Table 2. Results of different methods on the dataset PEMSM.

Classification	Methods	PEMSM (10/15/20 min)
		MAE	MAPE (%)	RMSE
Classical methods	HA	3.14/3.22/3.32	7.60/7.81/8.03	6.09/6.29/6.49
	SVR	1.78/2.01/2.21	4.45/5.03/5.54	3.44/3.92/4.36
	RF	1.82/2.10/2.34	4.24/4.95/5.60	3.19/3.76/4.25
	KNN	1.96/2.21/2.43	4.64/5.28/5.87	3.48/3.99/4.44
State-of-the-Art methods	STGCN	1.87/2.29/2.60	4.32/5.38/6.19	3.16/3.93/4.46
	FC-GAGA	1.86/2.20/2.45	4.36/5.26/5.98	3.38/4.16/4.77
	AGCRN	1.82/1.96/2.09	4.39/4.75/5.08	3.28/3.60/3.89
Our methods	GST-HCN	1.74/2.00/2.21	4.15/4.79/5.37	2.94/3.41/3.82
	HST-HCN	1.72/1.97/2.19	4.07/4.74/5.29	2.90/3.39/3.75
	LST-HCN	1.68/1.96/2.18	3.87/4.56/5.22	2.88/3.40/3.82
	ST-HCN	1.62/1.86/2.08	3.72/4.34/4.98	2.77/3.23/3.63

Notes.

The best experimental results for each setting are bolded.