i = 0
# Perform dimensionality reduction on the feature vectors
pca = PCA(n_components=dimensions[x])
pca.fit(X_train)
xTrainRed = pca.transform(X_train)
xTestRed = pca.transform(X_test)
predicted_values = fit_predict(xTrainRed, y_train, xTestRed, 40, neurons[x])
mape = statistics.mape((y_test + shifted_value) * 1000, (predicted_values + shifted_value) * 1000)
print('MAPE is ', mape)
mae = statistics.mae((y_test + shifted_value) * 1000, (predicted_values + shifted_value) * 1000)
print('MAE is ', mae)
mse = statistics.meanSquareError((y_test + shifted_value) * 1000, (predicted_values + shifted_value) * 1000)
print('MSE is ', mse)
rmse = math.sqrt(mse)
print('RMSE is ', rmse)
nrmse = statistics.normRmse((y_test + shifted_value) * 1000, (predicted_values + shifted_value) * 1000)
print('NRMSE is ', nrmse)
print (i+1)
preds.append(predicted_values)
# show
fig = plt.figure()
colors = ["g","r","b","c","m","y","k","w"]
legendVars = []
for j in range(len(preds)):
print(j)
x, = plt.plot(preds[j]+shifted_value, color=colors[j])
legendVars.append(x)
plt.xlabel('Hour')
plt.ylabel('Electricity load, kW')
plt.legend(legendVars, names)
plt.show()
plt.show()
'''
# show result of predictions
# pr = pred2.predicted_mean['2016-07-10 19:00:00':'2016-07-31 23:00:00']+shifted_value
# ax = (data['2016-07-10 19:00:00':'2016-07-31 23:00:00']+shifted_value).plot(figsize=(20, 16))
pr = pred2.predicted_mean['2016-12-01 22:00:00':'2017-05-16 21:00:00']
ax = (data['2016-12-01 22:00:00':'2017-05-16 21:00:00']).plot(figsize=(20, 16))
# plot the results
fig = plt.figure()
plt.plot(ax, label="$true$", c='green')
plt.plot(pr, label="$predict$", c='red')
plt.xlabel('Hour')
plt.ylabel('Electricity load (*1e3)')
plt.legend()
plt.show()
fig.savefig('../result/sarimax.jpg', bbox_inches='tight')
# quantify the accuracy of the prediction
prediction = pred2.predicted_mean['2016-12-01 22:00:00':'2017-05-16 21:00:00'].values
# flatten nested list
truth = list(itertools.chain.from_iterable(test_data.values))
# evaluation
mape = np.mean(np.abs((truth - prediction) / truth))
print('mape is {:.5f}'.format(mape))
mae = statistics.mae(truth,prediction)
print('MAE is ', mae)
mse = statistics.meanSquareError(truth,prediction)
print('MSE is ', mse)
rmse = statistics.Rmse(mse)
print('RMSE is ', rmse)
nrmse = statistics.normRmse(truth,prediction)
print('NRMSE is ', nrmse)
mape = statistics.mape([(y_test_true[i] + shifted_value) * 1000
for i in range(0, len(y_test_true))],
(predicted_values + shifted_value) * 1000)
print('MAPE is ', mape)
mae = statistics.mae([(y_test_true[i] + shifted_value) * 1000
for i in range(0, len(y_test_true))],
(predicted_values + shifted_value) * 1000)
print('MAE is ', mae)
mse = statistics.meanSquareError([(y_test_true[i] + shifted_value) * 1000
for i in range(0, len(y_test_true))],
(predicted_values + shifted_value) * 1000)
print('MSE is ', mse)
rmse = math.sqrt(mse)
print('RMSE is ', rmse)
nrmse = statistics.normRmse([(y_test_true[i] + shifted_value) * 1000
for i in range(0, len(y_test_true))],
(predicted_values + shifted_value) * 1000)
print('NRMSE is ', nrmse)
fig = plt.figure(figsize=(12, 9), dpi=100)
plt.subplot(2, 1, 1)
plt.plot(ca_y_test, label="$lowpass_real$", c='green')
plt.plot(ca_pred, label="$lowpass_prediction$", c='red')
plt.legend(['lowpass_real', 'lowpass_prediction'], loc='upper right')
plt.title('lowpass coefficient prediction result', fontsize=16)
plt.subplot(2, 1, 2)
# plt.plot(cd[train_row:], label="$highpass_real$", c='green')
plt.plot(cd_y_test, label="$highpass_real$", c='green')
plt.plot(cd_pred, label="$highpass_prediction$", c='red')
plt.legend(['highpass_real', 'highpass_prediction'], loc='upper right')
plt.title('highpass coefficient prediction result', fontsize=16)
# save model
model.save('../model/seq2seq_code.h5')
model = load_model('../model/seq2seq_code.h5')
# evaluate on some new patterns
result = model.predict(X_test, batch_size=50, verbose=0)
# calculate error,evaluate the result
predicted = one_hot_decode(result, series_min, series_max, n_unique)
# plot the results
fig = plt.figure()
plt.plot(test[:, n_in:] + shifted_value, label="$true$", c='green')
plt.plot(predicted + shifted_value, label="$predict$", c='red')
plt.xlabel('Hour')
plt.ylabel('Electricity load')
plt.legend()
plt.show()
fig.savefig('../result/seq2seq_code_result.jpg', bbox_inches='tight')
# evaluation
mape = statistics.mape(test[:, n_in:] + shifted_value,
predicted + shifted_value)
print('MAPE is ', mape)
nrmse = statistics.normRmse(test[:, n_in:] + shifted_value,
predicted + shifted_value)
print('NRMSE is ', nrmse)
mae = statistics.mae(test[:, n_in:] + shifted_value,
predicted + shifted_value)
print('MAE is ', mae)
mse = statistics.mse(test[:, n_in:] + shifted_value,
predicted + shifted_value)
print('MSE is ', mse)
rmse = sqrt(mse)
print('RMSE is ', rmse)
