Table 9. Comparison of prediction measures of different models.

Model	Metric
	MSE	MAE	sMAPE	RSE	CORR	SA
KCS-LSTM	4.1432	1.0966	0.0216	0.0691	0.99762	0.4949
ACO-LSTM	4.3614	1.2492	0.0245	0.0709	0.99777	0.4039
KCS-RNN	5.0949	1.2717	0.0247	0.0767	0.99724	0.4479
ACO-RNN	4.6749	1.2516	0.0245	0.0734	0.99759	0.4160
ACO-BP	6.5208	1.6372	0.0327	0.0867	0.99660	0.4387
BP	5.7137	1.4748	0.0282	0.0812	0.99714	0.4532
Informer	19.1286	3.5976	0.0782	0.1483	0.99602	0.3584
LSTNet	15.0160	3.1355	0.0625	0.1315	0.99677	0.3717
SCINet	4.3147	1.2676	0.0248	0.0705	0.99752	0.4444
KCS+LSTM	11.5543	2.8504	0.0622	0.1155	0.99782	0.3568
IMP	4.0%	12.2%	11.9%	2.0%	–	9.2%

Notes.

^aThe best results are highlighted with bold underline and second best results are shown in italic bold.

^bIMP shows the improvement of KCS-LSTM over the best model.