Evaluation of Short-Term Freeway Speed Prediction Based on Periodic Analysis Using Statistical Models and Machine Learning Models

Table 1

Improvement percentage of hybrid models for different prediction horizons in peak hours at station C (time scale: 5 min).

MAE	Prediction horizons (minutes)
	5	15	30	60
ST	0.1678	2.0201	6.5398	18.4136
VAR	0.7073	2.4666	7.7073	16.9290
ARIMA	0.6583	2.7986	13.5338	28.3774
SVM	2.7703	0.8935	9.5560	21.5672
MLP	0.3708	0.6867	5.3738	23.8708
RNN	1.4256	4.5202	8.9773	15.5103
MAPE (%)	Prediction horizons (minutes)
	5	15	30	60
ST	−0.0816	3.1284	11.1746	27.1694
VAR	5.5966	12.2921	17.4324	27.7299
ARIMA	3.1773	10.5383	23.3195	38.2678
SVM	5.0060	2.4855	16.7698	35.2814
MLP	2.4710	5.2158	10.1209	38.1002
RNN	−1.3680	4.7849	18.4671	25.7139
RMSE	Prediction horizons (minutes)
	5	15	30	60
ST	2.2066	7.9375	12.1864	21.1225
VAR	1.4003	−2.3589	4.3848	17.1806
ARIMA	5.0394	11.0708	20.5272	32.1107
SVM	5.2385	7.5790	16.1069	28.5130
MLP	3.0359	9.1199	15.6351	28.6400
RNN	6.6514	11.1524	14.1581	21.2511