",
       "      <th>Confirmed_y</th>\n",
       "      <th>Infant mortality (per 1000 births)</th>\n",
       "      <th>Literacy (%)</th>\n",
       "      <th>...</th>\n",
       "      <th>Birthrate</th>\n",
       "      <th>Deathrate</th>\n",
       "      <th>Agriculture</th>\n",
       "      <th>Industry</th>\n",
       "      <th>temperatureLow</th>\n",
       "      <th>humidity</th>\n",
       "      <th>temperatureHigh</th>\n",
       "      <th>Confirmed_x</th>\n",
       "      <th>Deaths</th>\n",
       "      <th>Recovered</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>20264082.0</td>\n",
       "      <td>7686850.0</td>\n",
       "      <td>2.6</td>\n",
       "      <td>0.34</td>\n",
       "      <td>3.98</td>\n",
       "      <td>171.0</td>\n",
       "      <td>4.69</td>\n",
       "      <td>100.0</td>\n",
       "      <td>...</td>\n",
       "      <td>12.14</td>\n",
       "      <td>7.51</td>\n",
       "      <td>0.038</td>\n",
       "      <td>0.262</td>\n",
       "      <td>79.41</td>\n",
       "      <td>0.11</td>\n",
       "      <td>105.65</td>\n",
       "      <td>112.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>20264082.0</td>\n",
       "      <td>7686850.0</td>\n",
       "      <td>2.6</td>\n",
       "      <td>0.34</td>\n",
       "      <td>3.98</td>\n",
       "      <td>171.0</td>\n",
       "      <td>4.69</td>\n",
       "      <td>100.0</td>\n",
       "      <td>...</td>\n",
       "      <td>12.14</td>\n",
       "      <td>7.51</td>\n",
       "      <td>0.038</td>\n",
       "      <td>0.262</td>\n",
       "      <td>72.64</td>\n",
       "      <td>0.11</td>\n",
       "      <td>106.29</td>\n",
       "      <td>112.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>20264082.0</td>\n",
       "      <td>7686850.0</td>\n",
       "      <td>2.6</td>\n",
       "      <td>0.34</td>\n",
       "      <td>3.98</td>\n",
       "      <td>171.0</td>\n",
       "      <td>4.69</td>\n",
       "      <td>100.0</td>\n",
       "      <td>...</td>\n",
       "      <td>12.14</td>\n",
       "      <td>7.51</td>\n",
       "      <td>0.038</td>\n",
       "      <td>0.262</td>\n",
       "      <td>68.74</td>\n",
       "      <td>0.16</td>\n",
       "      <td>99.76</td>\n",
       "      <td>112.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>20264082.0</td>\n",
       "      <td>7686850.0</td>\n",
       "      <td>2.6</td>\n",
       "      <td>0.34</td>\n",
       "      <td>3.98</td>\n",
       "      <td>171.0</td>\n",
       "      <td>4.69</td>\n",
       "      <td>100.0</td>\n",
       "      <td>...</td>\n",
       "      <td>12.14</td>\n",
       "      <td>7.51</td>\n",
       "      <td>0.038</td>\n",
       "      <td>0.262</td>\n",
       "      <td>71.79</td>\n",
       "      <td>0.16</td>\n",
       "      <td>103.39</td>\n",
       "      <td>112.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>20264082.0</td>\n",
       "      <td>7686850.0</td>\n",
       "      <td>2.6</td>\n",
       "      <td>0.34</td>\n",
       "      <td>3.98</td>\n",
       "      <td>171.0</td>\n",
       "      <td>4.69</td>\n",
       "      <td>100.0</td>\n",
       "      <td>...</td>\n",
       "      <td>12.14</td>\n",
       "      <td>7.51</td>\n",
       "      <td>0.038</td>\n",
       "      <td>0.262</td>\n",
       "      <td>85.01</td>\n",
       "      <td>0.28</td>\n",
       "      <td>110.70</td>\n",
       "      <td>112.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 22 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Outbreak  Country/Region  Population  Area (sq. mi.)  \\\n",
       "0         0               0  20264082.0       7686850.0   \n",
       "1         0               0  20264082.0       7686850.0   \n",
       "2         0               0  20264082.0       7686850.0   \n",
       "3         0               0  20264082.0       7686850.0   \n",
       "4         0               0  20264082.0       7686850.0   \n",
       "\n",
       "   Pop. Density (per sq. mi.)  Coastline (coast/area ratio)  Net migration  \\\n",
       "0                         2.6                          0.34           3.98   \n",
       "1                         2.6                          0.34           3.98   \n",
       "2                         2.6                          0.34           3.98   \n",
       "3                         2.6                          0.34           3.98   \n",
       "4                         2.6                          0.34           3.98   \n",
       "\n",
       "   Confirmed_y  Infant mortality (per 1000 births)  Literacy (%)  ...  \\\n",
       "0        171.0                                4.69         100.0  ...   \n",
       "1        171.0                                4.69         100.0  ...   \n",
       "2        171.0                                4.69         100.0  ...   \n",
       "3        171.0                                4.69         100.0  ...   \n",
       "4        171.0                                4.69         100.0  ...   \n",
       "\n",
       "   Birthrate  Deathrate  Agriculture  Industry  temperatureLow  humidity  \\\n",
       "0      12.14       7.51        0.038     0.262           79.41      0.11   \n",
       "1      12.14       7.51        0.038     0.262           72.64      0.11   \n",
       "2      12.14       7.51        0.038     0.262           68.74      0.16   \n",
       "3      12.14       7.51        0.038     0.262           71.79      0.16   \n",
       "4      12.14       7.51        0.038     0.262           85.01      0.28   \n",
       "\n",
       "   temperatureHigh  Confirmed_x  Deaths  Recovered  \n",
       "0           105.65        112.0     2.0        4.0  \n",
       "1           106.29        112.0     2.0        4.0  \n",
       "2            99.76        112.0     2.0        4.0  \n",
       "3           103.39        112.0     2.0        4.0  \n",
       "4           110.70        112.0     2.0        4.0  \n",
       "\n",
       "[5 rows x 22 columns]"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "col=[ 'Outbreak',  'Country/Region', 'Population','Area (sq. mi.)', 'Pop. Density (per sq. mi.)',\n",
    "       'Coastline (coast/area ratio)', 'Net migration','Confirmed_y',\n",
    "       'Infant mortality (per 1000 births)',  'Literacy (%)', 'Crops (%)',\n",
    "       'Climate', 'Birthrate', 'Deathrate', 'Agriculture', 'Industry','temperatureLow','humidity','temperatureHigh','Confirmed_x','Deaths','Recovered']\n",
    "newdf=df[col]\n",
    "newdf.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "def base_model(feature_len=3, after_day=3, input_shape=(8, 1)):\n",
    "    model = Sequential()\n",
    "    model.add(LSTM(units=100, return_sequences=False, input_shape=input_shape))\n",
    "    #model.add(LSTM(units=100, return_sequences=False, input_shape=input_shape))\n",
    "    # one to many\n",
    "    model.add(RepeatVector(after_day))\n",
    "    #model.add(LSTM(200, return_sequences=True))\n",
    "    model.add(LSTM(50, return_sequences=True))\n",
    "    model.add(TimeDistributed(Dense(units=feature_len, activation='linear')))\n",
    "    return model\n",
    "def normalize_data(data, scaler, feature_len):\n",
    "    minmaxscaler = scaler.fit(data)\n",
    "    normalize_data = minmaxscaler.transform(data)\n",
    "    return normalize_data\n",
    "from random import randint\n",
    "from numpy import array\n",
    "from numpy import argmax\n",
    "from numpy import array_equal\n",
    "from keras.utils import to_categorical\n",
    "from keras.models import Model\n",
    "from keras.layers import Input\n",
    "from keras.layers import LSTM,GRU\n",
    "from keras.layers import Dense\n",
    "def load_data(data, time_step=2, after_day=1, validate_percent=0.67):\n",
    "    seq_length = time_step + after_day\n",
    "    result = []\n",
    "    for index in range(len(data) - seq_length + 1):\n",
    "        result.append(data[index: index + seq_length])\n",
    "    result = np.array(result)\n",
    "    print('total data: ', result.shape)\n",
    "\n",
    "    train_size = int(len(result) * validate_percent)\n",
    "    train = result[:train_size, :]\n",
    "    validate = result[train_size:, :]\n",
    "\n",
    "    x_train = train[:, :time_step]\n",
    "    y_train = train[:, time_step:]\n",
    "    x_validate = validate[:, :time_step]\n",
    "    y_validate = validate[:, time_step:]\n",
    "    \n",
    "     \n",
    "\n",
    "    return [x_train, y_train, x_validate, y_validate]\n",
    "def seq2seq(feature_len=1, after_day=1, input_shape=(8, 1)):\n",
    "    '''\n",
    "    Encoder:\n",
    "    X = Input sequence\n",
    "    C = LSTM(X); The context vector\n",
    "\n",
    "    Decoder:\n",
    "    y(t) = LSTM(s(t-1), y(t-1)); where s is the hidden state of the LSTM(h and c)\n",
    "    y(0) = LSTM(s0, C); C is the context vector from the encoder.\n",
    "    '''\n",
    "\n",
    "    # Encoder\n",
    "    encoder_inputs = Input(shape=input_shape) # (timesteps, feature)\n",
    "    encoder = LSTM(units=100, return_state=True,  name='encoder')\n",
    "    encoder_outputs, state_h, state_c = encoder(encoder_inputs)\n",
    "    states = [state_h, state_c]\n",
    "\n",
    "    # Decoder\n",
    "    reshapor = Reshape((1, 100), name='reshapor')\n",
    "    decoder = LSTM(units=100, return_sequences=True, return_state=True, name='decoder')\n",
    "\n",
    "    # Densor\n",
    "    #tdensor = TimeDistributed(Dense(units=200, activation='linear', name='time_densor'))\n",
    "    densor_output = Dense(units=feature_len, activation='linear', name='output')\n",
    "\n",
    "    inputs = reshapor(encoder_outputs)\n",
    "    #inputs = tdensor(inputs)\n",
    "    all_outputs = []\n",
    "\n",
    "\n",
    "\n",
    "    for _ in range(after_day):\n",
    "        outputs, h, c = decoder(inputs, initial_state=states)\n",
    "\n",
    "        #inputs = tdensor(outputs)\n",
    "        inputs = outputs\n",
    "        states = [state_h, state_c]\n",
    "\n",
    "        outputs = densor_output(outputs)\n",
    "        all_outputs.append(outputs)\n",
    "\n",
    "    decoder_outputs = Lambda(lambda x: K.concatenate(x, axis=1))(all_outputs)\n",
    "    model = Model(inputs=encoder_inputs, outputs=decoder_outputs)\n",
    "\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "total data:  (44320, 6, 22)\n",
      "train data:  (35456, 2, 22) (35456, 4, 22)\n",
      "validate data:  (8864, 2, 22) (8864, 4, 22)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.preprocessing import MinMaxScaler\n",
    "scaler = MinMaxScaler(feature_range=(0, 1))\n",
    "data = normalize_data(newdf, scaler,df.shape[1])\n",
    "x_train, y_train, x_validate, y_validate = load_data(data,time_step=2, after_day=4, validate_percent=0.8)\n",
    "print('train data: ', x_train.shape, y_train.shape)\n",
    "print('validate data: ', x_validate.shape, y_validate.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras import backend as K\n",
    "from keras.models import Sequential, load_model, Model\n",
    "from keras.layers import LSTM, Dense, Activation, TimeDistributed, Dropout, Lambda, RepeatVector, Input, Reshape\n",
    "from keras.callbacks import ModelCheckpoint"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /home/rad/anaconda3/envs/tensorflow/lib/python3.7/site-packages/tensorflow/python/framework/op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Colocations handled automatically by placer.\n"
     ]
    }
   ],
   "source": [
    " # model complie\n",
    "from keras.optimizers import SGD,Adam\n",
    "# root mean squared error (rmse) for regression (only for Keras tensors)\n",
    "def rmse(y_true, y_pred):\n",
    "    from keras import backend\n",
    "    return backend.sqrt(backend.mean(backend.square(y_pred - y_true), axis=-1))\n",
    "\n",
    "# mean squared error (mse) for regression  (only for Keras tensors)\n",
    "def mse(y_true, y_pred):\n",
    "    from keras import backend\n",
    "    return backend.mean(backend.square(y_pred - y_true), axis=-1)\n",
    "\n",
    "# coefficient of determination (R^2) for regression  (only for Keras tensors)\n",
    "def r_square(y_true, y_pred):\n",
    "    from keras import backend as K\n",
    "    SS_res =  K.sum(K.square(y_true - y_pred)) \n",
    "    SS_tot = K.sum(K.square(y_true - K.mean(y_true))) \n",
    "    return ( 1 - SS_res/(SS_tot + K.epsilon()) )\n",
    "input_shape = (2, data.shape[1])\n",
    "model = base_model(data.shape[1], 4, input_shape)\n",
    "model.compile(loss='mse', optimizer='adam',metrics=['acc',rmse,mse,r_square])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /home/rad/anaconda3/envs/tensorflow/lib/python3.7/site-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use tf.cast instead.\n",
      "Train on 35456 samples, validate on 8864 samples\n",
      "Epoch 1/100\n",
