Forecast Energy usage of households
Create New

Forecast Energy usage of households



Forecast Energy usage of households
Forecast Energy usage of households

Project Rating

Overall Rating
2
0
0
0
0

         In this project, we want to analyze energy consumption in large households over a period of time. This will help in the effective planning and operation of power systems, developing optimal forecasting tools for energy operators to maximize profit and also to provide maximum satisfaction to energy consumers. For this, data for electric energy consumed in a city was collected for a particular period of time. Data was trained and the proposed model was validated using long short-term memory (LSTM) techniques. The model has been tested with actual energy consumption data and verified if this yields satisfactory performance. 

 

Why: Problem statement

In this project, we want to analyse the energy consumption in large households over a period of time. This will help in effective planning and operation of power systems, developing optimal forecasting tools for energy operators to maximize profit and also to provide maximum satisfaction to energy consumers. For this, data for electric energy consumed in a city was collected for a particular period of time. Data was trained and the proposed model was validated using long short-term memory (LSTM) techniques. The model has been tested with actual energy consumption data and verified if this yields satisfactory performance.

Due to increasing globalization and industrialization, energy consumption has been found to increase constantly. It has been found that buildings are responsible for the biggest proportion of energy consumption; for example in India and China, energy consumption in buildings represents around 40% of the total energy consumption. In order to control energy consumption in buildings, different policies have been proposed, from utilizing bioclimatic architecture to the use of predictive models within control approaches. Here we talk about using an artificial neural network approach for forecasting electric energy consumption. 

How: Solution description

We adapt two deep neural network architectures to energy disaggregation: 1) long short-term memory (LSTM); and 2) a network that regresses the start time, end time and average power demand of each appliance called Multilinear regression Algorithm. We use five metrics like ID, plans, date, time and meter readings to test the performance of these algorithms on real aggregate power data from appliances. Tests are performed against a house not seen during training and against houses seen during training. 

Models/ Algorithms proposed 

We use long short-term memory and Multilinear regression algorithm for analysis of electric energy consumption of the top households with the highest number of samples on an hourly basis based on the previous usage pattern. The major features taken for analysis include ID, plans, date, time and meter readings in KiloWatts per Hour. 

General Assumptions: We analyse that certain households consume more electric energy based on the usage of general appliances.

Technical Assumptions: We analyse if a particular appliance or group of appliances consume more energy compared to other appliances.

 

How is it different from competition

  Residential load forecasting has been playing an increasingly important role in developed cities. Due to the variability of residents’ activities, individual residential loads are usually too volatile to forecast accurately. It is shown that the forecasting accuracy can be notably improved by including appliance measurements in the training data. For this, energy consumption from residential households was analyzed using long short-term memory (LSTM) and Multilinear regression algorithm analysis. The effectiveness of the proposed method is validated through extensive comparison studies on a real-world dataset. We are not considering energy consumption from suburban areas which also consume substantial amount of electricity. Also, energy consumption from individual households are not taken into account.

Who are your customers

Common people and Industries 

Project Phases and Schedule

Phase 1: Data collection

Phase 2: Data cleaning 

Phase 3: Algorithm development

Phase 4: prediction 

Resources Required

  • Software: Anaconda tool 
  • Language: Python 3
  • Model: LSTM Model 
  • Algorithm: Multilinear regression algorithm 
  • Libraries used : Numpy, Pandas, Matplotlib, LSTM library 

