X Tutup
{ "cells": [ { "cell_type": "code", "execution_count": 3, "metadata": {}, "outputs": [], "source": [ "import pandas as pd\n", "import numpy as np\n", "import matplotlib.pyplot as plt\n", "%matplotlib inline\n", "import warnings \n", "warnings.filterwarnings('ignore')" ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [], "source": [ "data = pd.read_csv('ClimateChange_MonthlyStart.csv')" ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
Unnamed: 0SourceDateMean
00GCAG2016-120.7895
11GISTEMP2016-120.8100
22GCAG2016-110.7504
33GISTEMP2016-110.9300
44GCAG2016-100.7292
\n", "
" ], "text/plain": [ " Unnamed: 0 Source Date Mean\n", "0 0 GCAG 2016-12 0.7895\n", "1 1 GISTEMP 2016-12 0.8100\n", "2 2 GCAG 2016-11 0.7504\n", "3 3 GISTEMP 2016-11 0.9300\n", "4 4 GCAG 2016-10 0.7292" ] }, "execution_count": 5, "metadata": {}, "output_type": "execute_result" } ], "source": [ "data.head()" ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [], "source": [ "data.drop(columns='Unnamed: 0', inplace=True)" ] }, { "cell_type": "code", "execution_count": 7, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
SourceDateMean
0GCAG2016-120.7895
1GISTEMP2016-120.8100
2GCAG2016-110.7504
3GISTEMP2016-110.9300
4GCAG2016-100.7292
\n", "
" ], "text/plain": [ " Source Date Mean\n", "0 GCAG 2016-12 0.7895\n", "1 GISTEMP 2016-12 0.8100\n", "2 GCAG 2016-11 0.7504\n", "3 GISTEMP 2016-11 0.9300\n", "4 GCAG 2016-10 0.7292" ] }, "execution_count": 7, "metadata": {}, "output_type": "execute_result" } ], "source": [ "data.head()" ] }, { "cell_type": "code", "execution_count": 8, "metadata": {}, "outputs": [], "source": [ "data = data[data['Source']=='GCAG']" ] }, { "cell_type": "code", "execution_count": 9, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
SourceDateMean
0GCAG2016-120.7895
2GCAG2016-110.7504
4GCAG2016-100.7292
6GCAG2016-090.8767
8GCAG2016-080.8998
\n", "
" ], "text/plain": [ " Source Date Mean\n", "0 GCAG 2016-12 0.7895\n", "2 GCAG 2016-11 0.7504\n", "4 GCAG 2016-10 0.7292\n", "6 GCAG 2016-09 0.8767\n", "8 GCAG 2016-08 0.8998" ] }, "execution_count": 9, "metadata": {}, "output_type": "execute_result" } ], "source": [ "data.head()" ] }, { "cell_type": "code", "execution_count": 10, "metadata": {}, "outputs": [], "source": [ "data['Date'] = pd.to_datetime(data['Date'])" ] }, { "cell_type": "code", "execution_count": 11, "metadata": {}, "outputs": [], "source": [ "data.set_index('Date',drop=True,inplace=True)" ] }, { "cell_type": "code", "execution_count": 12, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "DatetimeIndex(['2016-12-01', '2016-11-01', '2016-10-01', '2016-09-01',\n", " '2016-08-01', '2016-07-01', '2016-06-01', '2016-05-01',\n", " '2016-04-01', '2016-03-01',\n", " ...\n", " '1880-10-01', '1880-09-01', '1880-08-01', '1880-07-01',\n", " '1880-06-01', '1880-05-01', '1880-04-01', '1880-03-01',\n", " '1880-02-01', '1880-01-01'],\n", " dtype='datetime64[ns]', name='Date', length=1644, freq=None)" ] }, "execution_count": 12, "metadata": {}, "output_type": "execute_result" } ], "source": [ "data.index" ] }, { "cell_type": "code", "execution_count": 13, "metadata": {}, "outputs": [], "source": [ "data = data.sort_index()" ] }, { "cell_type": "code", "execution_count": 14, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
SourceMean
Date
1880-01-01GCAG0.0009
1880-02-01GCAG-0.1229
1880-03-01GCAG-0.1357
1880-04-01GCAG-0.0499
1880-05-01GCAG-0.0738
\n", "
" ], "text/plain": [ " Source Mean\n", "Date \n", "1880-01-01 GCAG 0.0009\n", "1880-02-01 GCAG -0.1229\n", "1880-03-01 GCAG -0.1357\n", "1880-04-01 GCAG -0.0499\n", "1880-05-01 GCAG -0.0738" ] }, "execution_count": 14, "metadata": {}, "output_type": "execute_result" } ], "source": [ "data.head()" ] }, { "cell_type": "code", "execution_count": 15, "metadata": {}, "outputs": [], "source": [ "data.index.freq='MS'" ] }, { "cell_type": "code", "execution_count": 16, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "DatetimeIndex(['1880-01-01', '1880-02-01', '1880-03-01', '1880-04-01',\n", " '1880-05-01', '1880-06-01', '1880-07-01', '1880-08-01',\n", " '1880-09-01', '1880-10-01',\n", " ...\n", " '2016-03-01', '2016-04-01', '2016-05-01', '2016-06-01',\n", " '2016-07-01', '2016-08-01', '2016-09-01', '2016-10-01',\n", " '2016-11-01', '2016-12-01'],\n", " dtype='datetime64[ns]', name='Date', length=1644, freq='MS')" ] }, "execution_count": 16, "metadata": {}, "output_type": "execute_result" } ], "source": [ "data.index" ] }, { "cell_type": "code", "execution_count": 17, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "" ] }, "execution_count": 17, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "data.plot(figsize=(20,6))" ] }, { "cell_type": "code", "execution_count": 18, "metadata": {}, "outputs": [], "source": [ "from statsmodels.graphics.tsaplots import plot_acf, plot_pacf" ] }, { "cell_type": "code", "execution_count": 19, "metadata": { "scrolled": true }, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plot_acf(data['Mean'],lags=5);" ] }, { "cell_type": "code", "execution_count": 20, "metadata": {}, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plot_pacf(data['Mean'],lags=5);" ] }, { "cell_type": "code", "execution_count": 21, "metadata": {}, "outputs": [], "source": [ "from statsmodels.tsa.stattools import adfuller\n", "result = adfuller(data['Mean'],autolag='AIC')" ] }, { "cell_type": "code", "execution_count": 22, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(-0.4995472167184225,\n", " 0.8921048239299925,\n", " 24,\n", " 1619,\n", " {'1%': -3.434395520959224,\n", " '5%': -2.8633268625511046,\n", " '10%': -2.5677212878453477},\n", " -3042.5048877999643)" ] }, "execution_count": 22, "metadata": {}, "output_type": "execute_result" } ], "source": [ "result" ] }, { "cell_type": "code", "execution_count": 23, "metadata": {}, "outputs": [], "source": [ "result = adfuller(data['Mean'].diff(1).dropna(),autolag='AIC')" ] }, { "cell_type": "code", "execution_count": 24, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(-11.832255570392535,\n", " 7.940293414567011e-22,\n", " 23,\n", " 1619,\n", " {'1%': -3.434395520959224,\n", " '5%': -2.8633268625511046,\n", " '10%': -2.5677212878453477},\n", " -3041.860129875631)" ] }, "execution_count": 24, "metadata": {}, "output_type": "execute_result" } ], "source": [ "result" ] }, { "cell_type": "code", "execution_count": 25, "metadata": {}, "outputs": [], "source": [ "data.drop(columns='Source',inplace=True)\n", "train = data.iloc[:-12]\n", "test = data.iloc[-12:]" ] }, { "cell_type": "code", "execution_count": 27, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "/anaconda3/lib/python3.6/site-packages/pmdarima/arima/_auto_solvers.py:207: ModelFitWarning: Unable to fit ARIMA for order=(2, 1, 3); data is likely non-stationary. (if you do not want to see these warnings, run with error_action=\"ignore\")\n", " ModelFitWarning)\n", "/anaconda3/lib/python3.6/site-packages/pmdarima/arima/_auto_solvers.py:207: ModelFitWarning: Unable to fit ARIMA for order=(2, 1, 2); data is likely non-stationary. (if you do not want to see these warnings, run with error_action=\"ignore\")\n", " ModelFitWarning)\n", "/anaconda3/lib/python3.6/site-packages/pmdarima/arima/_auto_solvers.py:207: ModelFitWarning: Unable to fit ARIMA for order=(1, 1, 4); data is likely non-stationary. (if you do not want to see these warnings, run with error_action=\"ignore\")\n", " ModelFitWarning)\n" ] }, { "data": { "text/html": [ "\n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "
ARIMA Model Results
Dep. Variable: D.y No. Observations: 1643
Model: ARIMA(1, 1, 3) Log Likelihood 1560.759
Method: css-mle S.D. of innovations 0.094
Date: Wed, 15 May 2019 AIC -3109.519
Time: 10:58:45 BIC -3077.093
Sample: 1 HQIC -3097.494
\n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "
coef std err z P>|z| [0.025 0.975]
const 0.0006 0.000 2.058 0.040 2.69e-05 0.001
ar.L1.D.y 0.8897 0.026 34.371 0.000 0.839 0.940
ma.L1.D.y -1.3690 0.036 -37.750 0.000 -1.440 -1.298
ma.L2.D.y 0.3358 0.041 8.124 0.000 0.255 0.417
ma.L3.D.y 0.0460 0.028 1.639 0.101 -0.009 0.101
\n", "\n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "
Roots
Real Imaginary Modulus Frequency
AR.1 1.1239 +0.0000j 1.1239 0.0000
MA.1 1.0234 +0.0000j 1.0234 0.0000
MA.2 2.0482 +0.0000j 2.0482 0.0000
MA.3 -10.3722 +0.0000j 10.3722 0.5000
