Traffic Flow Prediction with Rainfall Impact Using a Deep Learning Method

Table 2

Comparisons between different predictors.

Model	Measurement	10-minute prediction	30-minute prediction
R-DBN	MAE (veh/h)	178.90	325.58
	MAPE (%)	13.14	19.75
	RMSE (veh/h)	255.79	356.49
R-LSTM	MAE (veh/h)	166.17	305.04
	MAPE (%)	11.69	17.88
	RMSE (veh/h)	240.98	296.91
R-BPNN	MAE (veh/h)	192.70	337.34
	MAPE (%)	15.59	21.80
	RMSE (veh/h)	298.22	377.86
R-ARIMA	MAE (veh/h)	236.24	409.75
	MAPE (%)	18.47	28.14
	RMSE (veh/h)	331.82	447.06
DBN	MAE (veh/h)	186.15	327.06
	MAPE (%)	13.58	20.19
	RMSE (veh/h)	265.16	374.33
LSTM	MAE (veh/h)	172.63	315.15
	MAPE (%)	12.84	19.55
	RMSE (veh/h)	255.04	346.83
BPNN	MAE (veh/h)	200.44	352.21
	MAPE (%)	16.76	22.81
	RMSE (veh/h)	302.74	375.32
ARIMA	MAE (veh/h)	227.85	369.54
	MAPE (%)	17.37	25.95
	RMSE (veh/h)	313.65	413.86