Table 5. Prediction accuracy of each combined model for subsampled data.

Model		Evaluation
	MAPE	RMSE	R2
Attention-BiLSTM	3.30%	47,263.84	0.85
PSO-Attention-BiLSTM	2.93%	41,568.20	0.88
APSO-Attention-BiLSTM	1.93%	32,931.93	0.93