Rail Transit Prediction Based on Multi-View Graph Attention Networks

Table 2

Accuracy results of rail passenger flow prediction experiment.

Methods	November		December
	MAE	RMSE	MAE	RMSE
ARIMA	14.22/18.81/24.39	30.06/33.52/35.24	18.39/20.52/24.22	30.24/33.81/35.06
SVR	13.92/16.52/16.12	26.89/28.75/28.56	14.12/15.75/16.92	26.56/28.52/28.89
LSTM	11.49/13.79/14.05	21.36/22.33/25.50	11.05/12.33/14.49	21.50/22.79/25.36
T-GCN	10.99/12.76/13.44	21.48/23.68/25.62	10.90/13.78/13.34	21.42/23.74/25.85
STGCN	9.06/10.65/11.32	21.74/22.22/23.55	8.20/10.26/11.25	20.72/22.28/23.25
DCRNN	8.83/9.71/11.79	21.52/23.15/26.04	8.24/9.65/11.87	21.28/23.24/26.95
MV-GAT	8.80/9.62/11.20	20.83/22.18/23.39	8.15/9.57/11.43	20.64/22.13/24.72