Multicomponent Spatial-Temporal Graph Attention Convolution Networks for Traffic Prediction with Spatially Sparse Data

Table 1

Performance comparison of different methods on PeMS-BAY with different number of nodes.

Model	24 nodes		56 nodes		88 nodes		232 nodes		325 nodes
	MAE	RMSE	MAE	RMSE	MAE	RMSE	MAE	RMSE	MAE	RMSE
HA	3.01	6.15	3.20	6.37	3.11	6.08	2.90	5.67	2.88	5.99
VAR	2.31	4.52	2.44	4.55	2.40	4.44	2.31	4.14	2.33	4.12
DCRNN	1.94	4.46	2.02	4.69	1.90	4.25	1.75	3.87	1.73	3.89
STGCN	2.22	4.84	2.25	4.82	2.18	4.66	1.89	4.15	1.89	4.31
ST-MetaNet	1.79	4.35	1.85	4.32	1.77	4.13	1.78	4.16	1.75	4.09
GraphWaveNet	1.79	4.21	1.79	4.10	1.73	3.87	1.60	3.58	1.58	3.54
MSTGACN (ours)	1.70	3.90	1.72	3.84	1.69	3.77	1.63	3.57	1.63	3.58