ST-LSTM: A Deep Learning Approach Combined Spatio-Temporal Features for Short-Term Forecast in Rail Transit

Table 2

Forecast performance of ST-LSTM network on different stations.

Station	ME	MAE	RMSE	MRE
321	366.96	71.08	99.12	13.52%
315	365.72	52.45	73.13	16.90%
318	280.70	32.45	47.31	11.13%
114	270.84	44.66	60.63	13.76%
322	215.52	30.41	41.12	12.10%
110	252.81	37.09	50.87	12.90%
606	238.99	33.44	48.46	11.97%
212	204.32	39.95	52.62	12.78%
108	215.70	32.76	47.10	13.94%
123	223.77	61.18	81.89	36.58%
335	69.81	6.58	10.10	29.34%
221	43.85	6.83	9.07	28.39%
305	58.25	9.55	12.93	32.25%
620	67.57	6.46	9.59	27.07%
621	45.15	6.69	8.91	30.47%
224	31.58	5.99	7.98	32.93%
219	36.12	5.40	7.40	31.86%
619	35.96	4.77	6.56	29.32%
118	23.63	4.41	5.72	29.88%
625	49.00	5.47	8.04	35.59%