Bi-GRCN: A Spatio-Temporal Traffic Flow Prediction Model Based on Graph Neural Network

Table 1

Prediction results of the Bi-GRCN model and other baseline methods.

T (min)	Metric	Models
		HA	ARIMA	SVR	GCN	GRU	Bi-GRCN
15	RMSE	4.389446	6.998807	4.228722	5.609166	4.802862	4.548942
	MAE	3.006430	4.991264	2.882457	4.410619	3.464756	3.239479
	Accuracy	0.699633	0.455428	0.710631	0.616168	0.671343	0.688718
	var	0.788113	−0.000145	0.803415	0.654015	0.746349	0.772483
30	RMSE	4.389446	6.998712	4.252060	5.635812	4.487037	4.369150
	MAE	3.006430	4.990788	2.935113	4.453700	3.218210	3.101176
	Accuracy	0.699633	0.455422	0.709021	0.614327	0.692941	0.701008
	var	0.788113	−0.000347	0.802071	0.650719	0.778851	0.790520
45	RMSE	4.389446	6.997727	4.280222	5.664346	4.372606	4.347006
	MAE	3.006430	4.990163	2.978076	4.462996	3.116036	3.077341
	Accuracy	0.699633	0.455437	0.707092	0.612372	0.700769	0.702521
	var	0.788113	−0.000512	0.800080	0.647190	0.789771	0.792385
60	RMSE	4.389446	6.989135	4.307416	5.677034	4.333217	4.245528
	MAE	3.006430	4.986023	3.011925	4.489186	3.086672	3.000904
	Accuracy	0.699633	0.455653	0.705231	0.611504	0.703465	0.709466
	var	0.788113	0.000986	0.798030	0.645671	0.793562	0.801980