      "35456/35456 [==============================] - 11s 304us/step - loss: 0.0736 - acc: 0.0072 - rmse: 0.2242 - mse: 0.0736 - r_square: 0.5028 - val_loss: 0.0385 - val_acc: 0.0000e+00 - val_rmse: 0.1442 - val_mse: 0.0385 - val_r_square: 0.7402\n",
      "Epoch 2/100\n",
      "35456/35456 [==============================] - 5s 142us/step - loss: 0.0206 - acc: 0.1016 - rmse: 0.1251 - mse: 0.0206 - r_square: 0.8634 - val_loss: 0.0315 - val_acc: 0.0000e+00 - val_rmse: 0.1315 - val_mse: 0.0315 - val_r_square: 0.7872\n",
      "Epoch 3/100\n",
      "35456/35456 [==============================] - 4s 126us/step - loss: 0.0131 - acc: 0.1591 - rmse: 0.0993 - mse: 0.0131 - r_square: 0.9130 - val_loss: 0.0259 - val_acc: 0.1947 - val_rmse: 0.1089 - val_mse: 0.0259 - val_r_square: 0.8227\n",
      "Epoch 4/100\n",
      "35456/35456 [==============================] - 5s 141us/step - loss: 0.0071 - acc: 0.2215 - rmse: 0.0732 - mse: 0.0071 - r_square: 0.9519 - val_loss: 0.0228 - val_acc: 0.0000e+00 - val_rmse: 0.1037 - val_mse: 0.0228 - val_r_square: 0.8446\n",
      "Epoch 5/100\n",
      "35456/35456 [==============================] - 5s 133us/step - loss: 0.0065 - acc: 0.2198 - rmse: 0.0712 - mse: 0.0065 - r_square: 0.9568 - val_loss: 0.0234 - val_acc: 0.0000e+00 - val_rmse: 0.1092 - val_mse: 0.0234 - val_r_square: 0.8413\n",
      "Epoch 6/100\n",
      "35456/35456 [==============================] - 5s 133us/step - loss: 0.0063 - acc: 0.0995 - rmse: 0.0706 - mse: 0.0063 - r_square: 0.9582 - val_loss: 0.0263 - val_acc: 0.0000e+00 - val_rmse: 0.1121 - val_mse: 0.0263 - val_r_square: 0.8207\n",
      "Epoch 7/100\n",
      "35456/35456 [==============================] - 7s 187us/step - loss: 0.0069 - acc: 0.1773 - rmse: 0.0711 - mse: 0.0069 - r_square: 0.9532 - val_loss: 0.0281 - val_acc: 0.0274 - val_rmse: 0.1081 - val_mse: 0.0281 - val_r_square: 0.8074\n",
      "Epoch 8/100\n",
      "35456/35456 [==============================] - 6s 182us/step - loss: 0.0074 - acc: 0.1278 - rmse: 0.0703 - mse: 0.0074 - r_square: 0.9484 - val_loss: 0.0218 - val_acc: 0.0274 - val_rmse: 0.0987 - val_mse: 0.0218 - val_r_square: 0.8513\n",
      "Epoch 9/100\n",
      "35456/35456 [==============================] - 6s 164us/step - loss: 0.0051 - acc: 0.1050 - rmse: 0.0636 - mse: 0.0051 - r_square: 0.9660 - val_loss: 0.0199 - val_acc: 0.0274 - val_rmse: 0.0950 - val_mse: 0.0199 - val_r_square: 0.8642\n",
      "Epoch 10/100\n",
      "35456/35456 [==============================] - 6s 162us/step - loss: 0.0049 - acc: 0.1355 - rmse: 0.0628 - mse: 0.0049 - r_square: 0.9679 - val_loss: 0.0196 - val_acc: 0.0000e+00 - val_rmse: 0.0945 - val_mse: 0.0196 - val_r_square: 0.8665\n",
      "Epoch 11/100\n",
      "35456/35456 [==============================] - 6s 163us/step - loss: 0.0046 - acc: 0.1597 - rmse: 0.0615 - mse: 0.0046 - r_square: 0.9697 - val_loss: 0.0189 - val_acc: 0.0000e+00 - val_rmse: 0.0929 - val_mse: 0.0189 - val_r_square: 0.8712\n",
      "Epoch 12/100\n",
      "35456/35456 [==============================] - 6s 161us/step - loss: 0.0044 - acc: 0.1532 - rmse: 0.0604 - mse: 0.0044 - r_square: 0.9709 - val_loss: 0.0186 - val_acc: 5.6408e-05 - val_rmse: 0.0916 - val_mse: 0.0186 - val_r_square: 0.8732\n",
      "Epoch 13/100\n",
      "35456/35456 [==============================] - 6s 157us/step - loss: 0.0041 - acc: 0.1500 - rmse: 0.0579 - mse: 0.0041 - r_square: 0.9734 - val_loss: 0.0185 - val_acc: 0.0096 - val_rmse: 0.0910 - val_mse: 0.0185 - val_r_square: 0.8739\n",
      "Epoch 14/100\n",
      "35456/35456 [==============================] - 5s 137us/step - loss: 0.0039 - acc: 0.2064 - rmse: 0.0564 - mse: 0.0039 - r_square: 0.9748 - val_loss: 0.0183 - val_acc: 0.0057 - val_rmse: 0.0908 - val_mse: 0.0183 - val_r_square: 0.8750\n",
      "Epoch 15/100\n",
      "35456/35456 [==============================] - 6s 155us/step - loss: 0.0037 - acc: 0.1593 - rmse: 0.0552 - mse: 0.0037 - r_square: 0.9755 - val_loss: 0.0198 - val_acc: 0.0137 - val_rmse: 0.0925 - val_mse: 0.0198 - val_r_square: 0.8646\n",
      "Epoch 16/100\n",
      "35456/35456 [==============================] - 8s 221us/step - loss: 0.0036 - acc: 0.2180 - rmse: 0.0544 - mse: 0.0036 - r_square: 0.9760 - val_loss: 0.0182 - val_acc: 0.0060 - val_rmse: 0.0895 - val_mse: 0.0182 - val_r_square: 0.8760\n",
      "Epoch 17/100\n",
      "35456/35456 [==============================] - 8s 214us/step - loss: 0.0035 - acc: 0.2262 - rmse: 0.0529 - mse: 0.0035 - r_square: 0.9771 - val_loss: 0.0190 - val_acc: 0.0255 - val_rmse: 0.0896 - val_mse: 0.0190 - val_r_square: 0.8702\n",
      "Epoch 18/100\n",
      "35456/35456 [==============================] - 9s 241us/step - loss: 0.0034 - acc: 0.2499 - rmse: 0.0526 - mse: 0.0034 - r_square: 0.9777 - val_loss: 0.0195 - val_acc: 0.0107 - val_rmse: 0.0900 - val_mse: 0.0195 - val_r_square: 0.8662\n",
      "Epoch 19/100\n",
      "35456/35456 [==============================] - 8s 221us/step - loss: 0.0031 - acc: 0.2621 - rmse: 0.0498 - mse: 0.0031 - r_square: 0.9799 - val_loss: 0.0199 - val_acc: 0.0089 - val_rmse: 0.0896 - val_mse: 0.0199 - val_r_square: 0.8633\n",
      "Epoch 20/100\n",
      "35456/35456 [==============================] - 8s 217us/step - loss: 0.0040 - acc: 0.2237 - rmse: 0.0556 - mse: 0.0040 - r_square: 0.9729 - val_loss: 0.0266 - val_acc: 0.0309 - val_rmse: 0.1000 - val_mse: 0.0266 - val_r_square: 0.8170\n",
      "Epoch 21/100\n",
      "35456/35456 [==============================] - 8s 232us/step - loss: 0.0034 - acc: 0.2231 - rmse: 0.0502 - mse: 0.0034 - r_square: 0.9771 - val_loss: 0.0189 - val_acc: 0.0263 - val_rmse: 0.0877 - val_mse: 0.0189 - val_r_square: 0.8707\n",
      "Epoch 22/100\n",
      "35456/35456 [==============================] - 8s 233us/step - loss: 0.0036 - acc: 0.2416 - rmse: 0.0523 - mse: 0.0036 - r_square: 0.9757 - val_loss: 0.0202 - val_acc: 0.0263 - val_rmse: 0.0898 - val_mse: 0.0202 - val_r_square: 0.8618\n",
      "Epoch 23/100\n",
      "35456/35456 [==============================] - 8s 238us/step - loss: 0.0029 - acc: 0.2444 - rmse: 0.0486 - mse: 0.0029 - r_square: 0.9812 - val_loss: 0.0182 - val_acc: 0.0551 - val_rmse: 0.0868 - val_mse: 0.0182 - val_r_square: 0.8754\n",
      "Epoch 24/100\n",
      "35456/35456 [==============================] - 9s 243us/step - loss: 0.0028 - acc: 0.2332 - rmse: 0.0476 - mse: 0.0028 - r_square: 0.9822 - val_loss: 0.0191 - val_acc: 0.0714 - val_rmse: 0.0879 - val_mse: 0.0191 - val_r_square: 0.8691\n",
      "Epoch 25/100\n",
      "35456/35456 [==============================] - 7s 202us/step - loss: 0.0026 - acc: 0.2219 - rmse: 0.0463 - mse: 0.0026 - r_square: 0.9830 - val_loss: 0.0181 - val_acc: 0.0701 - val_rmse: 0.0864 - val_mse: 0.0181 - val_r_square: 0.8765\n",
      "Epoch 26/100\n",
      "35456/35456 [==============================] - 6s 182us/step - loss: 0.0025 - acc: 0.2210 - rmse: 0.0456 - mse: 0.0025 - r_square: 0.9836 - val_loss: 0.0192 - val_acc: 0.0737 - val_rmse: 0.0877 - val_mse: 0.0192 - val_r_square: 0.8690\n",
      "Epoch 27/100\n",
      "35456/35456 [==============================] - 6s 177us/step - loss: 0.0025 - acc: 0.1884 - rmse: 0.0455 - mse: 0.0025 - r_square: 0.9835 - val_loss: 0.0180 - val_acc: 0.0806 - val_rmse: 0.0866 - val_mse: 0.0180 - val_r_square: 0.8775\n",
      "Epoch 28/100\n",
      "35456/35456 [==============================] - 6s 162us/step - loss: 0.0026 - acc: 0.1951 - rmse: 0.0457 - mse: 0.0026 - r_square: 0.9833 - val_loss: 0.0197 - val_acc: 0.1070 - val_rmse: 0.0894 - val_mse: 0.0197 - val_r_square: 0.8657\n",
      "Epoch 29/100\n",
      "35456/35456 [==============================] - 9s 248us/step - loss: 0.0027 - acc: 0.1492 - rmse: 0.0466 - mse: 0.0027 - r_square: 0.9824 - val_loss: 0.0176 - val_acc: 0.1177 - val_rmse: 0.0878 - val_mse: 0.0176 - val_r_square: 0.8798\n",
      "Epoch 30/100\n",
      "35456/35456 [==============================] - 6s 175us/step - loss: 0.0027 - acc: 0.1996 - rmse: 0.0464 - mse: 0.0027 - r_square: 0.9826 - val_loss: 0.0211 - val_acc: 0.1191 - val_rmse: 0.0924 - val_mse: 0.0211 - val_r_square: 0.8558\n",
      "Epoch 31/100\n",
      "35456/35456 [==============================] - 9s 245us/step - loss: 0.0028 - acc: 0.1496 - rmse: 0.0469 - mse: 0.0028 - r_square: 0.9818 - val_loss: 0.0176 - val_acc: 0.1187 - val_rmse: 0.0880 - val_mse: 0.0176 - val_r_square: 0.8803\n",
      "Epoch 32/100\n",
      "35456/35456 [==============================] - 8s 225us/step - loss: 0.0027 - acc: 0.2158 - rmse: 0.0462 - mse: 0.0027 - r_square: 0.9820 - val_loss: 0.0243 - val_acc: 0.1647 - val_rmse: 0.0947 - val_mse: 0.0243 - val_r_square: 0.8332\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 33/100\n",
      "35456/35456 [==============================] - 8s 229us/step - loss: 0.0026 - acc: 0.1827 - rmse: 0.0448 - mse: 0.0026 - r_square: 0.9824 - val_loss: 0.0181 - val_acc: 0.1514 - val_rmse: 0.0874 - val_mse: 0.0181 - val_r_square: 0.8767\n",
      "Epoch 34/100\n",
      "35456/35456 [==============================] - 8s 229us/step - loss: 0.0027 - acc: 0.2586 - rmse: 0.0451 - mse: 0.0027 - r_square: 0.9822 - val_loss: 0.0218 - val_acc: 0.1869 - val_rmse: 0.0924 - val_mse: 0.0218 - val_r_square: 0.8509\n",
      "Epoch 35/100\n",
      "35456/35456 [==============================] - 7s 186us/step - loss: 0.0024 - acc: 0.1792 - rmse: 0.0432 - mse: 0.0024 - r_square: 0.9844 - val_loss: 0.0186 - val_acc: 0.1874 - val_rmse: 0.0888 - val_mse: 0.0186 - val_r_square: 0.8731\n",
      "Epoch 36/100\n",
      "35456/35456 [==============================] - 6s 163us/step - loss: 0.0023 - acc: 0.1760 - rmse: 0.0429 - mse: 0.0023 - r_square: 0.9848 - val_loss: 0.0206 - val_acc: 0.2721 - val_rmse: 0.0915 - val_mse: 0.0206 - val_r_square: 0.8591\n",
      "Epoch 37/100\n",
      "35456/35456 [==============================] - 6s 179us/step - loss: 0.0023 - acc: 0.1841 - rmse: 0.0425 - mse: 0.0023 - r_square: 0.9851 - val_loss: 0.0188 - val_acc: 0.3035 - val_rmse: 0.0889 - val_mse: 0.0188 - val_r_square: 0.8717\n",
      "Epoch 38/100\n",
      "35456/35456 [==============================] - 6s 173us/step - loss: 0.0023 - acc: 0.1971 - rmse: 0.0424 - mse: 0.0023 - r_square: 0.9851 - val_loss: 0.0207 - val_acc: 0.3517 - val_rmse: 0.0915 - val_mse: 0.0207 - val_r_square: 0.8584\n",
      "Epoch 39/100\n",
      "35456/35456 [==============================] - 6s 178us/step - loss: 0.0023 - acc: 0.2158 - rmse: 0.0421 - mse: 0.0023 - r_square: 0.9853 - val_loss: 0.0190 - val_acc: 0.2841 - val_rmse: 0.0890 - val_mse: 0.0190 - val_r_square: 0.8703\n",
      "Epoch 40/100\n",
      "35456/35456 [==============================] - 6s 172us/step - loss: 0.0023 - acc: 0.2334 - rmse: 0.0424 - mse: 0.0023 - r_square: 0.9850 - val_loss: 0.0214 - val_acc: 0.2833 - val_rmse: 0.0926 - val_mse: 0.0214 - val_r_square: 0.8534\n",
      "Epoch 41/100\n",
      "35456/35456 [==============================] - 7s 193us/step - loss: 0.0023 - acc: 0.1805 - rmse: 0.0419 - mse: 0.0023 - r_square: 0.9852 - val_loss: 0.0198 - val_acc: 0.2064 - val_rmse: 0.0898 - val_mse: 0.0198 - val_r_square: 0.8651\n",
      "Epoch 42/100\n",
      "35456/35456 [==============================] - 6s 159us/step - loss: 0.0023 - acc: 0.2067 - rmse: 0.0422 - mse: 0.0023 - r_square: 0.9848 - val_loss: 0.0209 - val_acc: 0.2415 - val_rmse: 0.0917 - val_mse: 0.0209 - val_r_square: 0.8569\n",
      "Epoch 43/100\n",
      "35456/35456 [==============================] - 6s 163us/step - loss: 0.0023 - acc: 0.1765 - rmse: 0.0419 - mse: 0.0023 - r_square: 0.9853 - val_loss: 0.0202 - val_acc: 0.1649 - val_rmse: 0.0899 - val_mse: 0.0202 - val_r_square: 0.8619\n",
      "Epoch 44/100\n",
      "35456/35456 [==============================] - 6s 170us/step - loss: 0.0023 - acc: 0.1758 - rmse: 0.0419 - mse: 0.0023 - r_square: 0.9852 - val_loss: 0.0199 - val_acc: 0.1780 - val_rmse: 0.0896 - val_mse: 0.0199 - val_r_square: 0.8638\n",
      "Epoch 45/100\n",
      "35456/35456 [==============================] - 6s 172us/step - loss: 0.0022 - acc: 0.1718 - rmse: 0.0414 - mse: 0.0022 - r_square: 0.9859 - val_loss: 0.0200 - val_acc: 0.1443 - val_rmse: 0.0892 - val_mse: 0.0200 - val_r_square: 0.8632\n",
      "Epoch 46/100\n",
      "35456/35456 [==============================] - 6s 164us/step - loss: 0.0021 - acc: 0.1524 - rmse: 0.0412 - mse: 0.0021 - r_square: 0.9860 - val_loss: 0.0195 - val_acc: 0.1781 - val_rmse: 0.0881 - val_mse: 0.0195 - val_r_square: 0.8670\n",
      "Epoch 47/100\n",
      "35456/35456 [==============================] - 7s 207us/step - loss: 0.0021 - acc: 0.1540 - rmse: 0.0408 - mse: 0.0021 - r_square: 0.9863 - val_loss: 0.0199 - val_acc: 0.1860 - val_rmse: 0.0883 - val_mse: 0.0199 - val_r_square: 0.8639\n",
      "Epoch 48/100\n",
      "35456/35456 [==============================] - 5s 146us/step - loss: 0.0021 - acc: 0.1420 - rmse: 0.0406 - mse: 0.0021 - r_square: 0.9864 - val_loss: 0.0192 - val_acc: 0.2672 - val_rmse: 0.0872 - val_mse: 0.0192 - val_r_square: 0.8687\n",
      "Epoch 49/100\n",
