Analysis and Prediction of Overloaded Extra-Heavy Vehicles for Highway Safety Using Machine Learning

Table 4

Prediction results of BPNN.

Data name	Highway data	National and provincial highway data (month)	National and provincial highway data (quarter)	National and provincial highway data (year)
Number of hidden nodes	16	3	8	6
Iteration times	3	3	5	6
Error descent gradient	0.008	0.005	0.019	0.034
	0.987	0.967	0.962	0.972
MSE (%^2)	0.090	0.055	0.093	0.059
RMSE (%)	0.960	2.960	3.060	1.460
ME (%)	2.700	−0.150	38.750	-3.200
MAE (%)	5.200	2.120	44.330	3.840
MAPE (%)	0.089	0.124	0.116	1.046
RMSPE (%)	0.059	0.026	0.058	0.894
Running time (s)	2.016	1.938	0.001	2.953