. 2019 Dec 13;9:19038. doi: 10.1038/s41598-019-55320-6

Table 2.

The results of DLSTM⁵ and LSTM-SAE using data set of case study 2.

Model	No. of hidden layer	No. of hidden units	Dropout	lag	batch	RMSE	MAE	SMAPE
DLSTM	1	⁴⁷	0.2	30	60	23.993	12.124	10.919
DLSTM	2	^3,42	0.2	30	73	23.750	12.452	12.181
LSTM-SAE	1	²⁰	0.1	30	219	23.907	12.509	11.052
LSTM-SAE	2	^25,26	0.3	25	146	24.041	12.060	9.864