      "35456/35456 [==============================] - 7s 208us/step - loss: 0.0021 - acc: 0.1411 - rmse: 0.0402 - mse: 0.0021 - r_square: 0.9867 - val_loss: 0.0196 - val_acc: 0.2650 - val_rmse: 0.0877 - val_mse: 0.0196 - val_r_square: 0.8657\n",
      "Epoch 50/100\n",
      "35456/35456 [==============================] - 6s 169us/step - loss: 0.0021 - acc: 0.1776 - rmse: 0.0402 - mse: 0.0021 - r_square: 0.9866 - val_loss: 0.0195 - val_acc: 0.3162 - val_rmse: 0.0875 - val_mse: 0.0195 - val_r_square: 0.8661\n",
      "Epoch 51/100\n",
      "35456/35456 [==============================] - 6s 163us/step - loss: 0.0020 - acc: 0.1682 - rmse: 0.0399 - mse: 0.0020 - r_square: 0.9867 - val_loss: 0.0193 - val_acc: 0.2895 - val_rmse: 0.0873 - val_mse: 0.0193 - val_r_square: 0.8682\n",
      "Epoch 52/100\n",
      "35456/35456 [==============================] - 6s 159us/step - loss: 0.0021 - acc: 0.2092 - rmse: 0.0404 - mse: 0.0021 - r_square: 0.9863 - val_loss: 0.0204 - val_acc: 0.3483 - val_rmse: 0.0893 - val_mse: 0.0204 - val_r_square: 0.8600\n",
      "Epoch 53/100\n",
      "35456/35456 [==============================] - 7s 190us/step - loss: 0.0021 - acc: 0.1929 - rmse: 0.0405 - mse: 0.0021 - r_square: 0.9862 - val_loss: 0.0194 - val_acc: 0.2717 - val_rmse: 0.0882 - val_mse: 0.0194 - val_r_square: 0.8675\n",
      "Epoch 54/100\n",
      "35456/35456 [==============================] - 7s 184us/step - loss: 0.0022 - acc: 0.2129 - rmse: 0.0412 - mse: 0.0022 - r_square: 0.9857 - val_loss: 0.0204 - val_acc: 0.1621 - val_rmse: 0.0912 - val_mse: 0.0204 - val_r_square: 0.8604\n",
      "Epoch 55/100\n",
      "35456/35456 [==============================] - 8s 221us/step - loss: 0.0023 - acc: 0.2057 - rmse: 0.0415 - mse: 0.0023 - r_square: 0.9854 - val_loss: 0.0198 - val_acc: 0.1176 - val_rmse: 0.0916 - val_mse: 0.0198 - val_r_square: 0.8646\n",
      "Epoch 56/100\n",
      "35456/35456 [==============================] - 7s 192us/step - loss: 0.0023 - acc: 0.1918 - rmse: 0.0420 - mse: 0.0023 - r_square: 0.9850 - val_loss: 0.0203 - val_acc: 0.1106 - val_rmse: 0.0931 - val_mse: 0.0203 - val_r_square: 0.8615\n",
      "Epoch 57/100\n",
      "35456/35456 [==============================] - 6s 173us/step - loss: 0.0023 - acc: 0.2070 - rmse: 0.0418 - mse: 0.0023 - r_square: 0.9851 - val_loss: 0.0209 - val_acc: 0.1766 - val_rmse: 0.0931 - val_mse: 0.0209 - val_r_square: 0.8568\n",
      "Epoch 58/100\n",
      "35456/35456 [==============================] - 5s 153us/step - loss: 0.0023 - acc: 0.1139 - rmse: 0.0418 - mse: 0.0023 - r_square: 0.9852 - val_loss: 0.0191 - val_acc: 0.4130 - val_rmse: 0.0904 - val_mse: 0.0191 - val_r_square: 0.8696\n",
      "Epoch 59/100\n",
      "35456/35456 [==============================] - 6s 163us/step - loss: 0.0024 - acc: 0.2361 - rmse: 0.0427 - mse: 0.0024 - r_square: 0.9840 - val_loss: 0.0254 - val_acc: 0.7313 - val_rmse: 0.0974 - val_mse: 0.0254 - val_r_square: 0.8252\n",
      "Epoch 60/100\n",
      "35456/35456 [==============================] - 6s 157us/step - loss: 0.0025 - acc: 0.2295 - rmse: 0.0435 - mse: 0.0025 - r_square: 0.9837 - val_loss: 0.0193 - val_acc: 0.7145 - val_rmse: 0.0928 - val_mse: 0.0193 - val_r_square: 0.8686\n",
      "Epoch 61/100\n",
      "35456/35456 [==============================] - 6s 171us/step - loss: 0.0028 - acc: 0.2204 - rmse: 0.0453 - mse: 0.0028 - r_square: 0.9815 - val_loss: 0.0213 - val_acc: 0.8260 - val_rmse: 0.0954 - val_mse: 0.0213 - val_r_square: 0.8543\n",
      "Epoch 62/100\n",
      "35456/35456 [==============================] - 5s 153us/step - loss: 0.0026 - acc: 0.2400 - rmse: 0.0442 - mse: 0.0026 - r_square: 0.9829 - val_loss: 0.0214 - val_acc: 0.7675 - val_rmse: 0.0959 - val_mse: 0.0214 - val_r_square: 0.8542\n",
      "Epoch 63/100\n",
      "35456/35456 [==============================] - 6s 155us/step - loss: 0.0028 - acc: 0.2458 - rmse: 0.0455 - mse: 0.0028 - r_square: 0.9816 - val_loss: 0.0222 - val_acc: 0.4146 - val_rmse: 0.0943 - val_mse: 0.0222 - val_r_square: 0.8475\n",
      "Epoch 64/100\n",
      "35456/35456 [==============================] - 5s 150us/step - loss: 0.0023 - acc: 0.2063 - rmse: 0.0419 - mse: 0.0023 - r_square: 0.9851 - val_loss: 0.0208 - val_acc: 0.3146 - val_rmse: 0.0922 - val_mse: 0.0208 - val_r_square: 0.8572\n",
      "Epoch 65/100\n",
      "35456/35456 [==============================] - 5s 151us/step - loss: 0.0021 - acc: 0.2471 - rmse: 0.0403 - mse: 0.0021 - r_square: 0.9864 - val_loss: 0.0204 - val_acc: 0.6438 - val_rmse: 0.0893 - val_mse: 0.0204 - val_r_square: 0.8604\n",
      "Epoch 66/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "35456/35456 [==============================] - 5s 153us/step - loss: 0.0020 - acc: 0.2659 - rmse: 0.0396 - mse: 0.0020 - r_square: 0.9870 - val_loss: 0.0203 - val_acc: 0.3445 - val_rmse: 0.0905 - val_mse: 0.0203 - val_r_square: 0.8611\n",
      "Epoch 67/100\n",
      "35456/35456 [==============================] - 5s 154us/step - loss: 0.0019 - acc: 0.2782 - rmse: 0.0387 - mse: 0.0019 - r_square: 0.9875 - val_loss: 0.0199 - val_acc: 0.1782 - val_rmse: 0.0883 - val_mse: 0.0199 - val_r_square: 0.8633\n",
      "Epoch 68/100\n",
      "35456/35456 [==============================] - 5s 153us/step - loss: 0.0019 - acc: 0.2409 - rmse: 0.0382 - mse: 0.0019 - r_square: 0.9878 - val_loss: 0.0199 - val_acc: 0.1929 - val_rmse: 0.0898 - val_mse: 0.0199 - val_r_square: 0.8635\n",
      "Epoch 69/100\n",
      "35456/35456 [==============================] - 5s 147us/step - loss: 0.0019 - acc: 0.2457 - rmse: 0.0379 - mse: 0.0019 - r_square: 0.9880 - val_loss: 0.0199 - val_acc: 0.1193 - val_rmse: 0.0883 - val_mse: 0.0199 - val_r_square: 0.8634\n",
      "Epoch 70/100\n",
      "35456/35456 [==============================] - 5s 149us/step - loss: 0.0018 - acc: 0.1893 - rmse: 0.0373 - mse: 0.0018 - r_square: 0.9883 - val_loss: 0.0197 - val_acc: 0.1837 - val_rmse: 0.0892 - val_mse: 0.0197 - val_r_square: 0.8648\n",
      "Epoch 71/100\n",
      "35456/35456 [==============================] - 5s 153us/step - loss: 0.0018 - acc: 0.1963 - rmse: 0.0376 - mse: 0.0018 - r_square: 0.9880 - val_loss: 0.0201 - val_acc: 0.1931 - val_rmse: 0.0893 - val_mse: 0.0201 - val_r_square: 0.8619\n",
      "Epoch 72/100\n",
      "35456/35456 [==============================] - 6s 156us/step - loss: 0.0018 - acc: 0.1581 - rmse: 0.0374 - mse: 0.0018 - r_square: 0.9880 - val_loss: 0.0195 - val_acc: 0.2222 - val_rmse: 0.0904 - val_mse: 0.0195 - val_r_square: 0.8665\n",
      "Epoch 73/100\n",
      "35456/35456 [==============================] - 6s 162us/step - loss: 0.0020 - acc: 0.1646 - rmse: 0.0387 - mse: 0.0020 - r_square: 0.9869 - val_loss: 0.0205 - val_acc: 0.3552 - val_rmse: 0.0919 - val_mse: 0.0205 - val_r_square: 0.8599\n",
      "Epoch 74/100\n",
      "35456/35456 [==============================] - 5s 154us/step - loss: 0.0020 - acc: 0.1604 - rmse: 0.0385 - mse: 0.0020 - r_square: 0.9870 - val_loss: 0.0187 - val_acc: 0.2681 - val_rmse: 0.0893 - val_mse: 0.0187 - val_r_square: 0.8721\n",
      "Epoch 75/100\n",
      "35456/35456 [==============================] - 5s 155us/step - loss: 0.0020 - acc: 0.1633 - rmse: 0.0389 - mse: 0.0020 - r_square: 0.9868 - val_loss: 0.0201 - val_acc: 0.2651 - val_rmse: 0.0906 - val_mse: 0.0201 - val_r_square: 0.8627\n",
      "Epoch 76/100\n",
      "35456/35456 [==============================] - 6s 157us/step - loss: 0.0019 - acc: 0.1328 - rmse: 0.0375 - mse: 0.0019 - r_square: 0.9876 - val_loss: 0.0176 - val_acc: 0.2102 - val_rmse: 0.0856 - val_mse: 0.0176 - val_r_square: 0.8795\n",
      "Epoch 77/100\n",
      "35456/35456 [==============================] - 5s 154us/step - loss: 0.0018 - acc: 0.1444 - rmse: 0.0374 - mse: 0.0018 - r_square: 0.9882 - val_loss: 0.0195 - val_acc: 0.1126 - val_rmse: 0.0884 - val_mse: 0.0195 - val_r_square: 0.8661\n",
      "Epoch 78/100\n",
      "35456/35456 [==============================] - 5s 153us/step - loss: 0.0017 - acc: 0.0928 - rmse: 0.0363 - mse: 0.0017 - r_square: 0.9890 - val_loss: 0.0173 - val_acc: 0.0969 - val_rmse: 0.0811 - val_mse: 0.0173 - val_r_square: 0.8813\n",
      "Epoch 79/100\n",
      "35456/35456 [==============================] - 5s 155us/step - loss: 0.0017 - acc: 0.1988 - rmse: 0.0368 - mse: 0.0017 - r_square: 0.9887 - val_loss: 0.0195 - val_acc: 0.1117 - val_rmse: 0.0839 - val_mse: 0.0195 - val_r_square: 0.8660\n",
      "Epoch 80/100\n",
      "35456/35456 [==============================] - 6s 155us/step - loss: 0.0017 - acc: 0.0937 - rmse: 0.0360 - mse: 0.0017 - r_square: 0.9891 - val_loss: 0.0179 - val_acc: 0.0262 - val_rmse: 0.0823 - val_mse: 0.0179 - val_r_square: 0.8775\n",
      "Epoch 81/100\n",
      "35456/35456 [==============================] - 5s 152us/step - loss: 0.0017 - acc: 0.1512 - rmse: 0.0366 - mse: 0.0017 - r_square: 0.9888 - val_loss: 0.0203 - val_acc: 0.0736 - val_rmse: 0.0844 - val_mse: 0.0203 - val_r_square: 0.8609\n",
      "Epoch 82/100\n",
      "35456/35456 [==============================] - 5s 152us/step - loss: 0.0017 - acc: 0.0860 - rmse: 0.0357 - mse: 0.0017 - r_square: 0.9891 - val_loss: 0.0183 - val_acc: 0.0674 - val_rmse: 0.0804 - val_mse: 0.0183 - val_r_square: 0.8744\n",
      "Epoch 83/100\n",
      "35456/35456 [==============================] - 6s 175us/step - loss: 0.0016 - acc: 0.1699 - rmse: 0.0347 - mse: 0.0016 - r_square: 0.9895 - val_loss: 0.0198 - val_acc: 0.6443 - val_rmse: 0.0833 - val_mse: 0.0198 - val_r_square: 0.8641\n",
      "Epoch 84/100\n",
      "35456/35456 [==============================] - 7s 183us/step - loss: 0.0016 - acc: 0.1601 - rmse: 0.0345 - mse: 0.0016 - r_square: 0.9897 - val_loss: 0.0187 - val_acc: 0.6349 - val_rmse: 0.0833 - val_mse: 0.0187 - val_r_square: 0.8716\n",
      "Epoch 85/100\n",
      "35456/35456 [==============================] - 7s 183us/step - loss: 0.0016 - acc: 0.1547 - rmse: 0.0352 - mse: 0.0016 - r_square: 0.9895 - val_loss: 0.0195 - val_acc: 0.7588 - val_rmse: 0.0831 - val_mse: 0.0195 - val_r_square: 0.8662\n",
      "Epoch 86/100\n",
      "35456/35456 [==============================] - 6s 160us/step - loss: 0.0016 - acc: 0.1999 - rmse: 0.0345 - mse: 0.0016 - r_square: 0.9899 - val_loss: 0.0188 - val_acc: 0.6191 - val_rmse: 0.0822 - val_mse: 0.0188 - val_r_square: 0.8712\n",
      "Epoch 87/100\n",
      "35456/35456 [==============================] - 6s 165us/step - loss: 0.0015 - acc: 0.1884 - rmse: 0.0337 - mse: 0.0015 - r_square: 0.9902 - val_loss: 0.0198 - val_acc: 0.5620 - val_rmse: 0.0841 - val_mse: 0.0198 - val_r_square: 0.8645\n",
      "Epoch 88/100\n",
      "35456/35456 [==============================] - 6s 170us/step - loss: 0.0014 - acc: 0.2176 - rmse: 0.0327 - mse: 0.0014 - r_square: 0.9907 - val_loss: 0.0189 - val_acc: 0.0862 - val_rmse: 0.0813 - val_mse: 0.0189 - val_r_square: 0.8704\n",
      "Epoch 89/100\n",
      "35456/35456 [==============================] - 5s 154us/step - loss: 0.0014 - acc: 0.1511 - rmse: 0.0321 - mse: 0.0014 - r_square: 0.9909 - val_loss: 0.0199 - val_acc: 0.0652 - val_rmse: 0.0828 - val_mse: 0.0199 - val_r_square: 0.8637\n",
      "Epoch 90/100\n",
      "35456/35456 [==============================] - 6s 162us/step - loss: 0.0014 - acc: 0.1043 - rmse: 0.0328 - mse: 0.0014 - r_square: 0.9906 - val_loss: 0.0188 - val_acc: 0.3141 - val_rmse: 0.0831 - val_mse: 0.0188 - val_r_square: 0.8713\n",
      "Epoch 91/100\n",
      "35456/35456 [==============================] - 6s 162us/step - loss: 0.0014 - acc: 0.0977 - rmse: 0.0330 - mse: 0.0014 - r_square: 0.9905 - val_loss: 0.0194 - val_acc: 0.6766 - val_rmse: 0.0840 - val_mse: 0.0194 - val_r_square: 0.8672\n",
      "Epoch 92/100\n",
      "35456/35456 [==============================] - 6s 171us/step - loss: 0.0014 - acc: 0.1518 - rmse: 0.0323 - mse: 0.0014 - r_square: 0.9907 - val_loss: 0.0179 - val_acc: 0.7088 - val_rmse: 0.0811 - val_mse: 0.0179 - val_r_square: 0.8778\n",
      "Epoch 93/100\n",
      "35456/35456 [==============================] - 6s 165us/step - loss: 0.0014 - acc: 0.1603 - rmse: 0.0326 - mse: 0.0014 - r_square: 0.9906 - val_loss: 0.0188 - val_acc: 0.8164 - val_rmse: 0.0828 - val_mse: 0.0188 - val_r_square: 0.8711\n",
      "Epoch 94/100\n",
      "35456/35456 [==============================] - 5s 137us/step - loss: 0.0014 - acc: 0.1375 - rmse: 0.0318 - mse: 0.0014 - r_square: 0.9910 - val_loss: 0.0174 - val_acc: 0.6988 - val_rmse: 0.0805 - val_mse: 0.0174 - val_r_square: 0.8810\n",
      "Epoch 95/100\n",
      "35456/35456 [==============================] - 5s 138us/step - loss: 0.0014 - acc: 0.2072 - rmse: 0.0320 - mse: 0.0014 - r_square: 0.9910 - val_loss: 0.0190 - val_acc: 0.3233 - val_rmse: 0.0823 - val_mse: 0.0190 - val_r_square: 0.8699\n",
      "Epoch 96/100\n",
      "35456/35456 [==============================] - 5s 141us/step - loss: 0.0013 - acc: 0.1407 - rmse: 0.0315 - mse: 0.0013 - r_square: 0.9912 - val_loss: 0.0173 - val_acc: 0.0515 - val_rmse: 0.0792 - val_mse: 0.0173 - val_r_square: 0.8818\n",
      "Epoch 97/100\n",