Download:
Project Code Code copy
/* Your file Name : Energy_usage_prediction.ipynb */
/* Your coding Language : python */
/* Your code snippet start here */
{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "4941/4941 [==============================] - 9s 2ms/step - loss: 2.3036 - mean_absolute_error: 2.3036 - mean_absolute_percentage_error: 958087.2510\n",
      "Epoch 2/20\n",
      "4941/4941 [==============================] - 6s 1ms/step - loss: 1.2791 - mean_absolute_error: 1.2791 - mean_absolute_percentage_error: 1347564.1791\n",
      "Epoch 3/20\n",
      "4941/4941 [==============================] - 6s 1ms/step - loss: 1.2430 - mean_absolute_error: 1.2430 - mean_absolute_percentage_error: 1390990.4055\n",
      "Epoch 4/20\n",
      "4941/4941 [==============================] - 6s 1ms/step - loss: 1.2474 - mean_absolute_error: 1.2474 - mean_absolute_percentage_error: 1567744.6239\n",
      "Epoch 5/20\n",
      "4941/4941 [==============================] - 6s 1ms/step - loss: 1.2387 - mean_absolute_error: 1.2387 - mean_absolute_percentage_error: 1523228.4351\n",
      "Epoch 6/20\n",
      "4941/4941 [==============================] - 5s 1ms/step - loss: 1.2190 - mean_absolute_error: 1.2190 - mean_absolute_percentage_error: 1595154.4771\n",
      "Epoch 7/20\n",
      "4941/4941 [==============================] - 6s 1ms/step - loss: 1.2163 - mean_absolute_error: 1.2163 - mean_absolute_percentage_error: 1626107.8433\n",
      "Epoch 8/20\n",
      "4941/4941 [==============================] - 5s 1ms/step - loss: 1.2516 - mean_absolute_error: 1.2516 - mean_absolute_percentage_error: 1568342.0659\n",
      "Epoch 9/20\n",
      "4941/4941 [==============================] - 5s 1ms/step - loss: 1.1946 - mean_absolute_error: 1.1946 - mean_absolute_percentage_error: 1423129.3533\n",
      "Epoch 10/20\n",
      "4941/4941 [==============================] - 5s 1ms/step - loss: 1.2003 - mean_absolute_error: 1.2003 - mean_absolute_percentage_error: 1475296.8006\n",
      "Epoch 11/20\n",
      "4941/4941 [==============================] - 5s 1ms/step - loss: 1.2054 - mean_absolute_error: 1.2054 - mean_absolute_percentage_error: 1498650.4724\n",
      "Epoch 12/20\n",
      "4941/4941 [==============================] - 5s 1ms/step - loss: 1.2124 - mean_absolute_error: 1.2124 - mean_absolute_percentage_error: 1686772.0570\n",
      "Epoch 13/20\n",
      "4941/4941 [==============================] - 5s 1ms/step - loss: 1.2073 - mean_absolute_error: 1.2073 - mean_absolute_percentage_error: 1666159.9720\n",
      "Epoch 14/20\n",
      "4941/4941 [==============================] - 5s 1ms/step - loss: 1.1960 - mean_absolute_error: 1.1960 - mean_absolute_percentage_error: 1595460.3351\n",
      "Epoch 15/20\n",
      "4941/4941 [==============================] - 6s 1ms/step - loss: 1.1979 - mean_absolute_error: 1.1979 - mean_absolute_percentage_error: 1804018.6608\n",
      "Epoch 16/20\n",
      "4941/4941 [==============================] - 5s 1ms/step - loss: 1.2098 - mean_absolute_error: 1.2098 - mean_absolute_percentage_error: 1561521.9452\n",
      "Epoch 17/20\n",
      "4941/4941 [==============================] - 6s 1ms/step - loss: 1.1859 - mean_absolute_error: 1.1859 - mean_absolute_percentage_error: 1675428.1582\n",
      "Epoch 18/20\n",
      "4941/4941 [==============================] - 6s 1ms/step - loss: 1.1819 - mean_absolute_error: 1.1819 - mean_absolute_percentage_error: 1685081.5631\n",
      "Epoch 19/20\n",
      "4941/4941 [==============================] - 6s 1ms/step - loss: 1.1998 - mean_absolute_error: 1.1998 - mean_absolute_percentage_error: 1767809.2131\n",
      "Epoch 20/20\n",
      "4941/4941 [==============================] - 6s 1ms/step - loss: 1.2062 - mean_absolute_error: 1.2062 - mean_absolute_percentage_error: 1639093.6553\n",
      "550/550 [==============================] - 2s 3ms/step\n",
      "Epoch 1/20\n",
      "4942/4942 [==============================] - 10s 2ms/step - loss: 2.1671 - mean_absolute_error: 2.1671 - mean_absolute_percentage_error: 1143005.4167\n",
      "Epoch 2/20\n",
      "4942/4942 [==============================] - 7s 1ms/step - loss: 1.2824 - mean_absolute_error: 1.2824 - mean_absolute_percentage_error: 839380.2460\n",
      "Epoch 3/20\n",
      "4942/4942 [==============================] - 8s 2ms/step - loss: 1.2475 - mean_absolute_error: 1.2475 - mean_absolute_percentage_error: 916040.6669\n",
      "Epoch 4/20\n",
      "4942/4942 [==============================] - 7s 1ms/step - loss: 1.2429 - mean_absolute_error: 1.2429 - mean_absolute_percentage_error: 1263264.4392\n",
      "Epoch 5/20\n",
      "4942/4942 [==============================] - 7s 1ms/step - loss: 1.2030 - mean_absolute_error: 1.2030 - mean_absolute_percentage_error: 1130597.8666\n",
      "Epoch 6/20\n",
      "4942/4942 [==============================] - 7s 1ms/step - loss: 1.2147 - mean_absolute_error: 1.2147 - mean_absolute_percentage_error: 1042162.3122\n",
      "Epoch 7/20\n",
      "4942/4942 [==============================] - 7s 1ms/step - loss: 1.2360 - mean_absolute_error: 1.2360 - mean_absolute_percentage_error: 1025985.5136\n",
      "Epoch 8/20\n",
      "4942/4942 [==============================] - 7s 1ms/step - loss: 1.2213 - mean_absolute_error: 1.2213 - mean_absolute_percentage_error: 854871.7015\n",
      "Epoch 9/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2030 - mean_absolute_error: 1.2030 - mean_absolute_percentage_error: 1070723.3570\n",
      "Epoch 10/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2100 - mean_absolute_error: 1.2100 - mean_absolute_percentage_error: 1165382.3719\n",
      "Epoch 11/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2109 - mean_absolute_error: 1.2109 - mean_absolute_percentage_error: 1137655.4446\n",
      "Epoch 12/20\n",
      "4942/4942 [==============================] - 7s 1ms/step - loss: 1.2175 - mean_absolute_error: 1.2175 - mean_absolute_percentage_error: 1205465.6474\n",
      "Epoch 13/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2036 - mean_absolute_error: 1.2036 - mean_absolute_percentage_error: 1109000.0632\n",
      "Epoch 14/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1917 - mean_absolute_error: 1.1917 - mean_absolute_percentage_error: 1096029.0650\n",
      "Epoch 15/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1869 - mean_absolute_error: 1.1869 - mean_absolute_percentage_error: 1107820.0921\n",
      "Epoch 16/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1776 - mean_absolute_error: 1.1776 - mean_absolute_percentage_error: 1031655.8989\n",
