Table 5.

Mean absolute error for artificial dataset.

Method	Set	Average Value	Minimum Value	Maximum Value	Standard Deviation
Seasonal ARIMA	Training	14.148	NA
	Test	13.644
Support vector regression	Training	0.132 *	NA
	Test	8.541
Multi-layer perceptron	Training	1.552	0.399	5.461	1.570
	Test	8.283	7.595	10.173	0.724
Convolutional neural network	Training	0.580	0.221	0.867	0.238
	Test	10.012	9.711	10.319	0.190
LSTM-EEMD	Training	17.750	16.628	19.368	1.435
	Test	18.308	17.658	19.382	0.937
LSTM RNN many-to-one	Training	15.591	8.107	29.257	6.113
	Test	17.447	8.327	29.060	7.196
LSTM RNN many-to-many	Training	8.521	8.519	8.524	0.001
	Test	7.984 *	7.965	7.990	0.007

*: Best result.