      "35456/35456 [==============================] - 5s 135us/step - loss: 0.0014 - acc: 0.1918 - rmse: 0.0318 - mse: 0.0014 - r_square: 0.9911 - val_loss: 0.0192 - val_acc: 0.0103 - val_rmse: 0.0834 - val_mse: 0.0192 - val_r_square: 0.8685\n",
      "Epoch 98/100\n",
      "35456/35456 [==============================] - 4s 126us/step - loss: 0.0013 - acc: 0.0650 - rmse: 0.0313 - mse: 0.0013 - r_square: 0.9913 - val_loss: 0.0177 - val_acc: 0.0902 - val_rmse: 0.0802 - val_mse: 0.0177 - val_r_square: 0.8788\n",
      "Epoch 99/100\n",
      "35456/35456 [==============================] - 4s 127us/step - loss: 0.0014 - acc: 0.1271 - rmse: 0.0315 - mse: 0.0014 - r_square: 0.9911 - val_loss: 0.0212 - val_acc: 0.6932 - val_rmse: 0.0851 - val_mse: 0.0212 - val_r_square: 0.8545\n",
      "Epoch 100/100\n",
      "35456/35456 [==============================] - 4s 123us/step - loss: 0.0014 - acc: 0.1191 - rmse: 0.0315 - mse: 0.0014 - r_square: 0.9909 - val_loss: 0.0189 - val_acc: 0.7037 - val_rmse: 0.0839 - val_mse: 0.0189 - val_r_square: 0.8706\n"
     ]
    }
   ],
   "source": [
    "input_shape = (2, data.shape[1])\n",
    "model = base_model(data.shape[1], 4, input_shape)\n",
    "model.compile(loss='mse', optimizer='adam',metrics=['acc',rmse,mse,r_square])\n",
    "# mlp overfit on the moons dataset with patient early stopping and model checkpointing\n",
    "from sklearn.datasets import make_moons\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras.callbacks import EarlyStopping,ReduceLROnPlateau\n",
    "reduce_lr_acc = ReduceLROnPlateau(monitor='val_acc', factor=0.3, patience=7, verbose=1, min_delta=1e-4, mode='max')\n",
    "earlyStopping = EarlyStopping(monitor='val_acc', patience=100, verbose=0, mode='max')\n",
    "history = model.fit(x_train, y_train, batch_size=256,epochs=100,shuffle=False ,validation_data=(x_validate, y_validate))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "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": [
    "plt.plot(history.history['loss'],'k-*',label='train')\n",
    "plt.plot(history.history['val_loss'],'r-*', label='test')\n",
    "plt.legend()\n",
    "plt.title(\"LSTM loss during training on Covid19+SARS+Country+Weather\")\n",
    "plt.savefig(\"LSTMLOSS_Covid19 Sars Weather country.pdf\", bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "----------------------------------------------------------------------------------------------------\n",
      "Train Score: 0.00147075 MSE (0.03835037 RMSE ) , 3.21507140  ACC\n",
      "Validation Score: 0.01532461 MSE (0.12379260 RMSE ) , 0.46249296  ACC\n"
     ]
    }
   ],
   "source": [
    "import math \n",
    "print('-' * 100)\n",
    "train_score = model.evaluate(x=x_train, y=y_train, batch_size=3, verbose=0)\n",
    "print('Train Score: %.8f MSE (%.8f RMSE ) , %.8f  ACC' % (train_score[0], math.sqrt(train_score[0]),train_score[1]*100)  )\n",
    "validate_score = model.evaluate(x=x_validate, y=y_validate, batch_size=3, verbose=0)\n",
    "print('Validation Score: %.8f MSE (%.8f RMSE ) , %.8f  ACC' % (validate_score[0], math.sqrt(validate_score[0]),validate_score[1]*100))\n",
    "\n",
    "train_predict = model.predict(x_train)\n",
    "validate_predict = model.predict(x_validate)\n",
    "\n",
    "def inverse_normalize_data(data, scaler):\n",
    "    for i in range(len(data)):\n",
    "        data[i] = scaler.inverse_transform(data[i])\n",
    "\n",
    "    return data\n",
    "\n",
    "train_predict= inverse_normalize_data(train_predict, scaler)\n",
    "y_train = inverse_normalize_data(y_train, scaler)\n",
    "validate_predict = inverse_normalize_data(validate_predict, scaler)\n",
    "y_validate= inverse_normalize_data(y_validate, scaler)\n",
    "#test_predict = inverse_normalize_data(test_predict, scaler)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1080 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "day = ['First Day','Second Day','Third Day','Fourth Day']\n",
    "#dfx = pd.DataFrame(data=[y_validate[:], validate_predict])\n",
    "#dfx = dfx.T\n",
    "#dfx\n",
    "fig = plt.figure(figsize=(20, 15))\n",
    "ax1 = fig.add_subplot(231)\n",
    "ax2 = fig.add_subplot(232)\n",
    "ax3 = fig.add_subplot(233)\n",
    "\n",
    "ax1.plot(day,y_validate[:,:,0][0],color='red',label='Confirmed Actual')\n",
    "ax1.plot(day,validate_predict[:,:,0][0],color='maroon',label='Confirmed Prediction')\n",
    "ax1.title.set_text(\"Confirmed per/day\")\n",
    "ax1.legend()\n",
    "\n",
    "\n",
    "ax2.bar(day,y_validate[:,:,1][0],color='red',label='Deaths Actual')\n",
    "ax2.bar(day,validate_predict[:,:,1][0],color='maroon',label='Deaths Prediction')\n",
    "ax2.title.set_text(\"Deaths per/day\")\n",
    "ax2.legend()\n",
    "\n",
    "\n",
    "ax3.bar(day,y_validate[:,:,2][0],color='red',label='Recoverd Actual')\n",
    "ax3.bar(day,validate_predict[:,:,2][0],color='maroon',label='Recoverd Prediction')\n",
    "ax3.title.set_text(\"Recoverd per/day\")\n",
    "ax3.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# GRU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "gru_1 (GRU)                  (None, 100)               36900     \n",
      "_________________________________________________________________\n",
      "repeat_vector_3 (RepeatVecto (None, 4, 100)            0         \n",
      "_________________________________________________________________\n",
      "gru_2 (GRU)                  (None, 4, 50)             22650     \n",
      "_________________________________________________________________\n",
      "time_distributed_3 (TimeDist (None, 4, 22)             1122      \n",
      "=================================================================\n",
      "Total params: 60,672\n",
      "Trainable params: 60,672\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Train on 35456 samples, validate on 8864 samples\n",
      "Epoch 1/100\n",
      "35456/35456 [==============================] - 10s 291us/step - loss: 0.0095 - acc: 0.0937 - mse: 0.0095 - rmse: 0.0731 - r_square: 0.9359 - val_loss: 0.0587 - val_acc: 0.0000e+00 - val_mse: 0.0587 - val_rmse: 0.1877 - val_r_square: 0.6071\n",
      "Epoch 2/100\n",
      "35456/35456 [==============================] - 8s 219us/step - loss: 0.0114 - acc: 0.1443 - mse: 0.0114 - rmse: 0.0764 - r_square: 0.9213 - val_loss: 0.0598 - val_acc: 0.2156 - val_mse: 0.0598 - val_rmse: 0.1891 - val_r_square: 0.6004\n",
      "Epoch 3/100\n",
      "35456/35456 [==============================] - 8s 220us/step - loss: 0.0136 - acc: 0.3067 - mse: 0.0136 - rmse: 0.0813 - r_square: 0.9013 - val_loss: 0.0641 - val_acc: 0.4485 - val_mse: 0.0641 - val_rmse: 0.2050 - val_r_square: 0.5750\n",
      "Epoch 4/100\n",
      "35456/35456 [==============================] - 8s 219us/step - loss: 0.0243 - acc: 0.2636 - mse: 0.0243 - rmse: 0.1139 - r_square: 0.8322 - val_loss: 0.0573 - val_acc: 0.0000e+00 - val_mse: 0.0573 - val_rmse: 0.1871 - val_r_square: 0.6173\n",
      "Epoch 5/100\n",
      "35456/35456 [==============================] - 8s 225us/step - loss: 0.0192 - acc: 0.1621 - mse: 0.0192 - rmse: 0.1029 - r_square: 0.8690 - val_loss: 0.0570 - val_acc: 0.3055 - val_mse: 0.0570 - val_rmse: 0.1856 - val_r_square: 0.6191\n",
      "Epoch 6/100\n",
      "35456/35456 [==============================] - 8s 221us/step - loss: 0.0138 - acc: 0.0353 - mse: 0.0138 - rmse: 0.0895 - r_square: 0.9081 - val_loss: 0.0615 - val_acc: 0.0000e+00 - val_mse: 0.0615 - val_rmse: 0.1943 - val_r_square: 0.5891\n",
      "Epoch 7/100\n",
      "35456/35456 [==============================] - 8s 222us/step - loss: 0.0126 - acc: 0.0221 - mse: 0.0126 - rmse: 0.0839 - r_square: 0.9145 - val_loss: 0.0887 - val_acc: 0.0000e+00 - val_mse: 0.0887 - val_rmse: 0.2218 - val_r_square: 0.4002\n",
      "Epoch 8/100\n",
      "35456/35456 [==============================] - 8s 227us/step - loss: 0.0159 - acc: 0.3469 - mse: 0.0159 - rmse: 0.0954 - r_square: 0.8923 - val_loss: 0.0592 - val_acc: 0.0000e+00 - val_mse: 0.0592 - val_rmse: 0.1966 - val_r_square: 0.6069\n",
      "Epoch 9/100\n",
      "35456/35456 [==============================] - 8s 224us/step - loss: 0.0151 - acc: 0.1994 - mse: 0.0151 - rmse: 0.0958 - r_square: 0.9003 - val_loss: 0.0645 - val_acc: 0.0000e+00 - val_mse: 0.0645 - val_rmse: 0.1902 - val_r_square: 0.5658\n",
      "Epoch 10/100\n",
      "35456/35456 [==============================] - 8s 223us/step - loss: 0.0136 - acc: 0.2937 - mse: 0.0136 - rmse: 0.0828 - r_square: 0.9056 - val_loss: 0.0697 - val_acc: 0.8902 - val_mse: 0.0697 - val_rmse: 0.1997 - val_r_square: 0.5309\n",
      "Epoch 11/100\n",
      "35456/35456 [==============================] - 8s 227us/step - loss: 0.0148 - acc: 0.0171 - mse: 0.0148 - rmse: 0.0934 - r_square: 0.9007 - val_loss: 0.0554 - val_acc: 0.0000e+00 - val_mse: 0.0554 - val_rmse: 0.1962 - val_r_square: 0.6353\n",
      "Epoch 12/100\n",
      "35456/35456 [==============================] - 8s 231us/step - loss: 0.0154 - acc: 0.2186 - mse: 0.0154 - rmse: 0.0934 - r_square: 0.8971 - val_loss: 0.0529 - val_acc: 0.0364 - val_mse: 0.0529 - val_rmse: 0.1923 - val_r_square: 0.6514\n",
      "Epoch 13/100\n",
      "35456/35456 [==============================] - 8s 225us/step - loss: 0.0162 - acc: 0.3578 - mse: 0.0162 - rmse: 0.0990 - r_square: 0.8925 - val_loss: 0.0404 - val_acc: 0.0879 - val_mse: 0.0404 - val_rmse: 0.1779 - val_r_square: 0.7384\n",
      "Epoch 14/100\n",
      "35456/35456 [==============================] - 8s 231us/step - loss: 0.0088 - acc: 0.0396 - mse: 0.0088 - rmse: 0.0742 - r_square: 0.9418 - val_loss: 0.0359 - val_acc: 0.0000e+00 - val_mse: 0.0359 - val_rmse: 0.1642 - val_r_square: 0.7663\n",
      "Epoch 15/100\n",
      "35456/35456 [==============================] - 8s 228us/step - loss: 0.0108 - acc: 0.1793 - mse: 0.0108 - rmse: 0.0797 - r_square: 0.9300 - val_loss: 0.0487 - val_acc: 0.5018 - val_mse: 0.0487 - val_rmse: 0.1892 - val_r_square: 0.6811\n",
      "Epoch 16/100\n",
      "35456/35456 [==============================] - 8s 227us/step - loss: 0.0126 - acc: 0.0606 - mse: 0.0126 - rmse: 0.0854 - r_square: 0.9176 - val_loss: 0.0507 - val_acc: 0.4258 - val_mse: 0.0507 - val_rmse: 0.1924 - val_r_square: 0.6680\n",
      "Epoch 17/100\n",
      "35456/35456 [==============================] - 8s 229us/step - loss: 0.0144 - acc: 0.1021 - mse: 0.0144 - rmse: 0.0888 - r_square: 0.9031 - val_loss: 0.0437 - val_acc: 0.0000e+00 - val_mse: 0.0437 - val_rmse: 0.1853 - val_r_square: 0.7173\n",
      "Epoch 18/100\n",
      "35456/35456 [==============================] - 8s 232us/step - loss: 0.0118 - acc: 0.1853 - mse: 0.0118 - rmse: 0.0829 - r_square: 0.9240 - val_loss: 0.0509 - val_acc: 0.0825 - val_mse: 0.0509 - val_rmse: 0.1933 - val_r_square: 0.6664\n",
      "Epoch 19/100\n",
      "35456/35456 [==============================] - 8s 228us/step - loss: 0.0129 - acc: 0.2391 - mse: 0.0129 - rmse: 0.0865 - r_square: 0.9154 - val_loss: 0.0457 - val_acc: 0.1741 - val_mse: 0.0457 - val_rmse: 0.1877 - val_r_square: 0.7037\n",
      "Epoch 20/100\n",
      "35456/35456 [==============================] - 8s 228us/step - loss: 0.0119 - acc: 0.2111 - mse: 0.0119 - rmse: 0.0844 - r_square: 0.9232 - val_loss: 0.0471 - val_acc: 0.0641 - val_mse: 0.0471 - val_rmse: 0.1892 - val_r_square: 0.6934\n",
      "Epoch 21/100\n",
      "35456/35456 [==============================] - 8s 232us/step - loss: 0.0122 - acc: 0.1106 - mse: 0.0122 - rmse: 0.0845 - r_square: 0.9206 - val_loss: 0.0574 - val_acc: 0.0000e+00 - val_mse: 0.0574 - val_rmse: 0.1980 - val_r_square: 0.6205\n",
      "Epoch 22/100\n",
      "35456/35456 [==============================] - 8s 233us/step - loss: 0.0126 - acc: 0.1877 - mse: 0.0126 - rmse: 0.0840 - r_square: 0.9159 - val_loss: 0.0465 - val_acc: 0.8902 - val_mse: 0.0465 - val_rmse: 0.1883 - val_r_square: 0.6977\n",
      "Epoch 23/100\n",
      "35456/35456 [==============================] - 8s 216us/step - loss: 0.0129 - acc: 0.0469 - mse: 0.0129 - rmse: 0.0870 - r_square: 0.9157 - val_loss: 0.0539 - val_acc: 0.0000e+00 - val_mse: 0.0539 - val_rmse: 0.1944 - val_r_square: 0.6452\n",
      "Epoch 24/100\n",
      "35456/35456 [==============================] - 8s 222us/step - loss: 0.0140 - acc: 0.1183 - mse: 0.0140 - rmse: 0.0905 - r_square: 0.9079 - val_loss: 0.0501 - val_acc: 0.0000e+00 - val_mse: 0.0501 - val_rmse: 0.1905 - val_r_square: 0.6716\n",
      "Epoch 25/100\n",
      "35456/35456 [==============================] - 8s 212us/step - loss: 0.0135 - acc: 0.0860 - mse: 0.0135 - rmse: 0.0897 - r_square: 0.9110 - val_loss: 0.0499 - val_acc: 0.8902 - val_mse: 0.0499 - val_rmse: 0.1786 - val_r_square: 0.6678\n",
      "Epoch 26/100\n",
      "35456/35456 [==============================] - 8s 214us/step - loss: 0.0137 - acc: 0.0235 - mse: 0.0137 - rmse: 0.0903 - r_square: 0.9092 - val_loss: 0.0458 - val_acc: 0.0000e+00 - val_mse: 0.0458 - val_rmse: 0.1853 - val_r_square: 0.7012\n",
      "Epoch 27/100\n",