" ], "text/plain": [ "\n", "\"\"\"\n", " ARIMA Model Results \n", "==============================================================================\n", "Dep. Variable: D.y No. Observations: 1643\n", "Model: ARIMA(1, 1, 3) Log Likelihood 1560.759\n", "Method: css-mle S.D. of innovations 0.094\n", "Date: Wed, 15 May 2019 AIC -3109.519\n", "Time: 10:58:45 BIC -3077.093\n", "Sample: 1 HQIC -3097.494\n", " \n", "==============================================================================\n", " coef std err z P>|z| [0.025 0.975]\n", "------------------------------------------------------------------------------\n", "const 0.0006 0.000 2.058 0.040 2.69e-05 0.001\n", "ar.L1.D.y 0.8897 0.026 34.371 0.000 0.839 0.940\n", "ma.L1.D.y -1.3690 0.036 -37.750 0.000 -1.440 -1.298\n", "ma.L2.D.y 0.3358 0.041 8.124 0.000 0.255 0.417\n", "ma.L3.D.y 0.0460 0.028 1.639 0.101 -0.009 0.101\n", " Roots \n", "=============================================================================\n", " Real Imaginary Modulus Frequency\n", "-----------------------------------------------------------------------------\n", "AR.1 1.1239 +0.0000j 1.1239 0.0000\n", "MA.1 1.0234 +0.0000j 1.0234 0.0000\n", "MA.2 2.0482 +0.0000j 2.0482 0.0000\n", "MA.3 -10.3722 +0.0000j 10.3722 0.5000\n", "-----------------------------------------------------------------------------\n", "\"\"\"" ] }, "execution_count": 27, "metadata": {}, "output_type": "execute_result" } ], "source": [ "from pmdarima import auto_arima\n", "stepwise_fit = auto_arima(data['Mean'],start_p=0,start_q=0,max_p=5,max_q=5,seasonal=False)\n", "stepwise_fit.summary()\n", "\n", "# ARIMA results = (1,1,3)" ] }, { "cell_type": "code", "execution_count": 28, "metadata": {}, "outputs": [ { "data": { "text/html": [ "\n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "
ARIMA Model Results
Dep. Variable: D.Mean No. Observations: 1631
Model: ARIMA(1, 1, 3) Log Likelihood 1550.423
Method: css-mle S.D. of innovations 0.093
Date: Wed, 15 May 2019 AIC -3088.845
Time: 10:59:01 BIC -3056.464
Sample: 02-01-1880 HQIC -3076.832
- 12-01-2015
\n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "
coef std err z P>|z| [0.025 0.975]
const 0.0006 0.000 2.043 0.041 2.34e-05 0.001
ar.L1.D.Mean 0.8912 0.026 33.868 0.000 0.840 0.943
ma.L1.D.Mean -1.3750 0.037 -37.494 0.000 -1.447 -1.303
ma.L2.D.Mean 0.3408 0.042 8.189 0.000 0.259 0.422
ma.L3.D.Mean 0.0471 0.028 1.665 0.096 -0.008 0.102
\n", "\n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "
Roots
Real Imaginary Modulus Frequency
AR.1 1.1221 +0.0000j 1.1221 0.0000
MA.1 1.0238 +0.0000j 1.0238 0.0000
MA.2 2.0182 +0.0000j 2.0182 0.0000
MA.3 -10.2846 +0.0000j 10.2846 0.5000
" ], "text/plain": [ "\n", "\"\"\"\n", " ARIMA Model Results \n", "==============================================================================\n", "Dep. Variable: D.Mean No. Observations: 1631\n", "Model: ARIMA(1, 1, 3) Log Likelihood 1550.423\n", "Method: css-mle S.D. of innovations 0.093\n", "Date: Wed, 15 May 2019 AIC -3088.845\n", "Time: 10:59:01 BIC -3056.464\n", "Sample: 02-01-1880 HQIC -3076.832\n", " - 12-01-2015 \n", "================================================================================\n", " coef std err z P>|z| [0.025 0.975]\n", "--------------------------------------------------------------------------------\n", "const 0.0006 0.000 2.043 0.041 2.34e-05 0.001\n", "ar.L1.D.Mean 0.8912 0.026 33.868 0.000 0.840 0.943\n", "ma.L1.D.Mean -1.3750 0.037 -37.494 0.000 -1.447 -1.303\n", "ma.L2.D.Mean 0.3408 0.042 8.189 0.000 0.259 0.422\n", "ma.L3.D.Mean 0.0471 0.028 1.665 0.096 -0.008 0.102\n", " Roots \n", "=============================================================================\n", " Real Imaginary Modulus Frequency\n", "-----------------------------------------------------------------------------\n", "AR.1 1.1221 +0.0000j 1.1221 0.0000\n", "MA.1 1.0238 +0.0000j 1.0238 0.0000\n", "MA.2 2.0182 +0.0000j 2.0182 0.0000\n", "MA.3 -10.2846 +0.0000j 10.2846 0.5000\n", "-----------------------------------------------------------------------------\n", "\"\"\"" ] }, "execution_count": 28, "metadata": {}, "output_type": "execute_result" } ], "source": [ "from statsmodels.tsa.arima_model import ARIMA,ARIMAResults\n", "\n", "model = ARIMA(train['Mean'],order=(1,1,3))\n", "results = model.fit()\n", "results.summary()" ] }, { "cell_type": "code", "execution_count": 29, "metadata": {}, "outputs": [], "source": [ "start = len(train)\n", "end = len(train)+len(test)-1\n", "\n", "predictions = results.predict(start=start, end=end, typ='levels').rename('ARIMA Predictions')" ] }, { "cell_type": "code", "execution_count": 30, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "" ] }, "execution_count": 30, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "test.plot(figsize=(15,3),legend=True)\n", "predictions.plot(legend=True)" ] }, { "cell_type": "code", "execution_count": 31, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Mean 0.936292\n", "dtype: float64" ] }, "execution_count": 31, "metadata": {}, "output_type": "execute_result" } ], "source": [ "test.mean()" ] }, { "cell_type": "code", "execution_count": 32, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.9147031839305546" ] }, "execution_count": 32, "metadata": {}, "output_type": "execute_result" } ], "source": [ "predictions.mean()" ] }, { "cell_type": "code", "execution_count": 33, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "0.123226566211986\n", "0.3832349589356232\n" ] } ], "source": [ "from sklearn.metrics import mean_squared_error, r2_score\n", "\n", "MSE = mean_squared_error(test, predictions)\n", "r2 = r2_score(test, predictions)\n", "\n", "print(np.sqrt(MSE))\n", "print(r2)" ] }, { "cell_type": "code", "execution_count": 34, "metadata": { "scrolled": true }, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "/anaconda3/lib/python3.6/site-packages/statsmodels/base/model.py:508: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals\n", " \"Check mle_retvals\", ConvergenceWarning)\n", "/anaconda3/lib/python3.6/site-packages/statsmodels/base/model.py:508: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals\n", " \"Check mle_retvals\", ConvergenceWarning)\n", "/anaconda3/lib/python3.6/site-packages/statsmodels/base/model.py:508: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals\n", " \"Check mle_retvals\", ConvergenceWarning)\n", "/anaconda3/lib/python3.6/site-packages/statsmodels/base/model.py:508: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals\n", " \"Check mle_retvals\", ConvergenceWarning)\n", "/anaconda3/lib/python3.6/site-packages/statsmodels/base/model.py:508: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals\n", " \"Check mle_retvals\", ConvergenceWarning)\n", "/anaconda3/lib/python3.6/site-packages/statsmodels/base/model.py:508: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals\n", " \"Check mle_retvals\", ConvergenceWarning)\n", "/anaconda3/lib/python3.6/site-packages/statsmodels/base/model.py:508: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals\n", " \"Check mle_retvals\", ConvergenceWarning)\n", "/anaconda3/lib/python3.6/site-packages/statsmodels/base/model.py:508: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals\n", " \"Check mle_retvals\", ConvergenceWarning)\n", "/anaconda3/lib/python3.6/site-packages/statsmodels/base/model.py:508: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals\n", " \"Check mle_retvals\", ConvergenceWarning)\n", "/anaconda3/lib/python3.6/site-packages/statsmodels/base/model.py:508: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals\n", " \"Check mle_retvals\", ConvergenceWarning)\n", "/anaconda3/lib/python3.6/site-packages/pmdarima/arima/_auto_solvers.py:207: ModelFitWarning: Unable to fit ARIMA for order=(2, 1, 2) seasonal_order=(1, 0, 1, 12); data is likely non-stationary. (if you do not want to see these warnings, run with error_action=\"ignore\")\n", " ModelFitWarning)\n", "/anaconda3/lib/python3.6/site-packages/pmdarima/arima/_auto_solvers.py:207: ModelFitWarning: Unable to fit ARIMA for order=(2, 1, 4) seasonal_order=(1, 0, 1, 12); data is likely non-stationary. (if you do not want to see these warnings, run with error_action=\"ignore\")\n", " ModelFitWarning)\n", "/anaconda3/lib/python3.6/site-packages/statsmodels/base/model.py:508: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals\n", " \"Check mle_retvals\", ConvergenceWarning)\n", "/anaconda3/lib/python3.6/site-packages/pmdarima/arima/_auto_solvers.py:207: ModelFitWarning: Unable to fit ARIMA for order=(3, 1, 4) seasonal_order=(1, 0, 1, 12); data is likely non-stationary. (if you do not want to see these warnings, run with error_action=\"ignore\")\n", " ModelFitWarning)\n", "/anaconda3/lib/python3.6/site-packages/statsmodels/base/model.py:508: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals\n", " \"Check mle_retvals\", ConvergenceWarning)\n", "/anaconda3/lib/python3.6/site-packages/statsmodels/base/model.py:508: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals\n", " \"Check mle_retvals\", ConvergenceWarning)\n", "/anaconda3/lib/python3.6/site-packages/statsmodels/base/model.py:508: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals\n", " \"Check mle_retvals\", ConvergenceWarning)\n", "/anaconda3/lib/python3.6/site-packages/statsmodels/base/model.py:508: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals\n", " \"Check