      "Epoch 17/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2054 - mean_absolute_error: 1.2054 - mean_absolute_percentage_error: 1096490.3601\n",
      "Epoch 18/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.1878 - mean_absolute_error: 1.1878 - mean_absolute_percentage_error: 1108230.5376\n",
      "Epoch 19/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1825 - mean_absolute_error: 1.1825 - mean_absolute_percentage_error: 1006454.4162\n",
      "Epoch 20/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1839 - mean_absolute_error: 1.1839 - mean_absolute_percentage_error: 1256227.4337\n",
      "549/549 [==============================] - 1s 3ms/step\n",
      "Epoch 1/20\n",
      "4942/4942 [==============================] - 9s 2ms/step - loss: 2.2512 - mean_absolute_error: 2.2512 - mean_absolute_percentage_error: 851205.4632\n",
      "Epoch 2/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2891 - mean_absolute_error: 1.2891 - mean_absolute_percentage_error: 1240335.9637\n",
      "Epoch 3/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2311 - mean_absolute_error: 1.2311 - mean_absolute_percentage_error: 1420502.6207\n",
      "Epoch 4/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2252 - mean_absolute_error: 1.2252 - mean_absolute_percentage_error: 1621771.0300\n",
      "Epoch 5/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2119 - mean_absolute_error: 1.2119 - mean_absolute_percentage_error: 1535849.8630\n",
      "Epoch 6/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2363 - mean_absolute_error: 1.2363 - mean_absolute_percentage_error: 1447669.4566\n",
      "Epoch 7/20\n",
      "4942/4942 [==============================] - 7s 1ms/step - loss: 1.2120 - mean_absolute_error: 1.2120 - mean_absolute_percentage_error: 1717779.1071\n",
      "Epoch 8/20\n",
      "4942/4942 [==============================] - 7s 1ms/step - loss: 1.2188 - mean_absolute_error: 1.2188 - mean_absolute_percentage_error: 1533988.9414\n",
      "Epoch 9/20\n",
      "4942/4942 [==============================] - 8s 2ms/step - loss: 1.2088 - mean_absolute_error: 1.2088 - mean_absolute_percentage_error: 1648025.7270\n",
      "Epoch 10/20\n",
      "4942/4942 [==============================] - 9s 2ms/step - loss: 1.1879 - mean_absolute_error: 1.1879 - mean_absolute_percentage_error: 1494334.6159\n",
      "Epoch 11/20\n",
      "4942/4942 [==============================] - 8s 2ms/step - loss: 1.2139 - mean_absolute_error: 1.2139 - mean_absolute_percentage_error: 1683563.7424\n",
      "Epoch 12/20\n",
      "4942/4942 [==============================] - 8s 2ms/step - loss: 1.2262 - mean_absolute_error: 1.2262 - mean_absolute_percentage_error: 1505045.6311\n",
      "Epoch 13/20\n",
      "4942/4942 [==============================] - 7s 1ms/step - loss: 1.1944 - mean_absolute_error: 1.1944 - mean_absolute_percentage_error: 1475687.8672\n",
      "Epoch 14/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1886 - mean_absolute_error: 1.1886 - mean_absolute_percentage_error: 1520175.0712\n",
      "Epoch 15/20\n",
      "4942/4942 [==============================] - 7s 1ms/step - loss: 1.2018 - mean_absolute_error: 1.2018 - mean_absolute_percentage_error: 1579265.5445\n",
      "Epoch 16/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1870 - mean_absolute_error: 1.1870 - mean_absolute_percentage_error: 1515482.9699\n",
      "Epoch 17/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1881 - mean_absolute_error: 1.1881 - mean_absolute_percentage_error: 1716572.6518\n",
      "Epoch 18/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2053 - mean_absolute_error: 1.2053 - mean_absolute_percentage_error: 1381860.8933\n",
      "Epoch 19/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1716 - mean_absolute_error: 1.1716 - mean_absolute_percentage_error: 1629533.7124\n",
      "Epoch 20/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2009 - mean_absolute_error: 1.2009 - mean_absolute_percentage_error: 1747907.0087\n",
      "549/549 [==============================] - 2s 3ms/step\n",
      "Epoch 1/20\n",
      "4942/4942 [==============================] - 10s 2ms/step - loss: 2.3414 - mean_absolute_error: 2.3414 - mean_absolute_percentage_error: 1447326.6925\n",
      "Epoch 2/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2870 - mean_absolute_error: 1.2870 - mean_absolute_percentage_error: 778985.2606\n",
      "Epoch 3/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2452 - mean_absolute_error: 1.2452 - mean_absolute_percentage_error: 872967.0573\n",
      "Epoch 4/20\n",
      "4942/4942 [==============================] - 7s 2ms/step - loss: 1.2433 - mean_absolute_error: 1.2433 - mean_absolute_percentage_error: 977629.2591\n",
      "Epoch 5/20\n",
      "4942/4942 [==============================] - 7s 1ms/step - loss: 1.2171 - mean_absolute_error: 1.2171 - mean_absolute_percentage_error: 1135807.8099\n",
      "Epoch 6/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2093 - mean_absolute_error: 1.2093 - mean_absolute_percentage_error: 1023016.7983\n",
      "Epoch 7/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2323 - mean_absolute_error: 1.2323 - mean_absolute_percentage_error: 1095602.8101\n",
      "Epoch 8/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2039 - mean_absolute_error: 1.2039 - mean_absolute_percentage_error: 977805.6230\n",
      "Epoch 9/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2209 - mean_absolute_error: 1.2209 - mean_absolute_percentage_error: 1071091.8496\n",
      "Epoch 10/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2278 - mean_absolute_error: 1.2278 - mean_absolute_percentage_error: 1028599.0334\n",
      "Epoch 11/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.1937 - mean_absolute_error: 1.1937 - mean_absolute_percentage_error: 1289287.2909\n",
      "Epoch 12/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2121 - mean_absolute_error: 1.2121 - mean_absolute_percentage_error: 1113348.5269\n",
      "Epoch 13/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.1815 - mean_absolute_error: 1.1815 - mean_absolute_percentage_error: 1082312.2010\n",
      "Epoch 14/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2232 - mean_absolute_error: 1.2232 - mean_absolute_percentage_error: 978577.0254\n",
      "Epoch 15/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2014 - mean_absolute_error: 1.2014 - mean_absolute_percentage_error: 1060449.2574\n",
      "Epoch 16/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2197 - mean_absolute_error: 1.2197 - mean_absolute_percentage_error: 972684.4169\n",