      "35456/35456 [==============================] - 8s 219us/step - loss: 0.0129 - acc: 0.0957 - mse: 0.0129 - rmse: 0.0883 - r_square: 0.9157 - val_loss: 0.0435 - val_acc: 0.0000e+00 - val_mse: 0.0435 - val_rmse: 0.1823 - val_r_square: 0.7170\n",
      "Epoch 28/100\n",
      "35456/35456 [==============================] - 8s 218us/step - loss: 0.0131 - acc: 0.0939 - mse: 0.0131 - rmse: 0.0888 - r_square: 0.9153 - val_loss: 0.0465 - val_acc: 0.0000e+00 - val_mse: 0.0465 - val_rmse: 0.1878 - val_r_square: 0.6971\n",
      "Epoch 29/100\n",
      "35456/35456 [==============================] - 8s 215us/step - loss: 0.0119 - acc: 0.0470 - mse: 0.0119 - rmse: 0.0848 - r_square: 0.9222 - val_loss: 0.0443 - val_acc: 0.0000e+00 - val_mse: 0.0443 - val_rmse: 0.1839 - val_r_square: 0.7115\n",
      "Epoch 30/100\n",
      "35456/35456 [==============================] - 7s 192us/step - loss: 0.0125 - acc: 0.0639 - mse: 0.0125 - rmse: 0.0859 - r_square: 0.9182 - val_loss: 0.0509 - val_acc: 0.0000e+00 - val_mse: 0.0509 - val_rmse: 0.1926 - val_r_square: 0.6668\n",
      "Epoch 31/100\n",
      "35456/35456 [==============================] - 7s 196us/step - loss: 0.0116 - acc: 0.0796 - mse: 0.0116 - rmse: 0.0827 - r_square: 0.9238 - val_loss: 0.0478 - val_acc: 0.8902 - val_mse: 0.0478 - val_rmse: 0.1892 - val_r_square: 0.6880\n",
      "Epoch 32/100\n",
      "35456/35456 [==============================] - 7s 194us/step - loss: 0.0126 - acc: 0.1227 - mse: 0.0126 - rmse: 0.0852 - r_square: 0.9184 - val_loss: 0.0461 - val_acc: 0.0000e+00 - val_mse: 0.0461 - val_rmse: 0.1855 - val_r_square: 0.6993\n",
      "Epoch 33/100\n",
      "35456/35456 [==============================] - 7s 194us/step - loss: 0.0126 - acc: 0.1155 - mse: 0.0126 - rmse: 0.0856 - r_square: 0.9175 - val_loss: 0.0477 - val_acc: 0.0000e+00 - val_mse: 0.0477 - val_rmse: 0.1857 - val_r_square: 0.6867\n",
      "Epoch 34/100\n",
      "35456/35456 [==============================] - 7s 193us/step - loss: 0.0130 - acc: 0.0712 - mse: 0.0130 - rmse: 0.0875 - r_square: 0.9150 - val_loss: 0.0474 - val_acc: 0.0000e+00 - val_mse: 0.0474 - val_rmse: 0.1869 - val_r_square: 0.6900\n",
      "Epoch 35/100\n",
      "35456/35456 [==============================] - 7s 194us/step - loss: 0.0154 - acc: 0.0451 - mse: 0.0154 - rmse: 0.0969 - r_square: 0.8987 - val_loss: 0.0461 - val_acc: 0.0000e+00 - val_mse: 0.0461 - val_rmse: 0.1856 - val_r_square: 0.6989\n",
      "Epoch 36/100\n",
      "35456/35456 [==============================] - 7s 192us/step - loss: 0.0113 - acc: 0.1625 - mse: 0.0113 - rmse: 0.0811 - r_square: 0.9255 - val_loss: 0.0505 - val_acc: 0.8902 - val_mse: 0.0505 - val_rmse: 0.1902 - val_r_square: 0.6682\n",
      "Epoch 37/100\n",
      "35456/35456 [==============================] - 7s 195us/step - loss: 0.0120 - acc: 0.0343 - mse: 0.0120 - rmse: 0.0838 - r_square: 0.9213 - val_loss: 0.0467 - val_acc: 0.0000e+00 - val_mse: 0.0467 - val_rmse: 0.1884 - val_r_square: 0.6963\n",
      "Epoch 38/100\n",
      "35456/35456 [==============================] - 7s 196us/step - loss: 0.0127 - acc: 0.0740 - mse: 0.0127 - rmse: 0.0878 - r_square: 0.9162 - val_loss: 0.0455 - val_acc: 0.0000e+00 - val_mse: 0.0455 - val_rmse: 0.1849 - val_r_square: 0.7035\n",
      "Epoch 39/100\n",
      "35456/35456 [==============================] - 7s 194us/step - loss: 0.0132 - acc: 0.0808 - mse: 0.0132 - rmse: 0.0892 - r_square: 0.9145 - val_loss: 0.0422 - val_acc: 0.0000e+00 - val_mse: 0.0422 - val_rmse: 0.1811 - val_r_square: 0.7266\n",
      "Epoch 40/100\n",
      "35456/35456 [==============================] - 7s 195us/step - loss: 0.0136 - acc: 0.0505 - mse: 0.0136 - rmse: 0.0909 - r_square: 0.9118 - val_loss: 0.0443 - val_acc: 0.0000e+00 - val_mse: 0.0443 - val_rmse: 0.1829 - val_r_square: 0.7115\n",
      "Epoch 41/100\n",
      "35456/35456 [==============================] - 7s 199us/step - loss: 0.0144 - acc: 0.0379 - mse: 0.0144 - rmse: 0.0935 - r_square: 0.9070 - val_loss: 0.0432 - val_acc: 0.0000e+00 - val_mse: 0.0432 - val_rmse: 0.1811 - val_r_square: 0.7192\n",
      "Epoch 42/100\n",
      "35456/35456 [==============================] - 7s 201us/step - loss: 0.0140 - acc: 0.0308 - mse: 0.0140 - rmse: 0.0927 - r_square: 0.9089 - val_loss: 0.0445 - val_acc: 0.0000e+00 - val_mse: 0.0445 - val_rmse: 0.1823 - val_r_square: 0.7099\n",
      "Epoch 43/100\n",
      "35456/35456 [==============================] - 7s 196us/step - loss: 0.0146 - acc: 0.0235 - mse: 0.0146 - rmse: 0.0945 - r_square: 0.9052 - val_loss: 0.0443 - val_acc: 0.0000e+00 - val_mse: 0.0443 - val_rmse: 0.1820 - val_r_square: 0.7115\n",
      "Epoch 44/100\n",
      "35456/35456 [==============================] - 7s 195us/step - loss: 0.0139 - acc: 0.0209 - mse: 0.0139 - rmse: 0.0924 - r_square: 0.9094 - val_loss: 0.0447 - val_acc: 0.0000e+00 - val_mse: 0.0447 - val_rmse: 0.1824 - val_r_square: 0.7079\n",
      "Epoch 45/100\n",
      "35456/35456 [==============================] - 7s 192us/step - loss: 0.0147 - acc: 0.0161 - mse: 0.0147 - rmse: 0.0951 - r_square: 0.9044 - val_loss: 0.0444 - val_acc: 0.0000e+00 - val_mse: 0.0444 - val_rmse: 0.1821 - val_r_square: 0.7105\n",
      "Epoch 46/100\n",
      "35456/35456 [==============================] - 7s 200us/step - loss: 0.0140 - acc: 0.0144 - mse: 0.0140 - rmse: 0.0923 - r_square: 0.9092 - val_loss: 0.0450 - val_acc: 0.0000e+00 - val_mse: 0.0450 - val_rmse: 0.1826 - val_r_square: 0.7061\n",
      "Epoch 47/100\n",
      "35456/35456 [==============================] - 7s 192us/step - loss: 0.0148 - acc: 0.0129 - mse: 0.0148 - rmse: 0.0955 - r_square: 0.9038 - val_loss: 0.0451 - val_acc: 0.0000e+00 - val_mse: 0.0451 - val_rmse: 0.1833 - val_r_square: 0.7060\n",
      "Epoch 48/100\n",
      "35456/35456 [==============================] - 7s 194us/step - loss: 0.0131 - acc: 0.0199 - mse: 0.0131 - rmse: 0.0884 - r_square: 0.9146 - val_loss: 0.0436 - val_acc: 0.0000e+00 - val_mse: 0.0436 - val_rmse: 0.1816 - val_r_square: 0.7158\n",
      "Epoch 49/100\n",
      "35456/35456 [==============================] - 7s 198us/step - loss: 0.0139 - acc: 0.0307 - mse: 0.0139 - rmse: 0.0917 - r_square: 0.9101 - val_loss: 0.0463 - val_acc: 0.0000e+00 - val_mse: 0.0463 - val_rmse: 0.1865 - val_r_square: 0.6982\n",
      "Epoch 50/100\n",
      "35456/35456 [==============================] - 7s 203us/step - loss: 0.0110 - acc: 0.0379 - mse: 0.0110 - rmse: 0.0816 - r_square: 0.9274 - val_loss: 0.0452 - val_acc: 0.0000e+00 - val_mse: 0.0452 - val_rmse: 0.1857 - val_r_square: 0.7055\n",
      "Epoch 51/100\n",
      "35456/35456 [==============================] - 7s 197us/step - loss: 0.0127 - acc: 0.0090 - mse: 0.0127 - rmse: 0.0865 - r_square: 0.9178 - val_loss: 0.0459 - val_acc: 0.0000e+00 - val_mse: 0.0459 - val_rmse: 0.1850 - val_r_square: 0.7003\n",
      "Epoch 52/100\n",
      "35456/35456 [==============================] - 7s 197us/step - loss: 0.0122 - acc: 0.0000e+00 - mse: 0.0122 - rmse: 0.0843 - r_square: 0.9202 - val_loss: 0.0506 - val_acc: 0.0000e+00 - val_mse: 0.0506 - val_rmse: 0.1930 - val_r_square: 0.6688\n",
      "Epoch 53/100\n",
      "35456/35456 [==============================] - 7s 203us/step - loss: 0.0116 - acc: 0.1697 - mse: 0.0116 - rmse: 0.0826 - r_square: 0.9234 - val_loss: 0.0459 - val_acc: 0.0000e+00 - val_mse: 0.0459 - val_rmse: 0.1872 - val_r_square: 0.7016\n",
      "Epoch 54/100\n",
      "35456/35456 [==============================] - 7s 199us/step - loss: 0.0132 - acc: 0.1588 - mse: 0.0132 - rmse: 0.0892 - r_square: 0.9143 - val_loss: 0.0463 - val_acc: 0.0000e+00 - val_mse: 0.0463 - val_rmse: 0.1857 - val_r_square: 0.6976\n",
      "Epoch 55/100\n",
      "35456/35456 [==============================] - 7s 195us/step - loss: 0.0140 - acc: 0.1155 - mse: 0.0140 - rmse: 0.0919 - r_square: 0.9090 - val_loss: 0.0441 - val_acc: 0.0000e+00 - val_mse: 0.0441 - val_rmse: 0.1826 - val_r_square: 0.7128\n",
      "Epoch 56/100\n",
      "35456/35456 [==============================] - 7s 199us/step - loss: 0.0140 - acc: 0.0614 - mse: 0.0140 - rmse: 0.0923 - r_square: 0.9091 - val_loss: 0.0453 - val_acc: 0.0000e+00 - val_mse: 0.0453 - val_rmse: 0.1835 - val_r_square: 0.7039\n",
      "Epoch 57/100\n",
      "35456/35456 [==============================] - 7s 200us/step - loss: 0.0143 - acc: 0.0469 - mse: 0.0143 - rmse: 0.0933 - r_square: 0.9073 - val_loss: 0.0437 - val_acc: 0.0000e+00 - val_mse: 0.0437 - val_rmse: 0.1815 - val_r_square: 0.7155\n",
      "Epoch 58/100\n",
      "35456/35456 [==============================] - 7s 196us/step - loss: 0.0144 - acc: 0.0397 - mse: 0.0144 - rmse: 0.0939 - r_square: 0.9067 - val_loss: 0.0450 - val_acc: 0.0000e+00 - val_mse: 0.0450 - val_rmse: 0.1827 - val_r_square: 0.7060\n",
      "Epoch 59/100\n",
      "35456/35456 [==============================] - 7s 197us/step - loss: 0.0141 - acc: 0.0325 - mse: 0.0141 - rmse: 0.0926 - r_square: 0.9087 - val_loss: 0.0435 - val_acc: 0.0000e+00 - val_mse: 0.0435 - val_rmse: 0.1810 - val_r_square: 0.7167\n",
      "Epoch 60/100\n",
      "35456/35456 [==============================] - 7s 204us/step - loss: 0.0144 - acc: 0.0217 - mse: 0.0144 - rmse: 0.0940 - r_square: 0.9066 - val_loss: 0.0450 - val_acc: 0.0000e+00 - val_mse: 0.0450 - val_rmse: 0.1826 - val_r_square: 0.7056\n",
      "Epoch 61/100\n",
      "35456/35456 [==============================] - 7s 203us/step - loss: 0.0140 - acc: 0.0253 - mse: 0.0140 - rmse: 0.0920 - r_square: 0.9095 - val_loss: 0.0433 - val_acc: 0.0000e+00 - val_mse: 0.0433 - val_rmse: 0.1808 - val_r_square: 0.7182\n",
      "Epoch 62/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "35456/35456 [==============================] - 7s 192us/step - loss: 0.0143 - acc: 0.0235 - mse: 0.0143 - rmse: 0.0937 - r_square: 0.9073 - val_loss: 0.0452 - val_acc: 0.0000e+00 - val_mse: 0.0452 - val_rmse: 0.1829 - val_r_square: 0.7043\n",
      "Epoch 63/100\n",
      "35456/35456 [==============================] - 7s 191us/step - loss: 0.0136 - acc: 0.0253 - mse: 0.0136 - rmse: 0.0908 - r_square: 0.9118 - val_loss: 0.0432 - val_acc: 0.0000e+00 - val_mse: 0.0432 - val_rmse: 0.1819 - val_r_square: 0.7193\n",
      "Epoch 64/100\n",
      "35456/35456 [==============================] - 7s 199us/step - loss: 0.0137 - acc: 0.0235 - mse: 0.0137 - rmse: 0.0915 - r_square: 0.9113 - val_loss: 0.0448 - val_acc: 0.0000e+00 - val_mse: 0.0448 - val_rmse: 0.1824 - val_r_square: 0.7075\n",
      "Epoch 65/100\n",
      "35456/35456 [==============================] - 7s 193us/step - loss: 0.0134 - acc: 0.1588 - mse: 0.0134 - rmse: 0.0899 - r_square: 0.9133 - val_loss: 0.0432 - val_acc: 0.0000e+00 - val_mse: 0.0432 - val_rmse: 0.1812 - val_r_square: 0.7193\n",
      "Epoch 66/100\n",
      "35456/35456 [==============================] - 7s 193us/step - loss: 0.0124 - acc: 0.0144 - mse: 0.0124 - rmse: 0.0856 - r_square: 0.9197 - val_loss: 0.0455 - val_acc: 0.0000e+00 - val_mse: 0.0455 - val_rmse: 0.1847 - val_r_square: 0.7032\n",
      "Epoch 67/100\n",
      "35456/35456 [==============================] - 7s 195us/step - loss: 0.0124 - acc: 0.1155 - mse: 0.0124 - rmse: 0.0854 - r_square: 0.9194 - val_loss: 0.0444 - val_acc: 0.0000e+00 - val_mse: 0.0444 - val_rmse: 0.1838 - val_r_square: 0.7111\n",
      "Epoch 68/100\n",
      "35456/35456 [==============================] - 7s 200us/step - loss: 0.0129 - acc: 0.1227 - mse: 0.0129 - rmse: 0.0876 - r_square: 0.9162 - val_loss: 0.0475 - val_acc: 0.0000e+00 - val_mse: 0.0475 - val_rmse: 0.1884 - val_r_square: 0.6898\n",
      "Epoch 69/100\n",
      "35456/35456 [==============================] - 7s 196us/step - loss: 0.0118 - acc: 0.0975 - mse: 0.0118 - rmse: 0.0843 - r_square: 0.9223 - val_loss: 0.0439 - val_acc: 0.8902 - val_mse: 0.0439 - val_rmse: 0.1854 - val_r_square: 0.7157\n",
      "Epoch 70/100\n",
      "35456/35456 [==============================] - 7s 193us/step - loss: 0.0118 - acc: 0.0253 - mse: 0.0118 - rmse: 0.0845 - r_square: 0.9231 - val_loss: 0.0530 - val_acc: 0.0000e+00 - val_mse: 0.0530 - val_rmse: 0.1948 - val_r_square: 0.6512\n",
      "Epoch 71/100\n",
      "35456/35456 [==============================] - 7s 197us/step - loss: 0.0125 - acc: 0.0181 - mse: 0.0125 - rmse: 0.0864 - r_square: 0.9178 - val_loss: 0.0434 - val_acc: 0.0000e+00 - val_mse: 0.0434 - val_rmse: 0.1827 - val_r_square: 0.7186\n",
      "Epoch 72/100\n",
      "35456/35456 [==============================] - 7s 198us/step - loss: 0.0142 - acc: 0.0451 - mse: 0.0142 - rmse: 0.0931 - r_square: 0.9079 - val_loss: 0.0458 - val_acc: 0.0000e+00 - val_mse: 0.0458 - val_rmse: 0.1845 - val_r_square: 0.7010\n",
      "Epoch 73/100\n",
      "35456/35456 [==============================] - 7s 196us/step - loss: 0.0145 - acc: 0.0487 - mse: 0.0145 - rmse: 0.0939 - r_square: 0.9060 - val_loss: 0.0438 - val_acc: 0.0000e+00 - val_mse: 0.0438 - val_rmse: 0.1818 - val_r_square: 0.7150\n",
      "Epoch 74/100\n",
      "35456/35456 [==============================] - 7s 195us/step - loss: 0.0134 - acc: 0.1119 - mse: 0.0134 - rmse: 0.0910 - r_square: 0.9133 - val_loss: 0.0449 - val_acc: 0.0000e+00 - val_mse: 0.0449 - val_rmse: 0.1722 - val_r_square: 0.7022\n",