mle_retvals\", ConvergenceWarning)\n", "/anaconda3/lib/python3.6/site-packages/statsmodels/base/model.py:508: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals\n", " \"Check mle_retvals\", ConvergenceWarning)\n", "/anaconda3/lib/python3.6/site-packages/pmdarima/arima/_auto_solvers.py:207: ModelFitWarning: Unable to fit ARIMA for order=(2, 1, 2) seasonal_order=(2, 0, 1, 12); data is likely non-stationary. (if you do not want to see these warnings, run with error_action=\"ignore\")\n", " ModelFitWarning)\n", "/anaconda3/lib/python3.6/site-packages/pmdarima/arima/_auto_solvers.py:207: ModelFitWarning: Unable to fit ARIMA for order=(2, 1, 4) seasonal_order=(2, 0, 1, 12); data is likely non-stationary. (if you do not want to see these warnings, run with error_action=\"ignore\")\n", " ModelFitWarning)\n", "/anaconda3/lib/python3.6/site-packages/statsmodels/base/model.py:508: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals\n", " \"Check mle_retvals\", ConvergenceWarning)\n", "/anaconda3/lib/python3.6/site-packages/pmdarima/arima/_auto_solvers.py:207: ModelFitWarning: Unable to fit ARIMA for order=(3, 1, 4) seasonal_order=(2, 0, 1, 12); data is likely non-stationary. (if you do not want to see these warnings, run with error_action=\"ignore\")\n", " ModelFitWarning)\n" ] }, { "data": { "text/html": [ "\n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "
Statespace Model Results
Dep. Variable: y No. Observations: 1644
Model: SARIMAX(2, 1, 3)x(2, 0, 1, 12) Log Likelihood 1563.053
Date: Wed, 15 May 2019 AIC -3106.106
Time: 11:03:39 BIC -3052.063
Sample: 0 HQIC -3086.065
- 1644
Covariance Type: opg
\n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "
coef std err z P>|z| [0.025 0.975]
intercept 3.765e-05 3.83e-05 0.983 0.325 -3.74e-05 0.000
ar.L1 0.0013 0.086 0.015 0.988 -0.168 0.170
ar.L2 0.8372 0.076 10.956 0.000 0.687 0.987
ma.L1 -0.4619 0.087 -5.280 0.000 -0.633 -0.290
ma.L2 -0.8931 0.101 -8.873 0.000 -1.090 -0.696
ma.L3 0.3795 0.037 10.383 0.000 0.308 0.451
ar.S.L12 0.4557 0.184 2.477 0.013 0.095 0.816
ar.S.L24 0.0698 0.027 2.613 0.009 0.017 0.122
ma.S.L12 -0.4209 0.183 -2.303 0.021 -0.779 -0.063
sigma2 0.0087 0.000 36.263 0.000 0.008 0.009
\n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "
Ljung-Box (Q): 38.45 Jarque-Bera (JB): 128.35
Prob(Q): 0.54 Prob(JB): 0.00
Heteroskedasticity (H): 0.91 Skew: -0.06
Prob(H) (two-sided): 0.27 Kurtosis: 4.36


Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step)." ], "text/plain": [ "\n", "\"\"\"\n", " Statespace Model Results \n", "==========================================================================================\n", "Dep. Variable: y No. Observations: 1644\n", "Model: SARIMAX(2, 1, 3)x(2, 0, 1, 12) Log Likelihood 1563.053\n", "Date: Wed, 15 May 2019 AIC -3106.106\n", "Time: 11:03:39 BIC -3052.063\n", "Sample: 0 HQIC -3086.065\n", " - 1644 \n", "Covariance Type: opg \n", "==============================================================================\n", " coef std err z P>|z| [0.025 0.975]\n", "------------------------------------------------------------------------------\n", "intercept 3.765e-05 3.83e-05 0.983 0.325 -3.74e-05 0.000\n", "ar.L1 0.0013 0.086 0.015 0.988 -0.168 0.170\n", "ar.L2 0.8372 0.076 10.956 0.000 0.687 0.987\n", "ma.L1 -0.4619 0.087 -5.280 0.000 -0.633 -0.290\n", "ma.L2 -0.8931 0.101 -8.873 0.000 -1.090 -0.696\n", "ma.L3 0.3795 0.037 10.383 0.000 0.308 0.451\n", "ar.S.L12 0.4557 0.184 2.477 0.013 0.095 0.816\n", "ar.S.L24 0.0698 0.027 2.613 0.009 0.017 0.122\n", "ma.S.L12 -0.4209 0.183 -2.303 0.021 -0.779 -0.063\n", "sigma2 0.0087 0.000 36.263 0.000 0.008 0.009\n", "===================================================================================\n", "Ljung-Box (Q): 38.45 Jarque-Bera (JB): 128.35\n", "Prob(Q): 0.54 Prob(JB): 0.00\n", "Heteroskedasticity (H): 0.91 Skew: -0.06\n", "Prob(H) (two-sided): 0.27 Kurtosis: 4.36\n", "===================================================================================\n", "\n", "Warnings:\n", "[1] Covariance matrix calculated using the outer product of gradients (complex-step).\n", "\"\"\"" ] }, "execution_count": 34, "metadata": {}, "output_type": "execute_result" } ], "source": [ "from pmdarima import auto_arima\n", "stepwise_fit = auto_arima(data['Mean'],start_p=0,start_q=0,max_p=5,max_q=5,seasonal=True,m=12)\n", "stepwise_fit.summary()\n", "\n", "# SARIMA results = (1,1,2)(0,0,2,12)" ] }, { "cell_type": "code", "execution_count": 35, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "/anaconda3/lib/python3.6/site-packages/statsmodels/base/model.py:508: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals\n", " \"Check mle_retvals\", ConvergenceWarning)\n" ] } ], "source": [ "from statsmodels.tsa.statespace.sarimax import SARIMAX\n", "model = SARIMAX(train['Mean'],order=(2,1,3),seasonal_order=(2,0,1,12))\n", "results = model.fit()" ] }, { "cell_type": "code", "execution_count": 36, "metadata": {}, "outputs": [], "source": [ "start = len(train)\n", "end = len(train)+len(test)-1\n", "\n", "predictions = results.predict(start=start, end=end, typ='levels').rename('SARIMA Predictions')" ] }, { "cell_type": "code", "execution_count": 37, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "" ] }, "execution_count": 37, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "test.plot(figsize=(15,3),legend=True)\n", "predictions.plot(legend=True)" ] }, { "cell_type": "code", "execution_count": 38, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "0.12461538449037077\n", "0.36925418226757367\n" ] } ], "source": [ "from sklearn.metrics import mean_squared_error, r2_score\n", "\n", "MSE = mean_squared_error(test, predictions)\n", "r2 = r2_score(test, predictions)\n", "\n", "print(np.sqrt(MSE))\n", "print(r2)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## LSTM" ] }, { "cell_type": "code", "execution_count": 55, "metadata": {}, "outputs": [], "source": [ "from sklearn.preprocessing import StandardScaler\n", "\n", "sc = StandardScaler()\n", "sc.fit(train)\n", "\n", "train_sc = sc.transform(train)\n", "test_sc = sc.transform(test)\n", "\n", "X_train = train_sc[:-1]\n", "y_train = train_sc[1:]\n", "\n", "X_test = test_sc[:-1]\n", "y_test = test_sc[1:]" ] }, { "cell_type": "code", "execution_count": 56, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(1631, 1)" ] }, "execution_count": 56, "metadata": {}, "output_type": "execute_result" } ], "source": [ "X_train.shape" ] }, { "cell_type": "code", "execution_count": 58, "metadata": {}, "outputs": [], "source": [ "X_train_t = X_train[:,None]\n", "X_test_t = X_test[:,None]" ] }, { "cell_type": "code", "execution_count": 59, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(1631, 1, 1)" ] }, "execution_count": 59, "metadata": {}, "output_type": "execute_result" } ], "source": [ "X_train_t.shape" ] }, { "cell_type": "code", "execution_count": 60, "metadata": { "scrolled": true }, "outputs": [], "source": [ "from keras.models import Sequential\n", "from keras.layers import Dense, LSTM\n", "from keras.optimizers import Adam\n", "import keras.backend as K" ] }, { "cell_type": "code", "execution_count": 64, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Train on 1631 samples, validate on 11 samples\n", "Epoch 1/200\n", "1631/1631 [==============================] - 1s 534us/step - loss: 1.0062 - val_loss: 8.1564\n", "Epoch 2/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.9062 - val_loss: 7.1272\n", "Epoch 3/200\n", "1631/1631 [==============================] - 0s 88us/step - loss: 0.7829 - val_loss: 5.9224\n", "Epoch 4/200\n", "1631/1631 [==============================] - 0s 79us/step - loss: 0.6333 - val_loss: 4.3501\n", "Epoch 5/200\n", "1631/1631 [==============================] - 0s 78us/step - loss: 0.4675 - val_loss: 2.6891\n", "Epoch 6/200\n", "1631/1631 [==============================] - 0s 81us/step - loss: 0.3121 - val_loss: 1.2906\n", "Epoch 7/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1960 - val_loss: 0.4579\n", "Epoch 8/200\n", "1631/1631 [==============================] - 0s 81us/step - loss: 0.1347 - val_loss: 0.1578\n", "Epoch 9/200\n", "1631/1631 [==============================] - 0s 81us/step - loss: 0.1124 - val_loss: 0.1573\n", "Epoch 10/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1069 - val_loss: 0.2065\n", "Epoch 11/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1058 - val_loss: 0.2421\n", "Epoch 12/200\n", "1631/1631 [==============================] - 0s 81us/step - loss: 0.1054 - val_loss: 0.2518\n", "Epoch 13/200\n", "1631/1631 [==============================] - 0s 82us/step - loss: 0.1052 - val_loss: 0.2509\n", "Epoch 14/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1050 - val_loss: 0.2729\n", "Epoch 15/200\n", "1631/1631 [==============================] - 