      "Epoch 17/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.1806 - mean_absolute_error: 1.1806 - mean_absolute_percentage_error: 1015623.1604\n",
      "Epoch 18/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2045 - mean_absolute_error: 1.2045 - mean_absolute_percentage_error: 893191.7265\n",
      "Epoch 19/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.1818 - mean_absolute_error: 1.1818 - mean_absolute_percentage_error: 1188936.1663\n",
      "Epoch 20/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.1992 - mean_absolute_error: 1.1992 - mean_absolute_percentage_error: 1162818.1940\n",
      "549/549 [==============================] - 2s 3ms/step\n",
      "Epoch 1/20\n",
      "4942/4942 [==============================] - 9s 2ms/step - loss: 2.2649 - mean_absolute_error: 2.2649 - mean_absolute_percentage_error: 1246979.6331\n",
      "Epoch 2/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.3041 - mean_absolute_error: 1.3041 - mean_absolute_percentage_error: 1089864.5460\n",
      "Epoch 3/20\n",
      "4942/4942 [==============================] - 7s 1ms/step - loss: 1.2358 - mean_absolute_error: 1.2358 - mean_absolute_percentage_error: 1375058.2427\n",
      "Epoch 4/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2177 - mean_absolute_error: 1.2177 - mean_absolute_percentage_error: 1365810.5812\n",
      "Epoch 5/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2050 - mean_absolute_error: 1.2050 - mean_absolute_percentage_error: 1556208.0705\n",
      "Epoch 6/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2139 - mean_absolute_error: 1.2139 - mean_absolute_percentage_error: 1443203.0834\n",
      "Epoch 7/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2187 - mean_absolute_error: 1.2187 - mean_absolute_percentage_error: 1553986.9363\n",
      "Epoch 8/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2397 - mean_absolute_error: 1.2397 - mean_absolute_percentage_error: 1282765.1683\n",
      "Epoch 9/20\n",
      "4942/4942 [==============================] - 7s 1ms/step - loss: 1.1945 - mean_absolute_error: 1.1945 - mean_absolute_percentage_error: 1494544.6194\n",
      "Epoch 10/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2350 - mean_absolute_error: 1.2350 - mean_absolute_percentage_error: 1346599.9809\n",
      "Epoch 11/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1944 - mean_absolute_error: 1.1944 - mean_absolute_percentage_error: 1532066.4853\n",
      "Epoch 12/20\n",
      "4942/4942 [==============================] - 7s 1ms/step - loss: 1.2100 - mean_absolute_error: 1.2100 - mean_absolute_percentage_error: 1479336.2589\n",
      "Epoch 13/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2296 - mean_absolute_error: 1.2296 - mean_absolute_percentage_error: 1604460.8440\n",
      "Epoch 14/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.1785 - mean_absolute_error: 1.1785 - mean_absolute_percentage_error: 1481387.2109\n",
      "Epoch 15/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2011 - mean_absolute_error: 1.2011 - mean_absolute_percentage_error: 1405422.9639\n",
      "Epoch 16/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.1939 - mean_absolute_error: 1.1939 - mean_absolute_percentage_error: 1608347.7391\n",
      "Epoch 17/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1962 - mean_absolute_error: 1.1962 - mean_absolute_percentage_error: 1524255.6763\n",
      "Epoch 18/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1811 - mean_absolute_error: 1.1811 - mean_absolute_percentage_error: 1513635.1032\n",
      "Epoch 19/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1921 - mean_absolute_error: 1.1921 - mean_absolute_percentage_error: 1524962.8704\n",
      "Epoch 20/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1832 - mean_absolute_error: 1.1832 - mean_absolute_percentage_error: 1543914.3708\n",
      "549/549 [==============================] - 2s 4ms/step\n",
      "Epoch 1/20\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4942/4942 [==============================] - 9s 2ms/step - loss: 2.2950 - mean_absolute_error: 2.2950 - mean_absolute_percentage_error: 1180833.8528\n",
      "Epoch 2/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2651 - mean_absolute_error: 1.2651 - mean_absolute_percentage_error: 1199761.5137\n",
      "Epoch 3/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2440 - mean_absolute_error: 1.2440 - mean_absolute_percentage_error: 1585958.9568\n",
      "Epoch 4/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2233 - mean_absolute_error: 1.2233 - mean_absolute_percentage_error: 1395958.6660\n",
      "Epoch 5/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2262 - mean_absolute_error: 1.2262 - mean_absolute_percentage_error: 1445176.6910\n",
      "Epoch 6/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2184 - mean_absolute_error: 1.2184 - mean_absolute_percentage_error: 1808534.9972\n",
      "Epoch 7/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1893 - mean_absolute_error: 1.1893 - mean_absolute_percentage_error: 1586505.7839\n",
      "Epoch 8/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2129 - mean_absolute_error: 1.2129 - mean_absolute_percentage_error: 1674456.0901\n",
      "Epoch 9/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2173 - mean_absolute_error: 1.2173 - mean_absolute_percentage_error: 1681672.8229\n",
      "Epoch 10/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2199 - mean_absolute_error: 1.2199 - mean_absolute_percentage_error: 1436990.3023\n",
      "Epoch 11/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1898 - mean_absolute_error: 1.1898 - mean_absolute_percentage_error: 1500084.4355\n",
      "Epoch 12/20\n",
      "4942/4942 [==============================] - 7s 1ms/step - loss: 1.1888 - mean_absolute_error: 1.1888 - mean_absolute_percentage_error: 1712082.2711\n",
      "Epoch 13/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2030 - mean_absolute_error: 1.2030 - mean_absolute_percentage_error: 1731209.4699\n",
      "Epoch 14/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2168 - mean_absolute_error: 1.2168 - mean_absolute_percentage_error: 1378693.4425\n",
      "Epoch 15/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1935 - mean_absolute_error: 1.1935 - mean_absolute_percentage_error: 1424512.5357\n",
      "Epoch 16/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.1669 - mean_absolute_error: 1.1669 - mean_absolute_percentage_error: 1586298.4456\n",