      "Epoch 75/100\n",
      "35456/35456 [==============================] - 7s 199us/step - loss: 0.0139 - acc: 0.1119 - mse: 0.0139 - rmse: 0.0915 - r_square: 0.9088 - val_loss: 0.0435 - val_acc: 0.0000e+00 - val_mse: 0.0435 - val_rmse: 0.1831 - val_r_square: 0.7183\n",
      "Epoch 76/100\n",
      "35456/35456 [==============================] - 7s 198us/step - loss: 0.0133 - acc: 0.0939 - mse: 0.0133 - rmse: 0.0897 - r_square: 0.9139 - val_loss: 0.0468 - val_acc: 0.8902 - val_mse: 0.0468 - val_rmse: 0.1861 - val_r_square: 0.6940\n",
      "Epoch 77/100\n",
      "35456/35456 [==============================] - 7s 194us/step - loss: 0.0143 - acc: 0.0939 - mse: 0.0143 - rmse: 0.0930 - r_square: 0.9075 - val_loss: 0.0430 - val_acc: 0.8902 - val_mse: 0.0430 - val_rmse: 0.1816 - val_r_square: 0.7205\n",
      "Epoch 78/100\n",
      "35456/35456 [==============================] - 7s 210us/step - loss: 0.0132 - acc: 0.0740 - mse: 0.0132 - rmse: 0.0903 - r_square: 0.9146 - val_loss: 0.0473 - val_acc: 0.0000e+00 - val_mse: 0.0473 - val_rmse: 0.1867 - val_r_square: 0.6905\n",
      "Epoch 79/100\n",
      "35456/35456 [==============================] - 9s 245us/step - loss: 0.0138 - acc: 0.0975 - mse: 0.0138 - rmse: 0.0923 - r_square: 0.9105 - val_loss: 0.0451 - val_acc: 0.8902 - val_mse: 0.0451 - val_rmse: 0.1845 - val_r_square: 0.7065\n",
      "Epoch 80/100\n",
      "35456/35456 [==============================] - 9s 240us/step - loss: 0.0140 - acc: 0.0866 - mse: 0.0140 - rmse: 0.0921 - r_square: 0.9087 - val_loss: 0.0457 - val_acc: 0.8902 - val_mse: 0.0457 - val_rmse: 0.1841 - val_r_square: 0.7015\n",
      "Epoch 81/100\n",
      "35456/35456 [==============================] - 8s 233us/step - loss: 0.0143 - acc: 0.0758 - mse: 0.0143 - rmse: 0.0933 - r_square: 0.9072 - val_loss: 0.0444 - val_acc: 0.8902 - val_mse: 0.0444 - val_rmse: 0.1825 - val_r_square: 0.7108\n",
      "Epoch 82/100\n",
      "35456/35456 [==============================] - 8s 220us/step - loss: 0.0138 - acc: 0.0957 - mse: 0.0138 - rmse: 0.0919 - r_square: 0.9101 - val_loss: 0.0444 - val_acc: 0.8902 - val_mse: 0.0444 - val_rmse: 0.1822 - val_r_square: 0.7104\n",
      "Epoch 83/100\n",
      "35456/35456 [==============================] - 8s 217us/step - loss: 0.0167 - acc: 0.0560 - mse: 0.0167 - rmse: 0.0981 - r_square: 0.8852 - val_loss: 0.0435 - val_acc: 0.0000e+00 - val_mse: 0.0435 - val_rmse: 0.1817 - val_r_square: 0.7169\n",
      "Epoch 84/100\n",
      "35456/35456 [==============================] - 8s 215us/step - loss: 0.0155 - acc: 0.0678 - mse: 0.0155 - rmse: 0.0981 - r_square: 0.8983 - val_loss: 0.0446 - val_acc: 0.0000e+00 - val_mse: 0.0446 - val_rmse: 0.1826 - val_r_square: 0.7087\n",
      "Epoch 85/100\n",
      "35456/35456 [==============================] - 8s 220us/step - loss: 0.0134 - acc: 0.0560 - mse: 0.0134 - rmse: 0.0901 - r_square: 0.9139 - val_loss: 0.0428 - val_acc: 0.0000e+00 - val_mse: 0.0428 - val_rmse: 0.1822 - val_r_square: 0.7222\n",
      "Epoch 86/100\n",
      "35456/35456 [==============================] - 8s 222us/step - loss: 0.0129 - acc: 0.0614 - mse: 0.0129 - rmse: 0.0878 - r_square: 0.9171 - val_loss: 0.0433 - val_acc: 0.0000e+00 - val_mse: 0.0433 - val_rmse: 0.1808 - val_r_square: 0.7177\n",
      "Epoch 87/100\n",
      "35456/35456 [==============================] - 8s 219us/step - loss: 0.0130 - acc: 0.0542 - mse: 0.0130 - rmse: 0.0883 - r_square: 0.9166 - val_loss: 0.0442 - val_acc: 0.8902 - val_mse: 0.0442 - val_rmse: 0.1832 - val_r_square: 0.7123\n",
      "Epoch 88/100\n",
      "35456/35456 [==============================] - 8s 218us/step - loss: 0.0120 - acc: 0.0036 - mse: 0.0120 - rmse: 0.0842 - r_square: 0.9221 - val_loss: 0.0461 - val_acc: 0.0000e+00 - val_mse: 0.0461 - val_rmse: 0.1853 - val_r_square: 0.6985\n",
      "Epoch 89/100\n",
      "35456/35456 [==============================] - 8s 221us/step - loss: 0.0129 - acc: 0.0072 - mse: 0.0129 - rmse: 0.0876 - r_square: 0.9162 - val_loss: 0.0444 - val_acc: 0.0000e+00 - val_mse: 0.0444 - val_rmse: 0.1836 - val_r_square: 0.7112\n",
      "Epoch 90/100\n",
      "35456/35456 [==============================] - 8s 219us/step - loss: 0.0129 - acc: 0.0433 - mse: 0.0129 - rmse: 0.0873 - r_square: 0.9163 - val_loss: 0.0457 - val_acc: 0.0000e+00 - val_mse: 0.0457 - val_rmse: 0.1855 - val_r_square: 0.7018\n",
      "Epoch 91/100\n",
      "35456/35456 [==============================] - 8s 217us/step - loss: 0.0129 - acc: 0.0975 - mse: 0.0129 - rmse: 0.0872 - r_square: 0.9164 - val_loss: 0.0442 - val_acc: 0.0000e+00 - val_mse: 0.0442 - val_rmse: 0.1837 - val_r_square: 0.7128\n",
      "Epoch 92/100\n",
      "35456/35456 [==============================] - 8s 224us/step - loss: 0.0123 - acc: 0.1625 - mse: 0.0123 - rmse: 0.0854 - r_square: 0.9200 - val_loss: 0.0463 - val_acc: 0.0000e+00 - val_mse: 0.0463 - val_rmse: 0.1876 - val_r_square: 0.6979\n",
      "Epoch 93/100\n",
      "35456/35456 [==============================] - 8s 233us/step - loss: 0.0114 - acc: 0.0018 - mse: 0.0114 - rmse: 0.0818 - r_square: 0.9256 - val_loss: 0.0483 - val_acc: 0.0000e+00 - val_mse: 0.0483 - val_rmse: 0.1886 - val_r_square: 0.6842\n",
      "Epoch 94/100\n",
      "35456/35456 [==============================] - 9s 250us/step - loss: 0.0125 - acc: 0.0596 - mse: 0.0125 - rmse: 0.0865 - r_square: 0.9173 - val_loss: 0.0454 - val_acc: 0.0000e+00 - val_mse: 0.0454 - val_rmse: 0.1853 - val_r_square: 0.7043\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 95/100\n",
      "35456/35456 [==============================] - 8s 213us/step - loss: 0.0142 - acc: 0.0361 - mse: 0.0142 - rmse: 0.0931 - r_square: 0.9077 - val_loss: 0.0437 - val_acc: 0.0000e+00 - val_mse: 0.0437 - val_rmse: 0.1825 - val_r_square: 0.7153\n",
      "Epoch 96/100\n",
      "35456/35456 [==============================] - 8s 213us/step - loss: 0.0125 - acc: 0.0650 - mse: 0.0125 - rmse: 0.0866 - r_square: 0.9185 - val_loss: 0.0434 - val_acc: 0.0000e+00 - val_mse: 0.0434 - val_rmse: 0.1822 - val_r_square: 0.7178\n",
      "Epoch 97/100\n",
      "35456/35456 [==============================] - 7s 194us/step - loss: 0.0134 - acc: 0.0253 - mse: 0.0134 - rmse: 0.0898 - r_square: 0.9130 - val_loss: 0.0434 - val_acc: 0.0000e+00 - val_mse: 0.0434 - val_rmse: 0.1818 - val_r_square: 0.7178\n",
      "Epoch 98/100\n",
      "35456/35456 [==============================] - 7s 193us/step - loss: 0.0137 - acc: 0.0217 - mse: 0.0137 - rmse: 0.0912 - r_square: 0.9112 - val_loss: 0.0463 - val_acc: 0.0000e+00 - val_mse: 0.0463 - val_rmse: 0.1854 - val_r_square: 0.6977\n",
      "Epoch 99/100\n",
      "35456/35456 [==============================] - 7s 197us/step - loss: 0.0119 - acc: 0.0253 - mse: 0.0119 - rmse: 0.0828 - r_square: 0.9221 - val_loss: 0.0430 - val_acc: 0.0000e+00 - val_mse: 0.0430 - val_rmse: 0.1821 - val_r_square: 0.7211\n",
      "Epoch 100/100\n",
      "35456/35456 [==============================] - 7s 193us/step - loss: 0.0141 - acc: 0.0343 - mse: 0.0141 - rmse: 0.0921 - r_square: 0.9088 - val_loss: 0.0502 - val_acc: 0.0000e+00 - val_mse: 0.0502 - val_rmse: 0.1924 - val_r_square: 0.6717\n"
     ]
    }
   ],
   "source": [
    "def base_model2(feature_len=3, after_day=3, input_shape=(8, 1)):\n",
    "    model = Sequential()\n",
    "    model.add(GRU(units=100, return_sequences=False, input_shape=input_shape))\n",
    "    #model.add(LSTM(units=100, return_sequences=False, input_shape=input_shape))\n",
    "    # one to many\n",
    "    model.add(RepeatVector(after_day))\n",
    "    #model.add(LSTM(200, return_sequences=True))\n",
    "    model.add(GRU(50, return_sequences=True))\n",
    "    model.add(TimeDistributed(Dense(units=feature_len, activation='linear')))\n",
    "    return model\n",
    "opt = Adam(lr=0.01)\n",
    "input_shape = (2, data.shape[1])\n",
    "model = base_model2(data.shape[1], 4, input_shape)\n",
    "model.compile(loss='mse', optimizer=opt,metrics=['acc',mse,rmse,r_square])\n",
    "model.summary()\n",
    "history = model.fit(x_train, y_train, batch_size=64, epochs=100,shuffle=False ,validation_data=(x_validate, y_validate))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "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": [
    "plt.plot(history.history['loss'],'k-*',label='train')\n",
    "plt.plot(history.history['val_loss'],'r-*', label='test')\n",
    "plt.legend()\n",
    "plt.title(\"GRU loss during training on Covid19+SARS+Country+Weather\")\n",
    "plt.savefig(\"GRULOSS_Covid19 Sars Weather country.pdf\", bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 135,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "----------------------------------------------------------------------------------------------------\n",
      "Train Score: 0.02737362 MSE (0.16544975 RMSE ) , 0.00000000  ACC\n",
      "Validation Score: 0.04929391 MSE (0.22202232 RMSE ) , 0.00000000  ACC\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1080 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import math \n",
    "print('-' * 100)\n",
    "train_score = model.evaluate(x=x_train, y=y_train, batch_size=3, verbose=0)\n",
    "print('Train Score: %.8f MSE (%.8f RMSE ) , %.8f  ACC' % (train_score[0], math.sqrt(train_score[0]),train_score[1]*100)  )\n",
    "validate_score = model.evaluate(x=x_validate, y=y_validate, batch_size=3, verbose=0)\n",
    "print('Validation Score: %.8f MSE (%.8f RMSE ) , %.8f  ACC' % (validate_score[0], math.sqrt(validate_score[0]),validate_score[1]*100))\n",
    "\n",
    "train_predict = model.predict(x_train)\n",
    "validate_predict = model.predict(x_validate)\n",
    "\n",
    "def inverse_normalize_data(data, scaler):\n",
    "    for i in range(len(data)):\n",
    "        data[i] = scaler.inverse_transform(data[i])\n",
    "\n",
    "    return data\n",
    "\n",
    "train_predict= inverse_normalize_data(train_predict, scaler)\n",
    "y_train = inverse_normalize_data(y_train, scaler)\n",
    "validate_predict = inverse_normalize_data(validate_predict, scaler)\n",
    "y_validate= inverse_normalize_data(y_validate, scaler)\n",
    "#test_predict = inverse_normalize_data(test_predict, scaler)\n",
    "day = ['First Day','Second Day','Third Day','Fourth Day']\n",
    "#dfx = pd.DataFrame(data=[y_validate[:], validate_predict])\n",
    "#dfx = dfx.T\n",
    "#dfx\n",
    "fig = plt.figure(figsize=(20, 15))\n",
    "ax1 = fig.add_subplot(231)\n",
    "ax2 = fig.add_subplot(232)\n",
    "ax3 = fig.add_subplot(233)\n",
    "\n",
    "ax1.plot(day,y_validate[:,:,0][0],color='red',label='Confirmed Actual')\n",
    "ax1.plot(day,validate_predict[:,:,0][0],color='maroon',label='Confirmed Prediction')\n",
    "ax1.title.set_text(\"Confirmed per/day\")\n",
    "ax1.legend()\n",
    "\n",
    "\n",
    "ax2.bar(day,y_validate[:,:,1][0],color='red',label='Deaths Actual')\n",
    "ax2.bar(day,validate_predict[:,:,1][0],color='maroon',label='Deaths Prediction')\n",
    "ax2.title.set_text(\"Deaths per/day\")\n",
    "ax2.legend()\n",
    "\n",
    "\n",
    "ax3.bar(day,y_validate[:,:,2][0],color='red',label='Recoverd Actual')\n",
    "ax3.bar(day,validate_predict[:,:,2][0],color='maroon',label='Recoverd Prediction')\n",
    "ax3.title.set_text(\"Recoverd per/day\")\n",
    "ax3.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Seq2Seq model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /home/rad/anaconda3/envs/tensorflow/lib/python3.7/site-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use tf.cast instead.\n",
      "Train on 35456 samples, validate on 8864 samples\n",
      "Epoch 1/100\n",
      "35456/35456 [==============================] - 9s 253us/step - loss: 0.0579 - acc: 0.1878 - rmse: 0.1876 - mse: 0.0579 - r_square: 0.6067 - val_loss: 0.0310 - val_acc: 0.2459 - val_rmse: 0.1302 - val_mse: 0.0310 - val_r_square: 0.7905\n",
      "Epoch 2/100\n",
      "35456/35456 [==============================] - 5s 151us/step - loss: 0.0142 - acc: 0.1482 - rmse: 0.1002 - mse: 0.0142 - r_square: 0.9057 - val_loss: 0.0245 - val_acc: 0.0000e+00 - val_rmse: 0.1104 - val_mse: 0.0245 - val_r_square: 0.8334\n",
      "Epoch 3/100\n",
      "35456/35456 [==============================] - 5s 152us/step - loss: 0.0066 - acc: 0.1498 - rmse: 0.0686 - mse: 0.0066 - r_square: 0.9553 - val_loss: 0.0206 - val_acc: 9.5894e-04 - val_rmse: 0.0960 - val_mse: 0.0206 - val_r_square: 0.8594\n",
      "Epoch 4/100\n",
      "35456/35456 [==============================] - 5s 155us/step - loss: 0.0052 - acc: 0.1187 - rmse: 0.0636 - mse: 0.0052 - r_square: 0.9651 - val_loss: 0.0191 - val_acc: 0.0464 - val_rmse: 0.0943 - val_mse: 0.0191 - val_r_square: 0.8697\n",
      "Epoch 5/100\n",
      "35456/35456 [==============================] - 5s 153us/step - loss: 0.0053 - acc: 0.1043 - rmse: 0.0644 - mse: 0.0053 - r_square: 0.9651 - val_loss: 0.0189 - val_acc: 0.0255 - val_rmse: 0.0954 - val_mse: 0.0189 - val_r_square: 0.8714\n",
      "Epoch 6/100\n",