0s 84us/step - loss: 0.1048 - val_loss: 0.2660\n", "Epoch 16/200\n", "1631/1631 [==============================] - 0s 86us/step - loss: 0.1049 - val_loss: 0.2783\n", "Epoch 17/200\n", "1631/1631 [==============================] - 0s 83us/step - loss: 0.1047 - val_loss: 0.2721\n", "Epoch 18/200\n", "1631/1631 [==============================] - 0s 82us/step - loss: 0.1048 - val_loss: 0.2676\n", "Epoch 19/200\n", "1631/1631 [==============================] - 0s 94us/step - loss: 0.1047 - val_loss: 0.2828\n", "Epoch 20/200\n", "1631/1631 [==============================] - 0s 94us/step - loss: 0.1046 - val_loss: 0.2793\n", "Epoch 21/200\n", "1631/1631 [==============================] - 0s 98us/step - loss: 0.1046 - val_loss: 0.2677\n", "Epoch 22/200\n", "1631/1631 [==============================] - 0s 93us/step - loss: 0.1046 - val_loss: 0.2784\n", "Epoch 23/200\n", "1631/1631 [==============================] - 0s 93us/step - loss: 0.1046 - val_loss: 0.2565\n", "Epoch 24/200\n", "1631/1631 [==============================] - 0s 86us/step - loss: 0.1045 - val_loss: 0.2739\n", "Epoch 25/200\n", "1631/1631 [==============================] - 0s 92us/step - loss: 0.1046 - val_loss: 0.2662\n", "Epoch 26/200\n", "1631/1631 [==============================] - 0s 97us/step - loss: 0.1045 - val_loss: 0.2601\n", "Epoch 27/200\n", "1631/1631 [==============================] - 0s 92us/step - loss: 0.1044 - val_loss: 0.2574\n", "Epoch 28/200\n", "1631/1631 [==============================] - 0s 84us/step - loss: 0.1044 - val_loss: 0.2587\n", "Epoch 29/200\n", "1631/1631 [==============================] - 0s 84us/step - loss: 0.1044 - val_loss: 0.2587\n", "Epoch 30/200\n", "1631/1631 [==============================] - 0s 85us/step - loss: 0.1044 - val_loss: 0.2400\n", "Epoch 31/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1044 - val_loss: 0.2585\n", "Epoch 32/200\n", "1631/1631 [==============================] - 0s 74us/step - loss: 0.1044 - val_loss: 0.2451\n", "Epoch 33/200\n", "1631/1631 [==============================] - 0s 71us/step - loss: 0.1043 - val_loss: 0.2448\n", "Epoch 34/200\n", "1631/1631 [==============================] - 0s 72us/step - loss: 0.1043 - val_loss: 0.2385\n", "Epoch 35/200\n", "1631/1631 [==============================] - 0s 74us/step - loss: 0.1042 - val_loss: 0.2384\n", "Epoch 36/200\n", "1631/1631 [==============================] - 0s 79us/step - loss: 0.1043 - val_loss: 0.2449\n", "Epoch 37/200\n", "1631/1631 [==============================] - 0s 86us/step - loss: 0.1042 - val_loss: 0.2374\n", "Epoch 38/200\n", "1631/1631 [==============================] - 0s 82us/step - loss: 0.1043 - val_loss: 0.2297\n", "Epoch 39/200\n", "1631/1631 [==============================] - 0s 75us/step - loss: 0.1043 - val_loss: 0.2257\n", "Epoch 40/200\n", "1631/1631 [==============================] - 0s 79us/step - loss: 0.1042 - val_loss: 0.2217\n", "Epoch 41/200\n", "1631/1631 [==============================] - 0s 83us/step - loss: 0.1041 - val_loss: 0.2231\n", "Epoch 42/200\n", "1631/1631 [==============================] - 0s 90us/step - loss: 0.1042 - val_loss: 0.2193\n", "Epoch 43/200\n", "1631/1631 [==============================] - 0s 89us/step - loss: 0.1041 - val_loss: 0.2249\n", "Epoch 44/200\n", "1631/1631 [==============================] - 0s 82us/step - loss: 0.1043 - val_loss: 0.2101\n", "Epoch 45/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1042 - val_loss: 0.1890\n", "Epoch 46/200\n", "1631/1631 [==============================] - 0s 75us/step - loss: 0.1040 - val_loss: 0.2319\n", "Epoch 47/200\n", "1631/1631 [==============================] - 0s 77us/step - loss: 0.1041 - val_loss: 0.2037\n", "Epoch 48/200\n", "1631/1631 [==============================] - 0s 78us/step - loss: 0.1041 - val_loss: 0.1915\n", "Epoch 49/200\n", "1631/1631 [==============================] - 0s 82us/step - loss: 0.1040 - val_loss: 0.2012\n", "Epoch 50/200\n", "1631/1631 [==============================] - 0s 84us/step - loss: 0.1040 - val_loss: 0.1959\n", "Epoch 51/200\n", "1631/1631 [==============================] - 0s 88us/step - loss: 0.1041 - val_loss: 0.2025\n", "Epoch 52/200\n", "1631/1631 [==============================] - 0s 87us/step - loss: 0.1040 - val_loss: 0.1945\n", "Epoch 53/200\n", "1631/1631 [==============================] - 0s 84us/step - loss: 0.1039 - val_loss: 0.2070\n", "Epoch 54/200\n", "1631/1631 [==============================] - 0s 88us/step - loss: 0.1039 - val_loss: 0.1840\n", "Epoch 55/200\n", "1631/1631 [==============================] - 0s 79us/step - loss: 0.1039 - val_loss: 0.2001\n", "Epoch 56/200\n", "1631/1631 [==============================] - 0s 76us/step - loss: 0.1039 - val_loss: 0.1788\n", "Epoch 57/200\n", "1631/1631 [==============================] - 0s 76us/step - loss: 0.1040 - val_loss: 0.1812\n", "Epoch 58/200\n", "1631/1631 [==============================] - 0s 78us/step - loss: 0.1039 - val_loss: 0.1858\n", "Epoch 59/200\n", "1631/1631 [==============================] - 0s 77us/step - loss: 0.1039 - val_loss: 0.1887\n", "Epoch 60/200\n", "1631/1631 [==============================] - 0s 79us/step - loss: 0.1038 - val_loss: 0.1961\n", "Epoch 61/200\n", "1631/1631 [==============================] - 0s 86us/step - loss: 0.1037 - val_loss: 0.1795\n", "Epoch 62/200\n", "1631/1631 [==============================] - 0s 86us/step - loss: 0.1039 - val_loss: 0.1805\n", "Epoch 63/200\n", "1631/1631 [==============================] - 0s 81us/step - loss: 0.1037 - val_loss: 0.1723\n", "Epoch 64/200\n", "1631/1631 [==============================] - 0s 79us/step - loss: 0.1038 - val_loss: 0.1742\n", "Epoch 65/200\n", "1631/1631 [==============================] - 0s 77us/step - loss: 0.1042 - val_loss: 0.1605\n", "Epoch 66/200\n", "1631/1631 [==============================] - 0s 76us/step - loss: 0.1040 - val_loss: 0.1978\n", "Epoch 67/200\n", "1631/1631 [==============================] - 0s 72us/step - loss: 0.1038 - val_loss: 0.1688\n", "Epoch 68/200\n", "1631/1631 [==============================] - 0s 71us/step - loss: 0.1038 - val_loss: 0.1752\n", "Epoch 69/200\n", "1631/1631 [==============================] - 0s 72us/step - loss: 0.1037 - val_loss: 0.1625\n", "Epoch 70/200\n", "1631/1631 [==============================] - 0s 72us/step - loss: 0.1037 - val_loss: 0.1607\n", "Epoch 71/200\n", "1631/1631 [==============================] - 0s 72us/step - loss: 0.1039 - val_loss: 0.1529\n", "Epoch 72/200\n", "1631/1631 [==============================] - 0s 72us/step - loss: 0.1036 - val_loss: 0.1737\n", "Epoch 73/200\n", "1631/1631 [==============================] - 0s 70us/step - loss: 0.1037 - val_loss: 0.1681\n", "Epoch 74/200\n", "1631/1631 [==============================] - 0s 68us/step - loss: 0.1037 - val_loss: 0.1620\n", "Epoch 75/200\n", "1631/1631 [==============================] - 0s 67us/step - loss: 0.1036 - val_loss: 0.1568\n", "Epoch 76/200\n", "1631/1631 [==============================] - 0s 70us/step - loss: 0.1038 - val_loss: 0.1498\n", "Epoch 77/200\n", "1631/1631 [==============================] - 0s 68us/step - loss: 0.1036 - val_loss: 0.1603\n", "Epoch 78/200\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "1631/1631 [==============================] - 0s 72us/step - loss: 0.1037 - val_loss: 0.1635\n", "Epoch 79/200\n", "1631/1631 [==============================] - 0s 74us/step - loss: 0.1036 - val_loss: 0.1519\n", "Epoch 80/200\n", "1631/1631 [==============================] - 0s 79us/step - loss: 0.1036 - val_loss: 0.1502\n", "Epoch 81/200\n", "1631/1631 [==============================] - 0s 76us/step - loss: 0.1036 - val_loss: 0.1487\n", "Epoch 82/200\n", "1631/1631 [==============================] - 0s 74us/step - loss: 0.1038 - val_loss: 0.1385\n", "Epoch 83/200\n", "1631/1631 [==============================] - 0s 76us/step - loss: 0.1036 - val_loss: 0.1485\n", "Epoch 84/200\n", "1631/1631 [==============================] - 0s 77us/step - loss: 0.1035 - val_loss: 0.1632\n", "Epoch 85/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1036 - val_loss: 0.1525\n", "Epoch 86/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1036 - val_loss: 0.1505\n", "Epoch 87/200\n", "1631/1631 [==============================] - 0s 81us/step - loss: 0.1035 - val_loss: 0.1518\n", "Epoch 88/200\n", "1631/1631 [==============================] - 0s 74us/step - loss: 0.1036 - val_loss: 0.1424\n", "Epoch 89/200\n", "1631/1631 [==============================] - 0s 77us/step - loss: 0.1037 - val_loss: 0.1441\n", "Epoch 90/200\n", "1631/1631 [==============================] - 0s 74us/step - loss: 0.1036 - val_loss: 0.1414\n", "Epoch 91/200\n", "1631/1631 [==============================] - 0s 71us/step - loss: 0.1035 - val_loss: 0.1605\n", "Epoch 92/200\n", "1631/1631 [==============================] - 