      "Epoch 17/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.1873 - mean_absolute_error: 1.1873 - mean_absolute_percentage_error: 1514794.7062\n",
      "Epoch 18/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1985 - mean_absolute_error: 1.1985 - mean_absolute_percentage_error: 1698172.1576\n",
      "Epoch 19/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2039 - mean_absolute_error: 1.2039 - mean_absolute_percentage_error: 1448914.4372\n",
      "Epoch 20/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.1962 - mean_absolute_error: 1.1962 - mean_absolute_percentage_error: 1650049.5730\n",
      "549/549 [==============================] - 2s 3ms/step\n",
      "Epoch 1/20\n",
      "4942/4942 [==============================] - 10s 2ms/step - loss: 2.2919 - mean_absolute_error: 2.2919 - mean_absolute_percentage_error: 1136593.7492\n",
      "Epoch 2/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2565 - mean_absolute_error: 1.2565 - mean_absolute_percentage_error: 918373.8958\n",
      "Epoch 3/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2572 - mean_absolute_error: 1.2572 - mean_absolute_percentage_error: 1018069.7103\n",
      "Epoch 4/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2344 - mean_absolute_error: 1.2344 - mean_absolute_percentage_error: 1036662.2483\n",
      "Epoch 5/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2270 - mean_absolute_error: 1.2270 - mean_absolute_percentage_error: 957781.6679\n",
      "Epoch 6/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2034 - mean_absolute_error: 1.2034 - mean_absolute_percentage_error: 830109.3669\n",
      "Epoch 7/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2545 - mean_absolute_error: 1.2545 - mean_absolute_percentage_error: 1058296.6553\n",
      "Epoch 8/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2301 - mean_absolute_error: 1.2301 - mean_absolute_percentage_error: 1058655.6630\n",
      "Epoch 9/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2355 - mean_absolute_error: 1.2355 - mean_absolute_percentage_error: 975896.6418\n",
      "Epoch 10/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2775 - mean_absolute_error: 1.2775 - mean_absolute_percentage_error: 1111633.0219\n",
      "Epoch 11/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2024 - mean_absolute_error: 1.2024 - mean_absolute_percentage_error: 1093734.8459\n",
      "Epoch 12/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2095 - mean_absolute_error: 1.2095 - mean_absolute_percentage_error: 1087558.6781\n",
      "Epoch 13/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2175 - mean_absolute_error: 1.2175 - mean_absolute_percentage_error: 944420.7142\n",
      "Epoch 14/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2044 - mean_absolute_error: 1.2044 - mean_absolute_percentage_error: 1002430.5724\n",
      "Epoch 15/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1821 - mean_absolute_error: 1.1821 - mean_absolute_percentage_error: 970620.6491\n",
      "Epoch 16/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2081 - mean_absolute_error: 1.2081 - mean_absolute_percentage_error: 1275292.2942\n",
      "Epoch 17/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1862 - mean_absolute_error: 1.1862 - mean_absolute_percentage_error: 1114563.4548\n",
      "Epoch 18/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2137 - mean_absolute_error: 1.2137 - mean_absolute_percentage_error: 1039429.4117\n",
      "Epoch 19/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1875 - mean_absolute_error: 1.1875 - mean_absolute_percentage_error: 1053048.4373\n",
      "Epoch 20/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.1896 - mean_absolute_error: 1.1896 - mean_absolute_percentage_error: 1148208.7259\n",
      "549/549 [==============================] - 2s 3ms/step\n",
      "Epoch 1/20\n",
      "4942/4942 [==============================] - 10s 2ms/step - loss: 2.1664 - mean_absolute_error: 2.1664 - mean_absolute_percentage_error: 1268134.9161\n",
      "Epoch 2/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2543 - mean_absolute_error: 1.2543 - mean_absolute_percentage_error: 1389710.2019\n",
      "Epoch 3/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2523 - mean_absolute_error: 1.2523 - mean_absolute_percentage_error: 1667850.8040\n",
      "Epoch 4/20\n",
      "4942/4942 [==============================] - 7s 1ms/step - loss: 1.2436 - mean_absolute_error: 1.2436 - mean_absolute_percentage_error: 1662605.5527\n",
      "Epoch 5/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2378 - mean_absolute_error: 1.2378 - mean_absolute_percentage_error: 1654140.7113\n",
      "Epoch 6/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2203 - mean_absolute_error: 1.2203 - mean_absolute_percentage_error: 1522450.3839\n",
      "Epoch 7/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2294 - mean_absolute_error: 1.2294 - mean_absolute_percentage_error: 1697527.5367\n",
      "Epoch 8/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2226 - mean_absolute_error: 1.2226 - mean_absolute_percentage_error: 1448153.1053\n",
      "Epoch 9/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2186 - mean_absolute_error: 1.2186 - mean_absolute_percentage_error: 1496597.7696\n",
      "Epoch 10/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2244 - mean_absolute_error: 1.2244 - mean_absolute_percentage_error: 1542004.0707\n",
      "Epoch 11/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1994 - mean_absolute_error: 1.1994 - mean_absolute_percentage_error: 1520672.4900\n",
      "Epoch 12/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2223 - mean_absolute_error: 1.2223 - mean_absolute_percentage_error: 1432021.1030\n",
      "Epoch 13/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1982 - mean_absolute_error: 1.1982 - mean_absolute_percentage_error: 1643476.1411\n",
      "Epoch 14/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2171 - mean_absolute_error: 1.2171 - mean_absolute_percentage_error: 1523051.0293\n",
      "Epoch 15/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1941 - mean_absolute_error: 1.1941 - mean_absolute_percentage_error: 1502717.8356\n",
      "Epoch 16/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1845 - mean_absolute_error: 1.1845 - mean_absolute_percentage_error: 1731520.2133\n",
      "Epoch 17/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1797 - mean_absolute_error: 1.1797 - mean_absolute_percentage_error: 1634640.0155\n",
      "Epoch 18/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1973 - mean_absolute_error: 1.1973 - mean_absolute_percentage_error: 1594509.5817\n",