      "35456/35456 [==============================] - 6s 155us/step - loss: 0.0050 - acc: 0.1481 - rmse: 0.0626 - mse: 0.0050 - r_square: 0.9668 - val_loss: 0.0229 - val_acc: 0.0085 - val_rmse: 0.1004 - val_mse: 0.0229 - val_r_square: 0.8433\n",
      "Epoch 7/100\n",
      "35456/35456 [==============================] - 5s 153us/step - loss: 0.0059 - acc: 0.1313 - rmse: 0.0661 - mse: 0.0059 - r_square: 0.9601 - val_loss: 0.0287 - val_acc: 0.0761 - val_rmse: 0.1095 - val_mse: 0.0287 - val_r_square: 0.8034\n",
      "Epoch 8/100\n",
      "35456/35456 [==============================] - 5s 154us/step - loss: 0.0089 - acc: 0.0583 - rmse: 0.0758 - mse: 0.0089 - r_square: 0.9382 - val_loss: 0.0210 - val_acc: 0.0000e+00 - val_rmse: 0.0963 - val_mse: 0.0210 - val_r_square: 0.8560\n",
      "Epoch 9/100\n",
      "35456/35456 [==============================] - 5s 155us/step - loss: 0.0052 - acc: 0.0840 - rmse: 0.0610 - mse: 0.0052 - r_square: 0.9652 - val_loss: 0.0188 - val_acc: 0.0467 - val_rmse: 0.0927 - val_mse: 0.0188 - val_r_square: 0.8720\n",
      "Epoch 10/100\n",
      "35456/35456 [==============================] - 6s 155us/step - loss: 0.0046 - acc: 0.0928 - rmse: 0.0604 - mse: 0.0046 - r_square: 0.9696 - val_loss: 0.0176 - val_acc: 0.0465 - val_rmse: 0.0914 - val_mse: 0.0176 - val_r_square: 0.8806\n",
      "Epoch 11/100\n",
      "35456/35456 [==============================] - 6s 155us/step - loss: 0.0039 - acc: 0.1749 - rmse: 0.0561 - mse: 0.0039 - r_square: 0.9750 - val_loss: 0.0177 - val_acc: 0.0321 - val_rmse: 0.0917 - val_mse: 0.0177 - val_r_square: 0.8794\n",
      "Epoch 12/100\n",
      "35456/35456 [==============================] - 5s 155us/step - loss: 0.0043 - acc: 0.1476 - rmse: 0.0585 - mse: 0.0043 - r_square: 0.9723 - val_loss: 0.0178 - val_acc: 0.0461 - val_rmse: 0.0909 - val_mse: 0.0178 - val_r_square: 0.8791\n",
      "Epoch 13/100\n",
      "35456/35456 [==============================] - 6s 156us/step - loss: 0.0036 - acc: 0.1201 - rmse: 0.0536 - mse: 0.0036 - r_square: 0.9762 - val_loss: 0.0209 - val_acc: 0.0022 - val_rmse: 0.0937 - val_mse: 0.0209 - val_r_square: 0.8572\n",
      "Epoch 14/100\n",
      "35456/35456 [==============================] - 5s 155us/step - loss: 0.0043 - acc: 0.0579 - rmse: 0.0557 - mse: 0.0043 - r_square: 0.9709 - val_loss: 0.0184 - val_acc: 0.0351 - val_rmse: 0.0915 - val_mse: 0.0184 - val_r_square: 0.8748\n",
      "Epoch 15/100\n",
      "35456/35456 [==============================] - 5s 155us/step - loss: 0.0032 - acc: 0.0481 - rmse: 0.0494 - mse: 0.0032 - r_square: 0.9791 - val_loss: 0.0182 - val_acc: 0.0243 - val_rmse: 0.0889 - val_mse: 0.0182 - val_r_square: 0.8755\n",
      "Epoch 16/100\n",
      "35456/35456 [==============================] - 6s 156us/step - loss: 0.0035 - acc: 0.0798 - rmse: 0.0520 - mse: 0.0035 - r_square: 0.9766 - val_loss: 0.0186 - val_acc: 0.0613 - val_rmse: 0.0946 - val_mse: 0.0186 - val_r_square: 0.8731\n",
      "Epoch 17/100\n",
      "35456/35456 [==============================] - 6s 155us/step - loss: 0.0027 - acc: 0.1266 - rmse: 0.0468 - mse: 0.0027 - r_square: 0.9821 - val_loss: 0.0186 - val_acc: 0.2903 - val_rmse: 0.0909 - val_mse: 0.0186 - val_r_square: 0.8733\n",
      "Epoch 18/100\n",
      "35456/35456 [==============================] - 5s 155us/step - loss: 0.0030 - acc: 0.1160 - rmse: 0.0486 - mse: 0.0030 - r_square: 0.9803 - val_loss: 0.0176 - val_acc: 0.1041 - val_rmse: 0.0922 - val_mse: 0.0176 - val_r_square: 0.8798\n",
      "Epoch 19/100\n",
      "35456/35456 [==============================] - 5s 155us/step - loss: 0.0026 - acc: 0.1625 - rmse: 0.0451 - mse: 0.0026 - r_square: 0.9834 - val_loss: 0.0190 - val_acc: 0.3692 - val_rmse: 0.0909 - val_mse: 0.0190 - val_r_square: 0.8701\n",
      "Epoch 20/100\n",
      "35456/35456 [==============================] - 5s 155us/step - loss: 0.0028 - acc: 0.1475 - rmse: 0.0470 - mse: 0.0028 - r_square: 0.9813 - val_loss: 0.0173 - val_acc: 0.1062 - val_rmse: 0.0900 - val_mse: 0.0173 - val_r_square: 0.8818\n",
      "Epoch 21/100\n",
      "35456/35456 [==============================] - 6s 156us/step - loss: 0.0024 - acc: 0.1830 - rmse: 0.0434 - mse: 0.0024 - r_square: 0.9846 - val_loss: 0.0183 - val_acc: 0.2341 - val_rmse: 0.0889 - val_mse: 0.0183 - val_r_square: 0.8747\n",
      "Epoch 22/100\n",
      "35456/35456 [==============================] - 6s 156us/step - loss: 0.0026 - acc: 0.1892 - rmse: 0.0453 - mse: 0.0026 - r_square: 0.9831 - val_loss: 0.0173 - val_acc: 0.1015 - val_rmse: 0.0883 - val_mse: 0.0173 - val_r_square: 0.8822\n",
      "Epoch 23/100\n",
      "35456/35456 [==============================] - 6s 156us/step - loss: 0.0023 - acc: 0.2204 - rmse: 0.0425 - mse: 0.0023 - r_square: 0.9851 - val_loss: 0.0185 - val_acc: 0.0647 - val_rmse: 0.0885 - val_mse: 0.0185 - val_r_square: 0.8736\n",
      "Epoch 24/100\n",
      "35456/35456 [==============================] - 6s 155us/step - loss: 0.0026 - acc: 0.1966 - rmse: 0.0451 - mse: 0.0026 - r_square: 0.9828 - val_loss: 0.0175 - val_acc: 0.0120 - val_rmse: 0.0883 - val_mse: 0.0175 - val_r_square: 0.8806\n",
      "Epoch 25/100\n",
      "35456/35456 [==============================] - 6s 156us/step - loss: 0.0022 - acc: 0.2130 - rmse: 0.0417 - mse: 0.0022 - r_square: 0.9854 - val_loss: 0.0184 - val_acc: 0.0113 - val_rmse: 0.0886 - val_mse: 0.0184 - val_r_square: 0.8741\n",
      "Epoch 26/100\n",
      "35456/35456 [==============================] - 6s 157us/step - loss: 0.0026 - acc: 0.1628 - rmse: 0.0448 - mse: 0.0026 - r_square: 0.9826 - val_loss: 0.0178 - val_acc: 0.0012 - val_rmse: 0.0894 - val_mse: 0.0178 - val_r_square: 0.8782\n",
      "Epoch 27/100\n",
      "35456/35456 [==============================] - 5s 154us/step - loss: 0.0022 - acc: 0.1508 - rmse: 0.0414 - mse: 0.0022 - r_square: 0.9857 - val_loss: 0.0179 - val_acc: 0.0031 - val_rmse: 0.0883 - val_mse: 0.0179 - val_r_square: 0.8779\n",
      "Epoch 28/100\n",
      "35456/35456 [==============================] - 5s 153us/step - loss: 0.0025 - acc: 0.1529 - rmse: 0.0440 - mse: 0.0025 - r_square: 0.9834 - val_loss: 0.0179 - val_acc: 1.1282e-04 - val_rmse: 0.0901 - val_mse: 0.0179 - val_r_square: 0.8775\n",
      "Epoch 29/100\n",
      "35456/35456 [==============================] - 5s 153us/step - loss: 0.0022 - acc: 0.0968 - rmse: 0.0413 - mse: 0.0022 - r_square: 0.9859 - val_loss: 0.0175 - val_acc: 7.6151e-04 - val_rmse: 0.0881 - val_mse: 0.0175 - val_r_square: 0.8803\n",
      "Epoch 30/100\n",
      "35456/35456 [==============================] - 5s 154us/step - loss: 0.0024 - acc: 0.1282 - rmse: 0.0433 - mse: 0.0024 - r_square: 0.9843 - val_loss: 0.0177 - val_acc: 1.4102e-04 - val_rmse: 0.0895 - val_mse: 0.0177 - val_r_square: 0.8790\n",
      "Epoch 31/100\n",
      "35456/35456 [==============================] - 5s 154us/step - loss: 0.0022 - acc: 0.0933 - rmse: 0.0412 - mse: 0.0022 - r_square: 0.9860 - val_loss: 0.0177 - val_acc: 7.0510e-04 - val_rmse: 0.0883 - val_mse: 0.0177 - val_r_square: 0.8794\n",
      "Epoch 32/100\n",
      "35456/35456 [==============================] - 5s 154us/step - loss: 0.0023 - acc: 0.1287 - rmse: 0.0427 - mse: 0.0023 - r_square: 0.9846 - val_loss: 0.0175 - val_acc: 1.1282e-04 - val_rmse: 0.0890 - val_mse: 0.0175 - val_r_square: 0.8807\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 33/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0022 - acc: 0.0854 - rmse: 0.0416 - mse: 0.0022 - r_square: 0.9856 - val_loss: 0.0181 - val_acc: 2.8204e-04 - val_rmse: 0.0887 - val_mse: 0.0181 - val_r_square: 0.8766\n",
      "Epoch 34/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0023 - acc: 0.1245 - rmse: 0.0422 - mse: 0.0023 - r_square: 0.9849 - val_loss: 0.0173 - val_acc: 0.0000e+00 - val_rmse: 0.0872 - val_mse: 0.0173 - val_r_square: 0.8821\n",
      "Epoch 35/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0022 - acc: 0.0996 - rmse: 0.0411 - mse: 0.0022 - r_square: 0.9860 - val_loss: 0.0173 - val_acc: 2.8204e-05 - val_rmse: 0.0860 - val_mse: 0.0173 - val_r_square: 0.8816\n",
      "Epoch 36/100\n",
      "35456/35456 [==============================] - 5s 142us/step - loss: 0.0021 - acc: 0.0809 - rmse: 0.0407 - mse: 0.0021 - r_square: 0.9864 - val_loss: 0.0177 - val_acc: 5.6408e-04 - val_rmse: 0.0856 - val_mse: 0.0177 - val_r_square: 0.8791\n",
      "Epoch 37/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0021 - acc: 0.1162 - rmse: 0.0404 - mse: 0.0021 - r_square: 0.9865 - val_loss: 0.0171 - val_acc: 5.6408e-04 - val_rmse: 0.0841 - val_mse: 0.0171 - val_r_square: 0.8832\n",
      "Epoch 38/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0021 - acc: 0.1045 - rmse: 0.0405 - mse: 0.0021 - r_square: 0.9860 - val_loss: 0.0193 - val_acc: 0.0250 - val_rmse: 0.0869 - val_mse: 0.0193 - val_r_square: 0.8678\n",
      "Epoch 39/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0024 - acc: 0.0988 - rmse: 0.0415 - mse: 0.0024 - r_square: 0.9838 - val_loss: 0.0175 - val_acc: 0.0078 - val_rmse: 0.0844 - val_mse: 0.0175 - val_r_square: 0.8801\n",
      "Epoch 40/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0019 - acc: 0.1264 - rmse: 0.0383 - mse: 0.0019 - r_square: 0.9874 - val_loss: 0.0180 - val_acc: 0.0308 - val_rmse: 0.0838 - val_mse: 0.0180 - val_r_square: 0.8767\n",
      "Epoch 41/100\n",
      "35456/35456 [==============================] - 5s 144us/step - loss: 0.0021 - acc: 0.1171 - rmse: 0.0396 - mse: 0.0021 - r_square: 0.9857 - val_loss: 0.0182 - val_acc: 0.0488 - val_rmse: 0.0851 - val_mse: 0.0182 - val_r_square: 0.8753\n",
      "Epoch 42/100\n",
      "35456/35456 [==============================] - 5s 144us/step - loss: 0.0019 - acc: 0.1845 - rmse: 0.0382 - mse: 0.0019 - r_square: 0.9873 - val_loss: 0.0169 - val_acc: 0.0240 - val_rmse: 0.0818 - val_mse: 0.0169 - val_r_square: 0.8846\n",
      "Epoch 43/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0021 - acc: 0.1851 - rmse: 0.0396 - mse: 0.0021 - r_square: 0.9858 - val_loss: 0.0184 - val_acc: 0.0829 - val_rmse: 0.0854 - val_mse: 0.0184 - val_r_square: 0.8739\n",
      "Epoch 44/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0019 - acc: 0.1914 - rmse: 0.0378 - mse: 0.0019 - r_square: 0.9878 - val_loss: 0.0168 - val_acc: 0.0069 - val_rmse: 0.0817 - val_mse: 0.0168 - val_r_square: 0.8852\n",
      "Epoch 45/100\n",
      "35456/35456 [==============================] - 5s 144us/step - loss: 0.0018 - acc: 0.1677 - rmse: 0.0377 - mse: 0.0018 - r_square: 0.9879 - val_loss: 0.0174 - val_acc: 4.5126e-04 - val_rmse: 0.0837 - val_mse: 0.0174 - val_r_square: 0.8806\n",
      "Epoch 46/100\n",
      "35456/35456 [==============================] - 5s 144us/step - loss: 0.0018 - acc: 0.1757 - rmse: 0.0373 - mse: 0.0018 - r_square: 0.9884 - val_loss: 0.0168 - val_acc: 1.4102e-04 - val_rmse: 0.0824 - val_mse: 0.0168 - val_r_square: 0.8847\n",
      "Epoch 47/100\n",
      "35456/35456 [==============================] - 5s 144us/step - loss: 0.0018 - acc: 0.1522 - rmse: 0.0370 - mse: 0.0018 - r_square: 0.9886 - val_loss: 0.0171 - val_acc: 0.0000e+00 - val_rmse: 0.0825 - val_mse: 0.0171 - val_r_square: 0.8832\n",
      "Epoch 48/100\n",
      "35456/35456 [==============================] - 5s 144us/step - loss: 0.0018 - acc: 0.1316 - rmse: 0.0372 - mse: 0.0018 - r_square: 0.9884 - val_loss: 0.0171 - val_acc: 2.2563e-04 - val_rmse: 0.0823 - val_mse: 0.0171 - val_r_square: 0.8828\n",
      "Epoch 49/100\n",
      "35456/35456 [==============================] - 5s 148us/step - loss: 0.0018 - acc: 0.1077 - rmse: 0.0369 - mse: 0.0018 - r_square: 0.9883 - val_loss: 0.0170 - val_acc: 0.1086 - val_rmse: 0.0821 - val_mse: 0.0170 - val_r_square: 0.8837\n",
      "Epoch 50/100\n",
      "35456/35456 [==============================] - 5s 145us/step - loss: 0.0018 - acc: 0.1763 - rmse: 0.0369 - mse: 0.0018 - r_square: 0.9885 - val_loss: 0.0170 - val_acc: 0.1926 - val_rmse: 0.0814 - val_mse: 0.0170 - val_r_square: 0.8837\n",
      "Epoch 51/100\n",
      "35456/35456 [==============================] - 5s 144us/step - loss: 0.0017 - acc: 0.1113 - rmse: 0.0366 - mse: 0.0017 - r_square: 0.9885 - val_loss: 0.0170 - val_acc: 0.5639 - val_rmse: 0.0817 - val_mse: 0.0170 - val_r_square: 0.8832\n",
      "Epoch 52/100\n",
      "35456/35456 [==============================] - 5s 144us/step - loss: 0.0017 - acc: 0.2002 - rmse: 0.0363 - mse: 0.0017 - r_square: 0.9887 - val_loss: 0.0167 - val_acc: 0.6315 - val_rmse: 0.0799 - val_mse: 0.0167 - val_r_square: 0.8856\n",
      "Epoch 53/100\n",
      "35456/35456 [==============================] - 5s 144us/step - loss: 0.0018 - acc: 0.1545 - rmse: 0.0367 - mse: 0.0018 - r_square: 0.9884 - val_loss: 0.0174 - val_acc: 0.7369 - val_rmse: 0.0818 - val_mse: 0.0174 - val_r_square: 0.8808\n",
      "Epoch 54/100\n",
      "35456/35456 [==============================] - 5s 145us/step - loss: 0.0018 - acc: 0.1918 - rmse: 0.0363 - mse: 0.0018 - r_square: 0.9885 - val_loss: 0.0163 - val_acc: 0.6677 - val_rmse: 0.0785 - val_mse: 0.0163 - val_r_square: 0.8882\n",
      "Epoch 55/100\n",
