Highway Travel Time Prediction of Segments Based on ANPR Data considering Traffic Diversion

Table 9

Long-term prediction error results of different methods.

Methods	Error	20 min	40 min	60 min
XGBoost	RMSE	13.7142	14.3634	14.9989
	MAE	3.7057	3.8906	4.6398
	MAPE (%)	17.1187	17.4168	22.1957
LR	RMSE	13.8199	15.5390	16.7464
	MAE	5.4188	6.5753	7.6697
	MAPE (%)	31.2934	38.2185	45.0381
SARIMA	RMSE	12.9227	14.2330	15.3928
	MAE	3.9210	4.5686	5.5299
	MAPE (%)	22.3183	25.6994	31.2392
CNNs	RMSE	13.4614	14.0973	14.8385
	MAE	5.5265	5.2283	5.6061
	MAPE (%)	31.7741	28.9882	30.7723
LSTM-ST	RMSE	14.1691	14.3815	15.0072
	MAE	4.7177	4.9093	5.6427
	MAPE (%)	23.8613	25.2212	30.0549
LSTM-CNN	RMSE	12.0201	13.3627	14.4035
	MAE	3.3805	3.8236	4.7018
	MAPE (%)	17.7759	19.3002	24.3612