      "Epoch 19/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2033 - mean_absolute_error: 1.2033 - mean_absolute_percentage_error: 1756235.6218\n",
      "Epoch 20/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.1899 - mean_absolute_error: 1.1899 - mean_absolute_percentage_error: 1555676.3708\n",
      "549/549 [==============================] - 2s 3ms/step\n",
      "Epoch 1/20\n",
      "4942/4942 [==============================] - 10s 2ms/step - loss: 2.2431 - mean_absolute_error: 2.2431 - mean_absolute_percentage_error: 2490607.2721\n",
      "Epoch 2/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2989 - mean_absolute_error: 1.2989 - mean_absolute_percentage_error: 931353.1418\n",
      "Epoch 3/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2591 - mean_absolute_error: 1.2591 - mean_absolute_percentage_error: 1486822.8116\n",
      "Epoch 4/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2266 - mean_absolute_error: 1.2266 - mean_absolute_percentage_error: 1430696.7603\n",
      "Epoch 5/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1992 - mean_absolute_error: 1.1992 - mean_absolute_percentage_error: 1582205.2524\n",
      "Epoch 6/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2305 - mean_absolute_error: 1.2305 - mean_absolute_percentage_error: 1621122.9825\n",
      "Epoch 7/20\n",
      "4942/4942 [==============================] - 7s 2ms/step - loss: 1.2086 - mean_absolute_error: 1.2086 - mean_absolute_percentage_error: 1479801.7656\n",
      "Epoch 8/20\n",
      "4942/4942 [==============================] - 7s 1ms/step - loss: 1.2124 - mean_absolute_error: 1.2124 - mean_absolute_percentage_error: 1689577.7835\n",
      "Epoch 9/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1943 - mean_absolute_error: 1.1943 - mean_absolute_percentage_error: 1727032.2360\n",
      "Epoch 10/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2037 - mean_absolute_error: 1.2037 - mean_absolute_percentage_error: 1397511.6265\n",
      "Epoch 11/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2827 - mean_absolute_error: 1.2827 - mean_absolute_percentage_error: 1606254.3602\n",
      "Epoch 12/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2303 - mean_absolute_error: 1.2303 - mean_absolute_percentage_error: 1491113.6885\n",
      "Epoch 13/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2193 - mean_absolute_error: 1.2193 - mean_absolute_percentage_error: 1637395.8323\n",
      "Epoch 14/20\n",
      "4942/4942 [==============================] - 7s 1ms/step - loss: 1.1951 - mean_absolute_error: 1.1951 - mean_absolute_percentage_error: 1529086.7508\n",
      "Epoch 15/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1856 - mean_absolute_error: 1.1856 - mean_absolute_percentage_error: 1735363.6063\n",
      "Epoch 16/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.1878 - mean_absolute_error: 1.1878 - mean_absolute_percentage_error: 1629881.7118\n",
      "Epoch 17/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.1979 - mean_absolute_error: 1.1979 - mean_absolute_percentage_error: 1447749.6850\n",
      "Epoch 18/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2093 - mean_absolute_error: 1.2093 - mean_absolute_percentage_error: 1761806.7916\n",
      "Epoch 19/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2173 - mean_absolute_error: 1.2173 - mean_absolute_percentage_error: 1784962.7484\n",
      "Epoch 20/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1855 - mean_absolute_error: 1.1855 - mean_absolute_percentage_error: 1694449.2542\n",
      "549/549 [==============================] - 2s 4ms/step\n",
      "Epoch 1/20\n",
      "4942/4942 [==============================] - 10s 2ms/step - loss: 2.2533 - mean_absolute_error: 2.2533 - mean_absolute_percentage_error: 1901874.3576\n",
      "Epoch 2/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2692 - mean_absolute_error: 1.2692 - mean_absolute_percentage_error: 1172362.2617\n",
      "Epoch 3/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2321 - mean_absolute_error: 1.2321 - mean_absolute_percentage_error: 1591875.3381\n",
      "Epoch 4/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2030 - mean_absolute_error: 1.2030 - mean_absolute_percentage_error: 1546093.1836\n",
      "Epoch 5/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2161 - mean_absolute_error: 1.2161 - mean_absolute_percentage_error: 1517592.0926\n",
      "Epoch 6/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2053 - mean_absolute_error: 1.2053 - mean_absolute_percentage_error: 1801006.6922\n",
      "Epoch 7/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2121 - mean_absolute_error: 1.2121 - mean_absolute_percentage_error: 1270366.4434\n",
      "Epoch 8/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1979 - mean_absolute_error: 1.1979 - mean_absolute_percentage_error: 1434427.3434\n",
      "Epoch 9/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1951 - mean_absolute_error: 1.1951 - mean_absolute_percentage_error: 1515821.1765\n",
      "Epoch 10/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1931 - mean_absolute_error: 1.1931 - mean_absolute_percentage_error: 1557687.5355\n",
      "Epoch 11/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2361 - mean_absolute_error: 1.2361 - mean_absolute_percentage_error: 1846536.9774\n",
      "Epoch 12/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.1806 - mean_absolute_error: 1.1806 - mean_absolute_percentage_error: 1594548.2400\n",
      "Epoch 13/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2121 - mean_absolute_error: 1.2121 - mean_absolute_percentage_error: 1517350.8259\n",
      "Epoch 14/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1830 - mean_absolute_error: 1.1830 - mean_absolute_percentage_error: 1587549.6248\n",
      "Epoch 15/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.2061 - mean_absolute_error: 1.2061 - mean_absolute_percentage_error: 1456534.1041\n",
      "Epoch 16/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1878 - mean_absolute_error: 1.1878 - mean_absolute_percentage_error: 1558929.4604\n",
      "Epoch 17/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2105 - mean_absolute_error: 1.2105 - mean_absolute_percentage_error: 1819830.2651\n",
      "Epoch 18/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.2162 - mean_absolute_error: 1.2162 - mean_absolute_percentage_error: 1574591.7926\n",
      "Epoch 19/20\n",