      "35456/35456 [==============================] - 5s 145us/step - loss: 0.0018 - acc: 0.2200 - rmse: 0.0373 - mse: 0.0018 - r_square: 0.9881 - val_loss: 0.0182 - val_acc: 0.7660 - val_rmse: 0.0819 - val_mse: 0.0182 - val_r_square: 0.8749\n",
      "Epoch 56/100\n",
      "35456/35456 [==============================] - 5s 145us/step - loss: 0.0019 - acc: 0.2170 - rmse: 0.0373 - mse: 0.0019 - r_square: 0.9873 - val_loss: 0.0164 - val_acc: 0.7139 - val_rmse: 0.0775 - val_mse: 0.0164 - val_r_square: 0.8875\n",
      "Epoch 57/100\n",
      "35456/35456 [==============================] - 5s 155us/step - loss: 0.0018 - acc: 0.2924 - rmse: 0.0361 - mse: 0.0018 - r_square: 0.9884 - val_loss: 0.0187 - val_acc: 0.7792 - val_rmse: 0.0806 - val_mse: 0.0187 - val_r_square: 0.8714\n",
      "Epoch 58/100\n",
      "35456/35456 [==============================] - 5s 148us/step - loss: 0.0020 - acc: 0.2535 - rmse: 0.0377 - mse: 0.0020 - r_square: 0.9863 - val_loss: 0.0172 - val_acc: 0.3211 - val_rmse: 0.0792 - val_mse: 0.0172 - val_r_square: 0.8820\n",
      "Epoch 59/100\n",
      "35456/35456 [==============================] - 5s 145us/step - loss: 0.0016 - acc: 0.2713 - rmse: 0.0346 - mse: 0.0016 - r_square: 0.9895 - val_loss: 0.0161 - val_acc: 0.6601 - val_rmse: 0.0762 - val_mse: 0.0161 - val_r_square: 0.8892\n",
      "Epoch 60/100\n",
      "35456/35456 [==============================] - 5s 145us/step - loss: 0.0016 - acc: 0.2356 - rmse: 0.0350 - mse: 0.0016 - r_square: 0.9892 - val_loss: 0.0173 - val_acc: 0.2335 - val_rmse: 0.0791 - val_mse: 0.0173 - val_r_square: 0.8814\n",
      "Epoch 61/100\n",
      "35456/35456 [==============================] - 5s 146us/step - loss: 0.0015 - acc: 0.2733 - rmse: 0.0335 - mse: 0.0015 - r_square: 0.9902 - val_loss: 0.0159 - val_acc: 0.2728 - val_rmse: 0.0759 - val_mse: 0.0159 - val_r_square: 0.8911\n",
      "Epoch 62/100\n",
      "35456/35456 [==============================] - 5s 147us/step - loss: 0.0015 - acc: 0.2681 - rmse: 0.0334 - mse: 0.0015 - r_square: 0.9903 - val_loss: 0.0169 - val_acc: 0.1882 - val_rmse: 0.0782 - val_mse: 0.0169 - val_r_square: 0.8837\n",
      "Epoch 63/100\n",
      "35456/35456 [==============================] - 5s 145us/step - loss: 0.0015 - acc: 0.2464 - rmse: 0.0332 - mse: 0.0015 - r_square: 0.9904 - val_loss: 0.0160 - val_acc: 0.1420 - val_rmse: 0.0769 - val_mse: 0.0160 - val_r_square: 0.8902\n",
      "Epoch 64/100\n",
      "35456/35456 [==============================] - 5s 146us/step - loss: 0.0015 - acc: 0.2686 - rmse: 0.0332 - mse: 0.0015 - r_square: 0.9904 - val_loss: 0.0169 - val_acc: 0.1374 - val_rmse: 0.0781 - val_mse: 0.0169 - val_r_square: 0.8836\n",
      "Epoch 65/100\n",
      "35456/35456 [==============================] - 5s 146us/step - loss: 0.0016 - acc: 0.2308 - rmse: 0.0344 - mse: 0.0016 - r_square: 0.9898 - val_loss: 0.0164 - val_acc: 0.0324 - val_rmse: 0.0790 - val_mse: 0.0164 - val_r_square: 0.8872\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 66/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0015 - acc: 0.2566 - rmse: 0.0338 - mse: 0.0015 - r_square: 0.9898 - val_loss: 0.0169 - val_acc: 0.0962 - val_rmse: 0.0782 - val_mse: 0.0169 - val_r_square: 0.8837\n",
      "Epoch 67/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0017 - acc: 0.2184 - rmse: 0.0352 - mse: 0.0017 - r_square: 0.9891 - val_loss: 0.0167 - val_acc: 0.0236 - val_rmse: 0.0791 - val_mse: 0.0167 - val_r_square: 0.8851\n",
      "Epoch 68/100\n",
      "35456/35456 [==============================] - 5s 144us/step - loss: 0.0015 - acc: 0.2624 - rmse: 0.0337 - mse: 0.0015 - r_square: 0.9898 - val_loss: 0.0165 - val_acc: 0.1220 - val_rmse: 0.0779 - val_mse: 0.0165 - val_r_square: 0.8868\n",
      "Epoch 69/100\n",
      "35456/35456 [==============================] - 5s 145us/step - loss: 0.0015 - acc: 0.2186 - rmse: 0.0336 - mse: 0.0015 - r_square: 0.9902 - val_loss: 0.0166 - val_acc: 0.2018 - val_rmse: 0.0781 - val_mse: 0.0166 - val_r_square: 0.8858\n",
      "Epoch 70/100\n",
      "35456/35456 [==============================] - 5s 145us/step - loss: 0.0015 - acc: 0.2357 - rmse: 0.0334 - mse: 0.0015 - r_square: 0.9903 - val_loss: 0.0166 - val_acc: 0.4431 - val_rmse: 0.0783 - val_mse: 0.0166 - val_r_square: 0.8859\n",
      "Epoch 71/100\n",
      "35456/35456 [==============================] - 5s 144us/step - loss: 0.0015 - acc: 0.2466 - rmse: 0.0334 - mse: 0.0015 - r_square: 0.9902 - val_loss: 0.0166 - val_acc: 0.7840 - val_rmse: 0.0773 - val_mse: 0.0166 - val_r_square: 0.8859\n",
      "Epoch 72/100\n",
      "35456/35456 [==============================] - 5s 146us/step - loss: 0.0016 - acc: 0.2428 - rmse: 0.0342 - mse: 0.0016 - r_square: 0.9896 - val_loss: 0.0168 - val_acc: 0.7387 - val_rmse: 0.0786 - val_mse: 0.0168 - val_r_square: 0.8843\n",
      "Epoch 73/100\n",
      "35456/35456 [==============================] - 5s 145us/step - loss: 0.0015 - acc: 0.3060 - rmse: 0.0335 - mse: 0.0015 - r_square: 0.9899 - val_loss: 0.0161 - val_acc: 0.8428 - val_rmse: 0.0758 - val_mse: 0.0161 - val_r_square: 0.8891\n",
      "Epoch 74/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0014 - acc: 0.2340 - rmse: 0.0328 - mse: 0.0014 - r_square: 0.9905 - val_loss: 0.0165 - val_acc: 0.8562 - val_rmse: 0.0764 - val_mse: 0.0165 - val_r_square: 0.8865\n",
      "Epoch 75/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0014 - acc: 0.2915 - rmse: 0.0322 - mse: 0.0014 - r_square: 0.9909 - val_loss: 0.0158 - val_acc: 0.8443 - val_rmse: 0.0740 - val_mse: 0.0158 - val_r_square: 0.8915\n",
      "Epoch 76/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0014 - acc: 0.2325 - rmse: 0.0320 - mse: 0.0014 - r_square: 0.9909 - val_loss: 0.0164 - val_acc: 0.8406 - val_rmse: 0.0758 - val_mse: 0.0164 - val_r_square: 0.8872\n",
      "Epoch 77/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0014 - acc: 0.1724 - rmse: 0.0321 - mse: 0.0014 - r_square: 0.9909 - val_loss: 0.0155 - val_acc: 0.1002 - val_rmse: 0.0735 - val_mse: 0.0155 - val_r_square: 0.8934\n",
      "Epoch 78/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0015 - acc: 0.1890 - rmse: 0.0327 - mse: 0.0015 - r_square: 0.9905 - val_loss: 0.0168 - val_acc: 0.6369 - val_rmse: 0.0766 - val_mse: 0.0168 - val_r_square: 0.8841\n",
      "Epoch 79/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0016 - acc: 0.1223 - rmse: 0.0336 - mse: 0.0016 - r_square: 0.9897 - val_loss: 0.0158 - val_acc: 0.1094 - val_rmse: 0.0743 - val_mse: 0.0158 - val_r_square: 0.8914\n",
      "Epoch 80/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0016 - acc: 0.1678 - rmse: 0.0343 - mse: 0.0016 - r_square: 0.9894 - val_loss: 0.0176 - val_acc: 0.4168 - val_rmse: 0.0763 - val_mse: 0.0176 - val_r_square: 0.8785\n",
      "Epoch 81/100\n",
      "35456/35456 [==============================] - 5s 145us/step - loss: 0.0019 - acc: 0.1382 - rmse: 0.0360 - mse: 0.0019 - r_square: 0.9876 - val_loss: 0.0164 - val_acc: 0.2979 - val_rmse: 0.0753 - val_mse: 0.0164 - val_r_square: 0.8874\n",
      "Epoch 82/100\n",
      "35456/35456 [==============================] - 5s 144us/step - loss: 0.0014 - acc: 0.1946 - rmse: 0.0324 - mse: 0.0014 - r_square: 0.9906 - val_loss: 0.0161 - val_acc: 0.0842 - val_rmse: 0.0737 - val_mse: 0.0161 - val_r_square: 0.8894\n",
      "Epoch 83/100\n",
      "35456/35456 [==============================] - 5s 144us/step - loss: 0.0014 - acc: 0.2139 - rmse: 0.0321 - mse: 0.0014 - r_square: 0.9908 - val_loss: 0.0164 - val_acc: 0.2283 - val_rmse: 0.0755 - val_mse: 0.0164 - val_r_square: 0.8874\n",
      "Epoch 84/100\n",
      "35456/35456 [==============================] - 5s 145us/step - loss: 0.0013 - acc: 0.2061 - rmse: 0.0313 - mse: 0.0013 - r_square: 0.9913 - val_loss: 0.0158 - val_acc: 0.1830 - val_rmse: 0.0744 - val_mse: 0.0158 - val_r_square: 0.8910\n",
      "Epoch 85/100\n",
      "35456/35456 [==============================] - 5s 144us/step - loss: 0.0013 - acc: 0.1954 - rmse: 0.0312 - mse: 0.0013 - r_square: 0.9914 - val_loss: 0.0162 - val_acc: 0.1228 - val_rmse: 0.0751 - val_mse: 0.0162 - val_r_square: 0.8885\n",
      "Epoch 86/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0013 - acc: 0.1753 - rmse: 0.0305 - mse: 0.0013 - r_square: 0.9917 - val_loss: 0.0158 - val_acc: 0.0930 - val_rmse: 0.0746 - val_mse: 0.0158 - val_r_square: 0.8911\n",
      "Epoch 87/100\n",
      "35456/35456 [==============================] - 5s 145us/step - loss: 0.0013 - acc: 0.1648 - rmse: 0.0308 - mse: 0.0013 - r_square: 0.9916 - val_loss: 0.0162 - val_acc: 0.0922 - val_rmse: 0.0754 - val_mse: 0.0162 - val_r_square: 0.8886\n",
      "Epoch 88/100\n",
      "35456/35456 [==============================] - 5s 144us/step - loss: 0.0013 - acc: 0.1506 - rmse: 0.0305 - mse: 0.0013 - r_square: 0.9917 - val_loss: 0.0159 - val_acc: 0.0740 - val_rmse: 0.0747 - val_mse: 0.0159 - val_r_square: 0.8908\n",
      "Epoch 89/100\n",
      "35456/35456 [==============================] - 5s 145us/step - loss: 0.0013 - acc: 0.1465 - rmse: 0.0309 - mse: 0.0013 - r_square: 0.9915 - val_loss: 0.0162 - val_acc: 0.0908 - val_rmse: 0.0757 - val_mse: 0.0162 - val_r_square: 0.8885\n",
      "Epoch 90/100\n",
      "35456/35456 [==============================] - 5s 145us/step - loss: 0.0013 - acc: 0.2150 - rmse: 0.0309 - mse: 0.0013 - r_square: 0.9914 - val_loss: 0.0161 - val_acc: 0.0116 - val_rmse: 0.0752 - val_mse: 0.0161 - val_r_square: 0.8891\n",
      "Epoch 91/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0014 - acc: 0.1772 - rmse: 0.0315 - mse: 0.0014 - r_square: 0.9911 - val_loss: 0.0162 - val_acc: 0.0841 - val_rmse: 0.0760 - val_mse: 0.0162 - val_r_square: 0.8884\n",
      "Epoch 92/100\n",
      "35456/35456 [==============================] - 5s 145us/step - loss: 0.0014 - acc: 0.2787 - rmse: 0.0315 - mse: 0.0014 - r_square: 0.9910 - val_loss: 0.0165 - val_acc: 0.0037 - val_rmse: 0.0760 - val_mse: 0.0165 - val_r_square: 0.8866\n",
      "Epoch 93/100\n",
      "35456/35456 [==============================] - 5s 144us/step - loss: 0.0014 - acc: 0.1355 - rmse: 0.0316 - mse: 0.0014 - r_square: 0.9908 - val_loss: 0.0163 - val_acc: 0.1188 - val_rmse: 0.0760 - val_mse: 0.0163 - val_r_square: 0.8878\n",
      "Epoch 94/100\n",
      "35456/35456 [==============================] - 5s 144us/step - loss: 0.0013 - acc: 0.2034 - rmse: 0.0306 - mse: 0.0013 - r_square: 0.9915 - val_loss: 0.0164 - val_acc: 0.0267 - val_rmse: 0.0755 - val_mse: 0.0164 - val_r_square: 0.8872\n",
      "Epoch 95/100\n",
      "35456/35456 [==============================] - 5s 145us/step - loss: 0.0013 - acc: 0.1083 - rmse: 0.0311 - mse: 0.0013 - r_square: 0.9913 - val_loss: 0.0163 - val_acc: 0.3168 - val_rmse: 0.0762 - val_mse: 0.0163 - val_r_square: 0.8877\n",
      "Epoch 96/100\n",
      "35456/35456 [==============================] - 5s 149us/step - loss: 0.0013 - acc: 0.1390 - rmse: 0.0305 - mse: 0.0013 - r_square: 0.9915 - val_loss: 0.0162 - val_acc: 0.3257 - val_rmse: 0.0754 - val_mse: 0.0162 - val_r_square: 0.8887\n",
      "Epoch 97/100\n",
      "35456/35456 [==============================] - 5s 145us/step - loss: 0.0013 - acc: 0.1086 - rmse: 0.0308 - mse: 0.0013 - r_square: 0.9914 - val_loss: 0.0166 - val_acc: 0.5400 - val_rmse: 0.0769 - val_mse: 0.0166 - val_r_square: 0.8862\n",
      "Epoch 98/100\n",
      "35456/35456 [==============================] - 5s 145us/step - loss: 0.0013 - acc: 0.2006 - rmse: 0.0308 - mse: 0.0013 - r_square: 0.9914 - val_loss: 0.0162 - val_acc: 0.7508 - val_rmse: 0.0759 - val_mse: 0.0162 - val_r_square: 0.8889\n",
      "Epoch 99/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "35456/35456 [==============================] - 5s 142us/step - loss: 0.0013 - acc: 0.1682 - rmse: 0.0306 - mse: 0.0013 - r_square: 0.9914 - val_loss: 0.0162 - val_acc: 0.6407 - val_rmse: 0.0769 - val_mse: 0.0162 - val_r_square: 0.8886\n",
      "Epoch 100/100\n",
      "35456/35456 [==============================] - 5s 143us/step - loss: 0.0013 - acc: 0.1947 - rmse: 0.0310 - mse: 0.0013 - r_square: 0.9914 - val_loss: 0.0165 - val_acc: 0.6126 - val_rmse: 0.0768 - val_mse: 0.0165 - val_r_square: 0.8865\n"
     ]
    }
   ],
   "source": [
    "input_shape = (2, data.shape[1])\n",
    "model = seq2seq(data.shape[1], 4, input_shape)\n",
    "model.compile(loss='mse', optimizer='adam',metrics=['acc',rmse,mse,r_square])\n",
    "# mlp overfit on the moons dataset with patient early stopping and model checkpointing\n",
    "from sklearn.datasets import make_moons\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras.callbacks import EarlyStopping,ReduceLROnPlateau\n",
    "reduce_lr_acc = ReduceLROnPlateau(monitor='val_acc', factor=0.3, patience=7, verbose=1, min_delta=1e-4, mode='max')\n",
    "earlyStopping = EarlyStopping(monitor='val_acc', patience=100, verbose=0, mode='max')\n",
    "history = model.fit(x_train, y_train, batch_size=256,epochs=100,shuffle=False ,validation_data=(x_validate, y_validate))"
   ]
  },
  {
   "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": "3.8.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}