Metro Traffic Flow Prediction via Knowledge Graph and Spatiotemporal Graph Neural Network

Table 4

Traffic flow prediction results on the BJMF15 dataset.

Baselines		SVR	LSTM	GCN	TGCN	STGCN	KGR-STGNN
Prediction steps	Evaluation indicators
12	MAE	22.280	21.883	21.747	11.827	8.365	8.211
	RMSE	48.789	42.368	45.178	20.928	17.326	16.296
15	MAE	25.306	25.753	23.211	12.023	9.632	8.996
	RMSE	55.425	47.978	47.284	21.171	20.318	17.568
18	MAE	27.677	27.949	25.110	13.168	11.135	9.624
	RMSE	60.380	52.644	51.440	23.107	22.413	18.998
21	MAE	29.832	29.381	26.309	13.113	13.569	11.403
	RMSE	64.463	55.423	53.655	23.427	26.988	22.989