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

Table 1

Performance comparison of prediction models (Guiyang North Station).

Time interval (min)	Models	MAE	MAPE	RMSE	Accuracy	R^2
5	ARIMA	6.4677	0.2394	9.0327	0.8574	0.9399
	SVR	6.6324	0.1778	9.273	0.8569	0.9369
	BP	7.6623	0.3671	10.1135	0.844	0.9250
	KNN	8.8914	0.2836	12.2265	0.8116	0.8905
	LSTM	6.8288	0.2261	9.364	0.8547	0.9348
	SSA-LSTM-SVR	4.7361	0.1284	6.821	0.9106	0.9421
10	ARIMA	11.1200	0.19999	15.6300	0.8763	0.9541
	SVR	10.7635	0.1765	15.388	0.8812	0.9559
	BP	11.2143	0.2161	15.6944	0.8788	0.9541
	KNN	14.6499	0.2261	20.4401	0.8422	0.9222
	LSTM	11.2955	0.1866	15.9469	0.8768	0.9525
	SSA-LSTM-SVR	8.8346	0.1206	13.2418	0.9385	0.9623
15	ARIMA	15.4965	0.1973	21.6791	0.8853	0.9603
	SVR	15.9333	0.1588	22.16	0.8857	0.9587
	BP	21.4699	0.3244	27.8068	0.8566	0.9351
	KNN	21.4068	0.2179	30.2227	0.8441	0.9238
	LSTM	16.1834	0.1913	22.2071	0.8855	0.9588
	SSA-LSTM-SVR	10.1942	0.1376	16.9009	0.9320	0.9758
30	ARIMA	30.6037	0.2118	42.1165	0.8885	0.9622
	SVR	31.1061	0.1402	43.5484	0.8877	0.9599
	BP	48.4023	0.5103	59.7906	0.8457	0.9246
	KNN	39.3415	0.2543	54.7551	0.8577	0.9367
	LSTM	38.1273	0.3719	49.2125	0.8724	0.9489
	SSA-LSTM-SVR	25.2886	0.1273	33.985	0.9324	0.9502
45	ARIMA	47.3747	0.1911	66.9107	0.8818	0.9573
	SVR	47.8773	0.1496	68.8685	0.8812	0.9552
	BP	73.554	0.4118	94.3918	0.837	0.9159
	KNN	57.8264	0.2016	78.2123	0.8634	0.9421
	LSTM	69.918	0.2824	97.7382	0.8297	0.9093
	SSA-LSTM-SVR	42.4357	0.0864	56.9149	0.9309	0.9528
60	ARIMA	74.2852	0.2414	104.810	0.8612	0.9410
	SVR	77.6223	0.1775	113.9067	0.8525	0.9309
	BP	125.1846	0.5853	157.7238	0.7952	0.8675
	KNN	85.3978	0.2353	114.6119	0.8499	0.9305
	LSTM	117.9721	0.3366	169.38	0.778	0.847
	SSA-LSTM-SVR	68.2209	0.1246	93.5078	0.9107	0.9493