0s 72us/step - loss: 0.1037 - val_loss: 0.1280\n", "Epoch 93/200\n", "1631/1631 [==============================] - 0s 76us/step - loss: 0.1036 - val_loss: 0.1544\n", "Epoch 94/200\n", "1631/1631 [==============================] - 0s 77us/step - loss: 0.1037 - val_loss: 0.1435\n", "Epoch 95/200\n", "1631/1631 [==============================] - 0s 78us/step - loss: 0.1036 - val_loss: 0.1442\n", "Epoch 96/200\n", "1631/1631 [==============================] - 0s 81us/step - loss: 0.1036 - val_loss: 0.1550\n", "Epoch 97/200\n", "1631/1631 [==============================] - 0s 78us/step - loss: 0.1037 - val_loss: 0.1416\n", "Epoch 98/200\n", "1631/1631 [==============================] - 0s 79us/step - loss: 0.1034 - val_loss: 0.1488\n", "Epoch 99/200\n", "1631/1631 [==============================] - 0s 79us/step - loss: 0.1036 - val_loss: 0.1507\n", "Epoch 100/200\n", "1631/1631 [==============================] - 0s 79us/step - loss: 0.1036 - val_loss: 0.1351\n", "Epoch 101/200\n", "1631/1631 [==============================] - 0s 79us/step - loss: 0.1037 - val_loss: 0.1324\n", "Epoch 102/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1036 - val_loss: 0.1290\n", "Epoch 103/200\n", "1631/1631 [==============================] - 0s 79us/step - loss: 0.1036 - val_loss: 0.1329\n", "Epoch 104/200\n", "1631/1631 [==============================] - 0s 78us/step - loss: 0.1034 - val_loss: 0.1366\n", "Epoch 105/200\n", "1631/1631 [==============================] - 0s 74us/step - loss: 0.1034 - val_loss: 0.1376\n", "Epoch 106/200\n", "1631/1631 [==============================] - 0s 76us/step - loss: 0.1034 - val_loss: 0.1313\n", "Epoch 107/200\n", "1631/1631 [==============================] - 0s 73us/step - loss: 0.1035 - val_loss: 0.1272\n", "Epoch 108/200\n", "1631/1631 [==============================] - 0s 77us/step - loss: 0.1035 - val_loss: 0.1263\n", "Epoch 109/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1037 - val_loss: 0.1321\n", "Epoch 110/200\n", "1631/1631 [==============================] - 0s 84us/step - loss: 0.1036 - val_loss: 0.1252\n", "Epoch 111/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1034 - val_loss: 0.1345\n", "Epoch 112/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1034 - val_loss: 0.1239\n", "Epoch 113/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1036 - val_loss: 0.1337\n", "Epoch 114/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1034 - val_loss: 0.1284\n", "Epoch 115/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1038 - val_loss: 0.1255\n", "Epoch 116/200\n", "1631/1631 [==============================] - 0s 81us/step - loss: 0.1034 - val_loss: 0.1233\n", "Epoch 117/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1035 - val_loss: 0.1322\n", "Epoch 118/200\n", "1631/1631 [==============================] - 0s 74us/step - loss: 0.1034 - val_loss: 0.1265\n", "Epoch 119/200\n", "1631/1631 [==============================] - 0s 72us/step - loss: 0.1034 - val_loss: 0.1317\n", "Epoch 120/200\n", "1631/1631 [==============================] - 0s 72us/step - loss: 0.1034 - val_loss: 0.1269\n", "Epoch 121/200\n", "1631/1631 [==============================] - 0s 74us/step - loss: 0.1035 - val_loss: 0.1189\n", "Epoch 122/200\n", "1631/1631 [==============================] - 0s 78us/step - loss: 0.1036 - val_loss: 0.1203\n", "Epoch 123/200\n", "1631/1631 [==============================] - 0s 79us/step - loss: 0.1036 - val_loss: 0.1287\n", "Epoch 124/200\n", "1631/1631 [==============================] - 0s 74us/step - loss: 0.1035 - val_loss: 0.1285\n", "Epoch 125/200\n", "1631/1631 [==============================] - 0s 73us/step - loss: 0.1033 - val_loss: 0.1214\n", "Epoch 126/200\n", "1631/1631 [==============================] - 0s 71us/step - loss: 0.1033 - val_loss: 0.1223\n", "Epoch 127/200\n", "1631/1631 [==============================] - 0s 72us/step - loss: 0.1034 - val_loss: 0.1255\n", "Epoch 128/200\n", "1631/1631 [==============================] - 0s 78us/step - loss: 0.1035 - val_loss: 0.1244\n", "Epoch 129/200\n", "1631/1631 [==============================] - 0s 79us/step - loss: 0.1034 - val_loss: 0.1224\n", "Epoch 130/200\n", "1631/1631 [==============================] - 0s 81us/step - loss: 0.1033 - val_loss: 0.1168\n", "Epoch 131/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1033 - val_loss: 0.1161\n", "Epoch 132/200\n", "1631/1631 [==============================] - 0s 76us/step - loss: 0.1033 - val_loss: 0.1192\n", "Epoch 133/200\n", "1631/1631 [==============================] - 0s 72us/step - loss: 0.1034 - val_loss: 0.1147\n", "Epoch 134/200\n", "1631/1631 [==============================] - 0s 72us/step - loss: 0.1035 - val_loss: 0.1134\n", "Epoch 135/200\n", "1631/1631 [==============================] - 0s 71us/step - loss: 0.1033 - val_loss: 0.1132\n", "Epoch 136/200\n", "1631/1631 [==============================] - 0s 71us/step - loss: 0.1032 - val_loss: 0.1142\n", "Epoch 137/200\n", "1631/1631 [==============================] - 0s 73us/step - loss: 0.1034 - val_loss: 0.1146\n", "Epoch 138/200\n", "1631/1631 [==============================] - 0s 78us/step - loss: 0.1034 - val_loss: 0.1137\n", "Epoch 139/200\n", "1631/1631 [==============================] - 0s 79us/step - loss: 0.1034 - val_loss: 0.1172\n", "Epoch 140/200\n", "1631/1631 [==============================] - 0s 82us/step - loss: 0.1034 - val_loss: 0.1137\n", "Epoch 141/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1033 - val_loss: 0.1100\n", "Epoch 142/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1034 - val_loss: 0.1184\n", "Epoch 143/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1033 - val_loss: 0.1142\n", "Epoch 144/200\n", "1631/1631 [==============================] - 0s 82us/step - loss: 0.1033 - val_loss: 0.1073\n", "Epoch 145/200\n", "1631/1631 [==============================] - 0s 79us/step - loss: 0.1033 - val_loss: 0.1152\n", "Epoch 146/200\n", "1631/1631 [==============================] - 0s 74us/step - loss: 0.1033 - val_loss: 0.1097\n", "Epoch 147/200\n", "1631/1631 [==============================] - 0s 75us/step - loss: 0.1035 - val_loss: 0.1075\n", "Epoch 148/200\n", "1631/1631 [==============================] - 0s 77us/step - loss: 0.1033 - val_loss: 0.1112\n", "Epoch 149/200\n", "1631/1631 [==============================] - 0s 79us/step - loss: 0.1034 - val_loss: 0.1075\n", "Epoch 150/200\n", "1631/1631 [==============================] - 0s 82us/step - loss: 0.1033 - val_loss: 0.1098\n", "Epoch 151/200\n", "1631/1631 [==============================] - 0s 77us/step - loss: 0.1032 - val_loss: 0.1064\n", "Epoch 152/200\n", "1631/1631 [==============================] - 0s 72us/step - loss: 0.1033 - val_loss: 0.1072\n", "Epoch 153/200\n", "1631/1631 [==============================] - 0s 71us/step - loss: 0.1031 - val_loss: 0.1112\n", "Epoch 154/200\n", "1631/1631 [==============================] - 0s 73us/step - loss: 0.1032 - val_loss: 0.1133\n", "Epoch 155/200\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "1631/1631 [==============================] - 0s 79us/step - loss: 0.1034 - val_loss: 0.1126\n", "Epoch 156/200\n", "1631/1631 [==============================] - 0s 75us/step - loss: 0.1034 - val_loss: 0.1120\n", "Epoch 157/200\n", "1631/1631 [==============================] - 0s 73us/step - loss: 0.1033 - val_loss: 0.1117\n", "Epoch 158/200\n", "1631/1631 [==============================] - 0s 75us/step - loss: 0.1033 - val_loss: 0.1044\n", "Epoch 159/200\n", "1631/1631 [==============================] - 0s 75us/step - loss: 0.1032 - val_loss: 0.1084\n", "Epoch 160/200\n", "1631/1631 [==============================] - 0s 78us/step - loss: 0.1033 - val_loss: 0.1127\n", "Epoch 161/200\n", "1631/1631 [==============================] - 0s 78us/step - loss: 0.1034 - val_loss: 0.1140\n", "Epoch 162/200\n", "1631/1631 [==============================] - 0s 81us/step - loss: 0.1031 - val_loss: 0.1045\n", "Epoch 163/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1033 - val_loss: 0.1114\n", "Epoch 164/200\n", "1631/1631 [==============================] - 0s 82us/step - loss: 0.1032 - val_loss: 0.1052\n", "Epoch 165/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1033 - val_loss: 0.1056\n", "Epoch 166/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1035 - val_loss: 0.1074\n", "Epoch 167/200\n", "1631/1631 [==============================] - 0s 75us/step - loss: 0.1032 - val_loss: 0.1062\n", "Epoch 168/200\n", "1631/1631 [==============================] - 0s 75us/step - loss: 0.1032 - val_loss: 0.1034\n", "Epoch 169/200\n", "1631/1631 [==============================] - 0s 77us/step - loss: 0.1031 - val_loss: 0.1039\n", "Epoch 170/200\n", "1631/1631 [==============================] - 0s 81us/step - loss: 0.1032 - val_loss: 0.1057\n", "Epoch 171/200\n", "1631/1631 [==============================] - 0s 81us/step - loss: 0.1031 - val_loss: 0.1029\n", "Epoch 172/200\n", "1631/1631 [==============================] - 0s 81us/step - loss: 0.1032 - val_loss: 0.1041\n", "Epoch 173/200\n", "1631/1631 [==============================] - 0s 76us/step - loss: 0.1031 - val_loss: 0.1028\n", "Epoch 174/200\n", "1631/1631 [==============================] - 0s 75us/step - loss: 0.1031 - val_loss: 0.1034\n", "Epoch 175/200\n", "1631/1631 [==============================] - 0s 72us/step - loss: 0.1032 - val_loss: 0.1021\n", "Epoch 176/200\n", "1631/1631 [==============================] - 0s 74us/step - loss: 0.1032 - val_loss: 0.1071\n", "Epoch 177/200\n", "1631/1631 [==============================] - 0s 73us/step - loss: 0.1031 - val_loss: 0.1040\n", "Epoch 178/200\n", "1631/1631 [==============================] - 0s 75us/step - loss: 0.1032 - val_loss: 0.1052\n", "Epoch 179/200\n", "1631/1631 [==============================] - 0s 73us/step - loss: 0.1031 - val_loss: 0.1016\n", "Epoch 180/200\n", "1631/1631 [==============================] - 0s 71us/step - loss: 0.1033 - val_loss: 0.1039\n", "Epoch 181/200\n", "1631/1631 [==============================] - 0s 71us/step - loss: 0.1031 - val_loss: 0.1034\n", "Epoch 182/200\n", "1631/1631 [==============================] - 0s 72us/step - loss: 0.1032 - val_loss: 0.1019\n", "Epoch 183/200\n", "1631/1631 [==============================] - 0s 73us/step - loss: 0.1031 - val_loss: 0.1017\n", "Epoch 184/200\n", "1631/1631 [==============================] - 0s 76us/step - loss: 0.1031 - val_loss: 0.1012\n", "Epoch 185/200\n", "1631/1631 [==============================] - 0s 77us/step - loss: 0.1031 - val_loss: 0.1008\n", "Epoch 186/200\n", "1631/1631 [==============================] - 0s 77us/step - loss: 0.1031 - val_loss: 0.1021\n", "Epoch 187/200\n", "1631/1631 [==============================] - 0s 74us/step - loss: 0.1031 - val_loss: 0.1009\n", "Epoch 188/200\n", "1631/1631 [==============================] - 0s 72us/step - loss: 0.1031 - val_loss: 0.1011\n", "Epoch 189/200\n", "1631/1631 [==============================] - 0s 72us/step - loss: 0.1031 - val_loss: 0.1006\n", "Epoch 190/200\n", "1631/1631 [==============================] - 0s 76us/step - loss: 0.1035 - val_loss: 0.1007\n", "Epoch 191/200\n", "1631/1631 [==============================] - 0s 79us/step - loss: 0.1031 - val_loss: 0.1000\n", "Epoch 192/200\n", "1631/1631 [==============================] - 0s 82us/step - loss: 0.1031 - val_loss: 0.1009\n", "Epoch 193/200\n", "1631/1631 [==============================] - 0s 81us/step - loss: 0.1032 - val_loss: 0.1011\n", "Epoch 194/200\n", "1631/1631 [==============================] - 0s 75us/step - loss: 0.1032 - val_loss: 0.1022\n", "Epoch 195/200\n", "1631/1631 [==============================] - 0s 70us/step - loss: 0.1031 - val_loss: 0.1004\n", "Epoch 196/200\n", "1631/1631 [==============================] - 0s 70us/step - loss: 0.1032 - val_loss: 0.1007\n", "Epoch 197/200\n", "1631/1631 [==============================] - 0s 74us/step - loss: 0.1032 - val_loss: 0.1008\n", "Epoch 198/200\n", "1631/1631 [==============================] - 0s 80us/step - loss: 0.1032 - val_loss: 0.0998\n", "Epoch 199/200\n", "1631/1631 [==============================] - 0s 78us/step - loss: 0.1032 - val_loss: 0.1010\n", "Epoch 200/200\n", "1631/1631 [==============================] - 0s 81us/step - loss: 0.1032 - val_loss: 0.1005\n" ] }, { "data": { "text/plain": [ "" ] }, "execution_count": 64, "metadata": {}, "output_type": "execute_result" } ], "source": [ "np.random.seed(7)\n", "K.clear_session()\n", "model = Sequential()\n", "model.add(LSTM(7, input_shape=(1,1),activation='relu'))\n", "model.add(Dense(1))\n", "model.compile(loss='mean_squared_error',optimizer=Adam(lr=0.001))\n", "model.fit(X_train_t,y_train,validation_data=(X_test_t,y_test),epochs=200)" ] }, { "cell_type": "code", "execution_count": 65, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "[]" ] }, "execution_count": 65, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plt.plot(model.history.history['val_loss'][10:])" ] }, { "cell_type": "code", "execution_count": 66, "metadata": {}, "outputs": [], "source": [ "predictions = model.predict(X_test_t)\n", "predictions_sc = sc.inverse_transform(predictions)" ] }, { "cell_type": "code", "execution_count": 74, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "[]" ] }, "execution_count": 74, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plt.plot(test[1:].values)\n", "plt.plot(predictions_sc)" ] }, { "cell_type": "code", "execution_count": 75, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "0.10081410180544051\n", "0.600108817740227\n" ] } ], "source": [ "MSE = mean_squared_error(test[1:], predictions_sc)\n", "r2 = r2_score(test[1:], predictions_sc)\n", "\n", "print(np.sqrt(MSE))\n", "print(r2)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Feedforward NN" ] }, { "cell_type": "code", "execution_count": 76, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Train on 1631 samples, validate on 11 samples\n", "Epoch 1/200\n", "1631/1631 [==============================] - 0s 206us/step - loss: 0.3306 - val_loss: 0.1660\n", "Epoch 2/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.2353 - val_loss: 0.0946\n", "Epoch 3/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1728 - val_loss: 0.0904\n", "Epoch 4/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1367 - val_loss: 0.0910\n", "Epoch 5/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1194 - val_loss: 0.0914\n", "Epoch 6/200\n", "1631/1631 [==============================] - 0s 28us/step - loss: 0.1127 - val_loss: 0.0901\n", "Epoch 7/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1100 - val_loss: 0.0896\n", "Epoch 8/200\n", "1631/1631 [==============================] - 0s 28us/step - loss: 0.1087 - val_loss: 0.0897\n", "Epoch 9/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1077 - val_loss: 0.0895\n", "Epoch 10/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1072 - val_loss: 0.0890\n", "Epoch 11/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1064 - val_loss: 0.0928\n", "Epoch 12/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1061 - val_loss: 0.0901\n", "Epoch 13/200\n", "1631/1631 [==============================] - 0s 32us/step - loss: 0.1058 - val_loss: 0.0912\n", "Epoch 14/200\n", "1631/1631 [==============================] - 0s 42us/step - loss: 0.1054 - val_loss: 0.0897\n", "Epoch 15/200\n", "1631/1631 [==============================] - 0s 27us/step - loss: 0.1051 - val_loss: 0.0927\n", "Epoch 16/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1049 - val_loss: 0.0921\n", "Epoch 17/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1044 - val_loss: 0.0901\n", "Epoch 18/200\n", "1631/1631 [==============================] - 0s 33us/step - loss: 0.1043 - val_loss: 0.0903\n", "Epoch 19/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1041 - val_loss: 0.0937\n", "Epoch 20/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1039 - val_loss: 0.0892\n", "Epoch 21/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1037 - val_loss: 0.0920\n", "Epoch 22/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1038 - val_loss: 0.0903\n", "Epoch 23/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1036 - val_loss: 0.0895\n", "Epoch 24/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1034 - val_loss: 0.0898\n", "Epoch 25/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1033 - val_loss: 0.0897\n", "Epoch 26/200\n", "1631/1631 [==============================] - 0s 38us/step - loss: 0.1032 - val_loss: 0.0905\n", "Epoch 27/200\n", "1631/1631 [==============================] - 0s 36us/step - loss: 0.1034 - val_loss: 0.0899\n", "Epoch 28/200\n", "1631/1631 [==============================] - 0s 38us/step - loss: 0.1032 - val_loss: 0.0900\n", "Epoch 29/200\n", "1631/1631 [==============================] - 0s 35us/step - loss: 0.1031 - val_loss: 0.0915\n", "Epoch 30/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1031 - val_loss: 0.0899\n", "Epoch 31/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1031 - val_loss: 0.0902\n", "Epoch 32/200\n", "1631/1631 [==============================] - 0s 27us/step - loss: 0.1032 - val_loss: 0.0898\n", "Epoch 33/200\n", "1631/1631 [==============================] - 0s 27us/step - loss: 0.1030 - val_loss: 0.0898\n", "Epoch 34/200\n", "1631/1631 [==============================] - 0s 40us/step - loss: 0.1030 - val_loss: 0.0911\n", "Epoch 35/200\n", "1631/1631 [==============================] - 0s 27us/step - loss: 0.1032 - val_loss: 0.0903\n", "Epoch 36/200\n", "1631/1631 [==============================] - 0s 32us/step - loss: 0.1031 - val_loss: 0.0917\n", "Epoch 37/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0901\n", "Epoch 