      "4942/4942 [==============================] - 5s 1ms/step - loss: 1.1804 - mean_absolute_error: 1.1804 - mean_absolute_percentage_error: 1612513.5097\n",
      "Epoch 20/20\n",
      "4942/4942 [==============================] - 6s 1ms/step - loss: 1.1791 - mean_absolute_error: 1.1791 - mean_absolute_percentage_error: 1673861.4401\n",
      "549/549 [==============================] - 2s 3ms/step\n",
      "Epoch 1/20\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5491/5491 [==============================] - 12s 2ms/step - loss: 2.0268 - mean_absolute_error: 2.0268 - mean_absolute_percentage_error: 1694940.3899\n",
      "Epoch 2/20\n",
      "5491/5491 [==============================] - 7s 1ms/step - loss: 1.2636 - mean_absolute_error: 1.2636 - mean_absolute_percentage_error: 1230780.6051\n",
      "Epoch 3/20\n",
      "5491/5491 [==============================] - 6s 1ms/step - loss: 1.2201 - mean_absolute_error: 1.2201 - mean_absolute_percentage_error: 1354985.1789\n",
      "Epoch 4/20\n",
      "5491/5491 [==============================] - 9s 2ms/step - loss: 1.2233 - mean_absolute_error: 1.2233 - mean_absolute_percentage_error: 1405028.6958\n",
      "Epoch 5/20\n",
      "5491/5491 [==============================] - 6s 1ms/step - loss: 1.2640 - mean_absolute_error: 1.2640 - mean_absolute_percentage_error: 1427181.9791\n",
      "Epoch 6/20\n",
      "5491/5491 [==============================] - 6s 1ms/step - loss: 1.2077 - mean_absolute_error: 1.2077 - mean_absolute_percentage_error: 1502775.1066\n",
      "Epoch 7/20\n",
      "5491/5491 [==============================] - 6s 1ms/step - loss: 1.2104 - mean_absolute_error: 1.2104 - mean_absolute_percentage_error: 1469630.4395\n",
      "Epoch 8/20\n",
      "5491/5491 [==============================] - 6s 1ms/step - loss: 1.2017 - mean_absolute_error: 1.2017 - mean_absolute_percentage_error: 1239313.7552\n",
      "Epoch 9/20\n",
      "5491/5491 [==============================] - 9s 2ms/step - loss: 1.1878 - mean_absolute_error: 1.1878 - mean_absolute_percentage_error: 1593877.9280\n",
      "Epoch 10/20\n",
      "5491/5491 [==============================] - 9s 2ms/step - loss: 1.2006 - mean_absolute_error: 1.2006 - mean_absolute_percentage_error: 1265693.5792\n",
      "Epoch 11/20\n",
      "5491/5491 [==============================] - 8s 1ms/step - loss: 1.1944 - mean_absolute_error: 1.1944 - mean_absolute_percentage_error: 1538726.4211\n",
      "Epoch 12/20\n",
      "5491/5491 [==============================] - 7s 1ms/step - loss: 1.1870 - mean_absolute_error: 1.1870 - mean_absolute_percentage_error: 1428222.3631\n",
      "Epoch 13/20\n",
      "5491/5491 [==============================] - 6s 1ms/step - loss: 1.2114 - mean_absolute_error: 1.2114 - mean_absolute_percentage_error: 1555322.6877\n",
      "Epoch 14/20\n",
      "5491/5491 [==============================] - 7s 1ms/step - loss: 1.1889 - mean_absolute_error: 1.1889 - mean_absolute_percentage_error: 1566135.9547\n",
      "Epoch 15/20\n",
      "5491/5491 [==============================] - 7s 1ms/step - loss: 1.1881 - mean_absolute_error: 1.1881 - mean_absolute_percentage_error: 1551936.6098\n",
      "Epoch 16/20\n",
      "5491/5491 [==============================] - 7s 1ms/step - loss: 1.1909 - mean_absolute_error: 1.1909 - mean_absolute_percentage_error: 1325439.9229\n",
      "Epoch 17/20\n",
      "5491/5491 [==============================] - 7s 1ms/step - loss: 1.1721 - mean_absolute_error: 1.1721 - mean_absolute_percentage_error: 1337576.1529\n",
      "Epoch 18/20\n",
      "5491/5491 [==============================] - 6s 1ms/step - loss: 1.2001 - mean_absolute_error: 1.2001 - mean_absolute_percentage_error: 1321996.0577\n",
      "Epoch 19/20\n",
      "5491/5491 [==============================] - 8s 1ms/step - loss: 1.1640 - mean_absolute_error: 1.1640 - mean_absolute_percentage_error: 1313896.4735\n",
      "Epoch 20/20\n",
      "5491/5491 [==============================] - 6s 1ms/step - loss: 1.1876 - mean_absolute_error: 1.1876 - mean_absolute_percentage_error: 1560968.9449\n",
      "{'batch_size': 40, 'epochs': 20, 'optimizer': 'adam', 'units': 150}\n",
      "-1.1194756116182767\n",
      "1/1 [==============================] - 2s 2s/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 11ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 14ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 21ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 19ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 21ms/step\n",
      "1/1 [==============================] - 0s 11ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 12ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 11ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 13ms/step\n",
      "1/1 [==============================] - 0s 11ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 30ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 11ms/step\n",
      "1/1 [==============================] - 0s 12ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 13ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 14ms/step\n",
      "1/1 [==============================] - 0s 12ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 29ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 12ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 14ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 18ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 11ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 12ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 12ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 11ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 12ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 11ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 20ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 13ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 12ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 11ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 19ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 11ms/step\n",
      "1/1 [==============================] - 0s 26ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 26ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 24ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 11ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 12ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 41ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 3ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 41ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 7ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 8ms/step\n",
