lstm-for-epf.py

import numpy as np
import pandas as pd
from matplotlib import pyplot as plt

from tensorflow.contrib import learn
from sklearn.metrics import mean_squared_error, mean_absolute_error
from lstm_predictor import generate_data, load_csvdata, lstm_model


LOG_DIR = './ops_logs'
TIMESTEPS = 10
RNN_LAYERS = [{'steps': TIMESTEPS}]
DENSE_LAYERS = [10, 10]
TRAINING_STEPS = 100000
BATCH_SIZE = 100
PRINT_STEPS = TRAINING_STEPS / 100

dateparse = lambda dates: pd.datetime.strptime(dates, '%d/%m/%Y %H:%M')
rawdata = pd.read_csv("./input/ElectricityPrice/RealMarketPriceDataPT.csv", 
                   parse_dates={'timeline': ['date', '(UTC)']}, 
                   index_col='timeline', date_parser=dateparse)


X, y = load_csvdata(rawdata, TIMESTEPS, seperate=False)


regressor = learn.TensorFlowEstimator(model_fn=lstm_model(TIMESTEPS, RNN_LAYERS, DENSE_LAYERS), 
                                      n_classes=0,
                                      verbose=1,  
                                      steps=TRAINING_STEPS, 
                                      optimizer='Adagrad',
                                      learning_rate=0.03, 
                                      batch_size=BATCH_SIZE)




validation_monitor = learn.monitors.ValidationMonitor(X['val'], y['val'],
                                                      every_n_steps=PRINT_STEPS,
                                                      early_stopping_rounds=1000)

regressor.fit(X['train'], y['train'], monitors=[validation_monitor], logdir=LOG_DIR)


predicted = regressor.predict(X['test'])
mse = mean_absolute_error(y['test'], predicted)
print ("Error: %f" % mse)

plot_predicted, = plt.plot(predicted, label='predicted')
plot_test, = plt.plot(y['test'], label='test')
plt.legend(handles=[plot_predicted, plot_test])