38/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0905\n", "Epoch 39/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1032 - val_loss: 0.0905\n", "Epoch 40/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1031 - val_loss: 0.0918\n", "Epoch 41/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1029 - val_loss: 0.0902\n", "Epoch 42/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1031 - val_loss: 0.0918\n", "Epoch 43/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1031 - val_loss: 0.0905\n", "Epoch 44/200\n", "1631/1631 [==============================] - 0s 35us/step - loss: 0.1030 - val_loss: 0.0922\n", "Epoch 45/200\n", "1631/1631 [==============================] - 0s 36us/step - loss: 0.1030 - val_loss: 0.0911\n", "Epoch 46/200\n", "1631/1631 [==============================] - 0s 38us/step - loss: 0.1031 - val_loss: 0.0904\n", "Epoch 47/200\n", "1631/1631 [==============================] - 0s 38us/step - loss: 0.1033 - val_loss: 0.0907\n", "Epoch 48/200\n", "1631/1631 [==============================] - 0s 35us/step - loss: 0.1030 - val_loss: 0.0906\n", "Epoch 49/200\n", "1631/1631 [==============================] - 0s 34us/step - loss: 0.1030 - val_loss: 0.0904\n", "Epoch 50/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1031 - val_loss: 0.0902\n", "Epoch 51/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1029 - val_loss: 0.0903\n", "Epoch 52/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1032 - val_loss: 0.0911\n", "Epoch 53/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0902\n", "Epoch 54/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1029 - val_loss: 0.0910\n", "Epoch 55/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1029 - val_loss: 0.0903\n", "Epoch 56/200\n", "1631/1631 [==============================] - 0s 33us/step - loss: 0.1029 - val_loss: 0.0904\n", "Epoch 57/200\n", "1631/1631 [==============================] - 0s 33us/step - loss: 0.1030 - val_loss: 0.0903\n", "Epoch 58/200\n", "1631/1631 [==============================] - 0s 35us/step - loss: 0.1032 - val_loss: 0.0904\n", "Epoch 59/200\n", "1631/1631 [==============================] - 0s 37us/step - loss: 0.1030 - val_loss: 0.0904\n", "Epoch 60/200\n", "1631/1631 [==============================] - 0s 35us/step - loss: 0.1032 - val_loss: 0.0902\n", "Epoch 61/200\n", "1631/1631 [==============================] - 0s 36us/step - loss: 0.1030 - val_loss: 0.0902\n", "Epoch 62/200\n", "1631/1631 [==============================] - 0s 38us/step - loss: 0.1029 - val_loss: 0.0903\n", "Epoch 63/200\n", "1631/1631 [==============================] - 0s 36us/step - loss: 0.1030 - val_loss: 0.0902\n", "Epoch 64/200\n", "1631/1631 [==============================] - 0s 36us/step - loss: 0.1029 - val_loss: 0.0913\n", "Epoch 65/200\n", "1631/1631 [==============================] - 0s 36us/step - loss: 0.1029 - val_loss: 0.0903\n", "Epoch 66/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0912\n", "Epoch 67/200\n", "1631/1631 [==============================] - 0s 34us/step - loss: 0.1029 - val_loss: 0.0912\n", "Epoch 68/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1030 - val_loss: 0.0908\n", "Epoch 69/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1029 - val_loss: 0.0915\n", "Epoch 70/200\n", "1631/1631 [==============================] - 0s 27us/step - loss: 0.1030 - val_loss: 0.0932\n", "Epoch 71/200\n", "1631/1631 [==============================] - 0s 28us/step - loss: 0.1033 - val_loss: 0.0902\n", "Epoch 72/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1031 - val_loss: 0.0904\n", "Epoch 73/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1033 - val_loss: 0.0912\n", "Epoch 74/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1032 - val_loss: 0.0905\n", "Epoch 75/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1030 - val_loss: 0.0908\n", "Epoch 76/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1029 - val_loss: 0.0917\n", "Epoch 77/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1030 - val_loss: 0.0913\n", "Epoch 78/200\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0908\n", "Epoch 79/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1031 - val_loss: 0.0912\n", "Epoch 80/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1031 - val_loss: 0.0904\n", "Epoch 81/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1029 - val_loss: 0.0905\n", "Epoch 82/200\n", "1631/1631 [==============================] - 0s 28us/step - loss: 0.1029 - val_loss: 0.0914\n", "Epoch 83/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1031 - val_loss: 0.0945\n", "Epoch 84/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1029 - val_loss: 0.0904\n", "Epoch 85/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1029 - val_loss: 0.0904\n", "Epoch 86/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1031 - val_loss: 0.0902\n", "Epoch 87/200\n", "1631/1631 [==============================] - 0s 28us/step - loss: 0.1032 - val_loss: 0.0923\n", "Epoch 88/200\n", "1631/1631 [==============================] - 0s 28us/step - loss: 0.1030 - val_loss: 0.0921\n", "Epoch 89/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1030 - val_loss: 0.0907\n", "Epoch 90/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1031 - val_loss: 0.0933\n", "Epoch 91/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1030 - val_loss: 0.0907\n", "Epoch 92/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1031 - val_loss: 0.0914\n", "Epoch 93/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1029 - val_loss: 0.0903\n", "Epoch 94/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1028 - val_loss: 0.0905\n", "Epoch 95/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0903\n", "Epoch 96/200\n", "1631/1631 [==============================] - 0s 128us/step - loss: 0.1031 - val_loss: 0.0918\n", "Epoch 97/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1033 - val_loss: 0.0904\n", "Epoch 98/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1031 - val_loss: 0.0904\n", "Epoch 99/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1030 - val_loss: 0.0903\n", "Epoch 100/200\n", " 32/1631 [..............................] - ETA: 0s - loss: 0.0751" ] }, { "name": "stderr", "output_type": "stream", "text": [ "/anaconda3/lib/python3.6/site-packages/keras/callbacks.py:122: UserWarning: Method on_batch_end() is slow compared to the batch update (0.158251). Check your callbacks.\n", " % delta_t_median)\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "1631/1631 [==============================] - 0s 30us/step - loss: 0.1029 - val_loss: 0.0903\n", "Epoch 101/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0912\n", "Epoch 102/200\n", "1631/1631 [==============================] - 0s 36us/step - loss: 0.1029 - val_loss: 0.0908\n", "Epoch 103/200\n", "1631/1631 [==============================] - 0s 39us/step - loss: 0.1029 - val_loss: 0.0906\n", "Epoch 104/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0934\n", "Epoch 105/200\n", "1631/1631 [==============================] - 0s 27us/step - loss: 0.1030 - val_loss: 0.0905\n", "Epoch 106/200\n", "1631/1631 [==============================] - 0s 28us/step - loss: 0.1031 - val_loss: 0.0904\n", "Epoch 107/200\n", "1631/1631 [==============================] - 0s 27us/step - loss: 0.1029 - val_loss: 0.0904\n", "Epoch 108/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1028 - val_loss: 0.0928\n", "Epoch 109/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1031 - val_loss: 0.0919\n", "Epoch 110/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1032 - val_loss: 0.0904\n", "Epoch 111/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1033 - val_loss: 0.0906\n", "Epoch 112/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1029 - val_loss: 0.0906\n", "Epoch 113/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1030 - val_loss: 0.0905\n", "Epoch 114/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1029 - val_loss: 0.0904\n", "Epoch 115/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1029 - val_loss: 0.0906\n", "Epoch 116/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1029 - val_loss: 0.0928\n", "Epoch 117/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1032 - val_loss: 0.0932\n", "Epoch 118/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1031 - val_loss: 0.0923\n", "Epoch 119/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1028 - val_loss: 0.0904\n", "Epoch 120/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1029 - val_loss: 0.0908\n", "Epoch 121/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1029 - val_loss: 0.0923\n", "Epoch 122/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1031 - val_loss: 0.0927\n", "Epoch 123/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0926\n", "Epoch 124/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1031 - val_loss: 0.0907\n", "Epoch 125/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0944\n", "Epoch 126/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1029 - val_loss: 0.0902\n", "Epoch 127/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1029 - val_loss: 0.0923\n", "Epoch 128/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1030 - val_loss: 0.0904\n", "Epoch 129/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1031 - val_loss: 0.0904\n", "Epoch 130/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1030 - val_loss: 0.0905\n", "Epoch 131/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1032 - val_loss: 0.0913\n", "Epoch 132/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0903\n", "Epoch 133/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1030 - val_loss: 0.0904\n", "Epoch 134/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1029 - val_loss: 0.0907\n", "Epoch 135/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1029 - val_loss: 0.0905\n", "Epoch 136/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1029 - val_loss: 0.0905\n", "Epoch 137/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0936\n", "Epoch 138/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1029 - val_loss: 0.0909\n", "Epoch 139/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1031 - val_loss: 0.0928\n", "Epoch 140/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1031 - val_loss: 0.0933\n", "Epoch 141/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1031 - val_loss: 0.0904\n", "Epoch 142/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0934\n", "Epoch 143/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1029 - val_loss: 0.0903\n", "Epoch 144/200\n", "1631/1631 [==============================] - 0s 28us/step - loss: 0.1029 - val_loss: 0.0906\n", "Epoch 145/200\n", "1631/1631 [==============================] - 0s 28us/step - loss: 0.1029 - val_loss: 0.0936\n", "Epoch 146/200\n", "1631/1631 [==============================] - 0s 28us/step - loss: 0.1031 - val_loss: 0.0903\n", "Epoch 147/200\n", "1631/1631 [==============================] - 0s 27us/step - loss: 0.1028 - val_loss: 0.0902\n", "Epoch 148/200\n", "1631/1631 [==============================] - 0s 27us/step - loss: 0.1028 - val_loss: 0.0914\n", "Epoch 149/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0904\n", "Epoch 150/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1028 - val_loss: 0.0902\n", "Epoch 151/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0903\n", "Epoch 152/200\n", "1631/1631 [==============================] - 0s 32us/step - loss: 0.1031 - val_loss: 0.0906\n", "Epoch 153/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1032 - val_loss: 0.0904\n", "Epoch 154/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1031 - val_loss: 0.0902\n", "Epoch 155/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1028 - val_loss: 0.0903\n", "Epoch 156/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0904\n", "Epoch 157/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1028 - val_loss: 0.0903\n", "Epoch 158/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0902\n", "Epoch 159/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1029 - val_loss: 0.0907\n", "Epoch 160/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1031 - val_loss: 0.0922\n", "Epoch 161/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1028 - val_loss: 0.0902\n", "Epoch 162/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1029 - val_loss: 0.0904\n", "Epoch 163/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1032 - val_loss: 0.0906\n", "Epoch 164/200\n", "1631/1631 [==============================] - 0s 32us/step - loss: 0.1029 - val_loss: 0.0904\n", "Epoch 165/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1031 - val_loss: 0.0902\n", "Epoch 166/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1028 - val_loss: 0.0904\n", "Epoch 167/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1032 - val_loss: 0.0903\n", "Epoch 168/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1030 - val_loss: 0.0934\n", "Epoch 169/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1029 - val_loss: 0.0902\n", "Epoch 170/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1029 - val_loss: 0.0916\n", "Epoch 171/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1029 - val_loss: 0.0946\n", "Epoch 172/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1031 - val_loss: 0.0903\n", "Epoch 173/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1030 - val_loss: 0.0902\n", "Epoch 174/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1028 - val_loss: 0.0907\n", "Epoch 175/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1029 - val_loss: 0.0904\n", "Epoch 176/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0903\n", "Epoch 177/200\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "1631/1631 [==============================] - 0s 32us/step - loss: 0.1031 - val_loss: 0.0918\n", "Epoch 178/200\n", "1631/1631 [==============================] - 0s 32us/step - loss: 0.1029 - val_loss: 0.0912\n", "Epoch 179/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1029 - val_loss: 0.0917\n", "Epoch 180/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1030 - val_loss: 0.0908\n", "Epoch 181/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0904\n", "Epoch 182/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1032 - val_loss: 0.0920\n", "Epoch 183/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1031 - val_loss: 0.0903\n", "Epoch 184/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0908\n", "Epoch 185/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0907\n", "Epoch 186/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1030 - val_loss: 0.0908\n", "Epoch 187/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1029 - val_loss: 0.0903\n", "Epoch 188/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1029 - val_loss: 0.0914\n", "Epoch 189/200\n", "1631/1631 [==============================] - 0s 28us/step - loss: 0.1030 - val_loss: 0.0903\n", "Epoch 190/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1030 - val_loss: 0.0907\n", "Epoch 191/200\n", "1631/1631 [==============================] - 0s 28us/step - loss: 0.1031 - val_loss: 0.0906\n", "Epoch 192/200\n", "1631/1631 [==============================] - 0s 28us/step - loss: 0.1028 - val_loss: 0.0926\n", "Epoch 193/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1030 - val_loss: 0.0902\n", "Epoch 194/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1029 - val_loss: 0.0904\n", "Epoch 195/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0928\n", "Epoch 196/200\n", "1631/1631 [==============================] - 0s 31us/step - loss: 0.1029 - val_loss: 0.0906\n", "Epoch 197/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1028 - val_loss: 0.0923\n", "Epoch 198/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1032 - val_loss: 0.0904\n", "Epoch 199/200\n", "1631/1631 [==============================] - 0s 30us/step - loss: 0.1030 - val_loss: 0.0906\n", "Epoch 200/200\n", "1631/1631 [==============================] - 0s 29us/step - loss: 0.1029 - val_loss: 0.0907\n" ] }, { "data": { "text/plain": [ "" ] }, "execution_count": 76, "metadata": {}, "output_type": "execute_result" } ], "source": [ "model = Sequential()\n", "model.add(Dense(7, input_dim=1, activation='relu'))\n", "model.add(Dense(1))\n", "model.compile(loss='mean_squared_error', optimizer=Adam(lr=0.001))\n", "model.fit(X_train,y_train,validation_data=(X_test,y_test),epochs=200)" ] }, { "cell_type": "code", "execution_count": 78, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "[]" ] }, "execution_count": 78, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plt.plot(model.history.history['val_loss'])" ] }, { "cell_type": "code", "execution_count": 79, "metadata": {}, "outputs": [], "source": [ "predictions = model.predict(X_test)\n", "predictions = sc.inverse_transform(predictions)" ] }, { "cell_type": "code", "execution_count": 80, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "[]" ] }, "execution_count": 80, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plt.plot(predictions)" ] }, { "cell_type": "code", "execution_count": 81, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "[]" ] }, "execution_count": 81, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plt.plot(test[1:].values)\n", "plt.plot(predictions)" ] }, { "cell_type": "code", "execution_count": 82, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "0.09576390116629162\n", "0.6391697683986877\n" ] } ], "source": [ "MSE = mean_squared_error(test[1:], predictions)\n", "r2 = r2_score(test[1:], predictions)\n", "\n", "print(np.sqrt(MSE))\n", "print(r2)" ] }, { "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 }
X Tutup