      "1/1 [==============================] - 0s 6ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n",
      "1/1 [==============================] - 0s 5ms/step\n",
      "1/1 [==============================] - 0s 4ms/step\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1080 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.preprocessing import LabelEncoder, OneHotEncoder\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "from keras.wrappers.scikit_learn import KerasRegressor\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.model_selection import KFold\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from math import sqrt\n",
    "import keras\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras.layers import Dropout\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "data = pd.read_csv('data/hourly/myfile.csv')\n",
    "data = data.drop(['Date and time','Month','Weekday'],axis=1)\n",
    "\n",
    "sc = MinMaxScaler(feature_range = (0,1))  #StandardScaler for standatization and MinMaxScaler forn normalization \n",
    "#sc = StandardScaler()\n",
    "#data = sc.fit_transform(data)\n",
    "\n",
    "X = data.iloc[:, 1:4].values\n",
    "y = data.iloc[:, 0].values\n",
    "# dayEncoder = LabelEncoder()\n",
    "# X[:, 1] = dayEncoder.fit_transform(X[:, 1])\n",
    "# monthEncoder = LabelEncoder()\n",
    "# X[:, 2] = monthEncoder.fit_transform(X[:, 2])\n",
    "\n",
    "# onehotencoder = OneHotEncoder(categorical_features = 'all')\n",
    "\n",
    "# day = X[:,1].reshape(-1,1) \n",
    "# day = onehotencoder.fit_transform(day).toarray()\n",
    "# #remove first dummy variable to avoid trap\n",
    "# day = day[:,1:]\n",
    "\n",
    "# month = X[:,2].reshape(-1,1)\n",
    "# month = onehotencoder.fit_transform(month).toarray()\n",
    "# #remove first dummy variable to avoid trap\n",
    "# month = month[:,1:]\n",
    "\n",
    "# temp=X[:,0].reshape(-1,1)\n",
    "\n",
    "# X = np.concatenate((day, month), axis=1)\n",
    "# X = np.concatenate((X, temp), axis=1)\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2, random_state = 0)\n",
    "\n",
    "X_train = sc.fit_transform(X_train)\n",
    "X_test = sc.transform(X_test)\n",
    "\n",
    "neurons = [150]\n",
    "dimensions = np.shape(X)[1]\n",
    "layers = 6\n",
    "dropout = 0.2\n",
    "def baseline_model(optimizer,units):\n",
    "# Initialising the ANN\n",
    "    model = Sequential()\n",
    "     # Adding the input layer \n",
    "    model.add(Dense(units=units, kernel_initializer='uniform', activation='relu', input_dim=dimensions))\n",
    "    for i in range(0, layers):\n",
    "        #add hidden layers\n",
    "        model.add(Dense(units=units, kernel_initializer='uniform', activation='relu'))\n",
    "        model.add(Dropout(dropout))\n",
    "    #model.add(Dropout(0.2))\n",
    "    # Adding the output layer\n",
    "    model.add(Dense(units=1, kernel_initializer='uniform', activation='relu'))\n",
    "    \n",
    "    # Compiling the ANN\n",
    "    model.compile(optimizer=optimizer, loss='mean_absolute_error', metrics=['mae', 'mape'])\n",
    "    return model\n",
    "\n",
    "# evaluate model with standardized dataset\n",
    "estimator = KerasRegressor(build_fn=baseline_model, verbose=1)\n",
    "\n",
    "#hyperparameters tuning\n",
    "parameters = {#'batch_size': [100,200,400],\n",
    "              'batch_size': [40],\n",
    "              'epochs': [20],\n",
    "              'optimizer': ['adam'],\n",
    "              'units': neurons,\n",
    "              }\n",
    "\n",
    "grid_search = GridSearchCV(estimator = estimator,\n",
    "                           param_grid = parameters,\n",
    "                           scoring = 'neg_mean_absolute_error',\n",
    "                           cv = 10)\n",
    "\n",
    "\n",
    "grid_search = grid_search.fit(X_train, y_train)\n",
    "best_parameters = grid_search.best_params_\n",
    "best_accuracy = grid_search.best_score_\n",
    "print(best_parameters)\n",
    "print(best_accuracy)\n",
    "\n",
    "\n",
    "real_energy_usage = []\n",
    "temp = []\n",
    "error = []\n",
    "predicted = []\n",
    "\n",
    "plt.rcParams['figure.figsize'] = (20, 15)\n",
    "for i,day in enumerate(y_test):\n",
    "    prediction = grid_search.predict((np.array([X_test[i]])))\n",
    "    predicted.append(np.asscalar(prediction))\n",
    "    # test.append(y_test[i])\n",
    "    \n",
    "    # temp.append(X_test[i][0])\n",
    "    error.append(abs(y_test[i] - prediction))\n",
    "\n",
    "\n",
    "\n",
    "real_energy_usage = y_test[:]\n",
    "#real_energy_usage = test\n",
    "#real_energy_usage = np.expand_dims(np.asarray(real_energy_usage),axis=1)       \n",
    "#real_energy_usage_tbl = np.zeros(shape=(len(real_energy_usage), 1 + 2 ))\n",
    "#real_energy_usage_tbl[:,0] = real_energy_usage[:,0]\n",
    "#real_energy_usage  = sc.inverse_transform(real_energy_usage_tbl)[:,0]\n",
    "\n",
    "temp = X_test[:,0]\n",
    "temp = np.expand_dims(np.asarray(temp),axis=1)       \n",
    "temp_tbl = np.zeros(shape=(len(temp), 1 + 2 ))\n",
    "temp_tbl[:,0] = temp[:,0]\n",
    "temp = sc.inverse_transform(temp_tbl)[:,0]\n",
    "\n",
    "\n",
    "rmse = sqrt(mean_squared_error(real_energy_usage ,predicted))\n",
    "avg_error =  sum(error) / float(len(error))\n",
    "\n",
    "plt.title('Daily Electricity Usage vs Temp (multi-varaiate linear regression)')\n",
    "plt.text(0,0,\"Best Score\" + str(round(best_accuracy,4)) + \" | \" \n",
    "         + \"Rmse:\" + str(round(rmse, 2)) +  \" | \"\n",
    "         + \"avg_error:\" + str(round(avg_error, 2)) , fontsize=10)\n",
    "plt.text(0,0.5,\"units: \" + str(neurons[0]) + \" | \" +  \"layers: \" + str(layers) +\n",
    "         \" | \" + \"input_dim\" + str(dimensions) + \" | \"\n",
    "         + \"dropout: \" + str(dropout) + \" | \" + str(best_parameters),fontsize=9)\n",
    "plt.scatter(temp,predicted)\n",
    "plt.scatter(temp,real_energy_usage)\n",
    "plt.xlabel('Temperature (C)')\n",
    "plt.ylabel('kWh')\n",
    "plt.legend(['predicted', 'real'], loc='upper right')\n",
    "\n",
    "\n",
    "#grid_search.best_estimator_.model.save('dnn_100.h5')\n",
    "plt.show()\n",
    "\n"
   ]
  },
  {
   "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": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}

Comments

Leave a Comment

Post a Comment

About This Project

Project period

09/19/2019 - 09/30/2019

Views

37

Team Members

Courses

Get trained in Learny Technologies and develop your own applications.

New batch starts from Nov 13th.