Short-Term Traffic Flow Prediction of Expressway: A Hybrid Method Based on Singular Spectrum Analysis Decomposition

Table 2

The performance of different methods on Guiyang City Road Toll Stations.

		Time interval being 15 min/30 min/60 min
Traffic volume (TV)	Models	MAE	MAPE	RMSE	Accuracy	R^2
15 stations (TV ≥ 6000)	ARIMA	12.33/25.51/61.83	0.29/0.31/0.43	17.37/35.46/86.77	0.84/0.84/0.81	0.92/0.92/0.89
	SVR	12.68/25.74/61.53	0.19/0.20/0.23	17.96/36.89/93.12	0.84/0.84/0.80	0.92/0.91/0.87
	KNN	15.83/30.30/56.46	0.27/0.25/0.24	22.65/43.35/78.75	0.80/0.81/0.81	0.87/0.87/0.90
	BP	16.15/37.43/92.90	0.57/0.67/0.88	20.86/47.83/118.68	0.82/0.80/0.75	0.90/0.87/0.82
	LSTM	13.71/27.16/84.17	0.39/0.40/0.50	18.54/37.87/119.20	0.84/0.83/0.75	0.92/0.91/0.82
	SSA-LSTM-SVR	9.47/18.64/45.51	0.15/0.14/0.18	15.92/29.05/56.83	0.89/0.89/0.86	0.94/0.94/0.91
15 stations (2000 ≤ TV < 6000)	ARIMA	6.86/12.33/24.49	0.84/0.38/0.41	9.58/17.13/33.55	0.80/0.82/0.82	0.89/0.90/0.90
	SVR	7.02/12.44/24.50	0.23/0.23/0.25	9.83/17.6334.79	0.80/0.82/0.82	0.88/0.90/0.89
	KNN	8.41/14.44/26.95	0.35/0.31/0.36	12.01/20.56/38.92	0.75/0.79/0.80	0.82/0.86/0.88
	BP	8.12/17.05/35.58	0.64/0.74/0.82	10.49/21.06/43.30	0.78/0.78/0.77	0.86/0.86/0.84
	LSTM	6.95/13.31/30.57	0.38/0.44/0.56	9.58/17.59/41.23	0.80/0.82/0.79	0.88/0.90/0.87
	SSA-LSTM-SVR	5.22/10.01/20.51	0.19/0.18/0.20	7.27/15.36/28.07	0.85/0.88/0.88	0.90/0.93/0.93
15 stations (TV < 2000)	ARIMA	4.51/8.03/16.57	0.43/0.50/0.62	6.33/11.41/23.45	0.74/0.76/0.75	0.82/0.85/0.83
	SVR	4.55/8.08/16.68	0.26/0.26/0.31	6.46/11.73/24.66	0.74/0.76/0.75	0.82/0.84/0.82
	KNN	5.79/10.18/18.45	0.44/0.41/0.69	8.38/15.14/27.76	0.66/0.69/0.71	0.69/0.73/0.76
	BP	5.15/9.83/21.65	0.64/0.81/1.09	6.85/12.86/27.49	0.72/0.74/0.72	0.79/0.81/0.78
	LSTM	4.56/8.81/19.00	0.44/0.62/0.90	6.49/12.15/26.33	0.74/0.75/0.72	0.81/0.83/0.79
	SSA-LSTM-SVR	3.57/7.58/12.21	0.21/0.18/0.19	5.49/10.54/18.46	0.80/0.83/0.84	0.85/0.88/0.86