+ '.png',
dpi=700)
elif mode == 2:
plt.savefig(
'/home/vacek/Cloud/cloud-predictor/GoogleClusterTaskEvents/v2/Prediction-cs-100/LSTM-convnet/results/Normalized_RAM_Train_Loss'
+ '.png',
dpi=700)
plt.pause(3)
plt.close()
''' saving the trained model'''
if mode == 1:
model.save('model-CPU.h5') # creates a HDF5 file
elif mode == 2:
model.save('model-RAM.h5') # creates a HDF5 file
predicted = Train_LSTM.predict_point_by_point(model, X_test)
y_test = np.reshape(y_test, (y_test.size, ))
print(len(predicted), len(y_test), '------------')
del X_train, X_test
print('-----\n--------------\n--------------------------')
sleep(3)
print('Training duration (s) : ', time.time() - global_start_time)
print(' --------------------------------- ')
from sklearn.metrics import mean_squared_error
from math import sqrt
rms = sqrt(mean_squared_error(y_test, predicted))
ms = mean_squared_error(y_test, predicted)
import matplotlib.pyplot as plt
loss = history.history['loss']
val_loss = history.history['val_loss']
epochs = range(1, len(loss) + 1)
plt.figure()
plt.plot(epochs, loss, 'bo', label='Training loss')
plt.plot(epochs, val_loss, 'b', color='red', label='Validation loss')
plt.title('Training and validation loss')
plt.legend()
plt.pause(3)
plt.close()
# predictions = lstm.predict_sequences_multiple(model, X_test, seq_len, 50)
predicted = Train_LSTM.predict_point_by_point(model, X_test)
print(len(predicted), '------------')
# note that we can predict point by point each step using last true data!!!
predicted2 = []
for ii in range((X_test.shape[0])):
if ii > 1:
tmp = X_test[ii, :, :]
tmp[-1, :] = y_test[ii - 1]
input_seq = np.reshape(tmp, (1, tmp.shape[0], tmp.shape[1]))
else:
input_seq = np.reshape(
X_test[ii, :, :],
(1, X_test[ii, :, :].shape[0], X_test[ii, :, :].shape[1]))
# print(input_seq.shape)
y = Train_LSTM.predict_point_by_point(model, input_seq)