Analysis of the Severity of Accidents on Rural Roads Using Statistical and Artificial Neural Network Methods

Table 9

Comparing the performance of mathematical models with proposed models.

Type of model	Learning method	CCR	Active function	MSE	RMSE	MCR
Logistic regression	—	74.9	—	—	—	25.1
MLP	Gradient descent (GD)	68.2	Tanh	—	—	31.8
MLP	Mini-Batch	70.5	Tanh	0.260	0.509	29.5
M1	Scale conjugate gradient (SCG)	69.7	Tanh	0.295	0.543	30.3
M2	Levenberg–Marquardt (LM)	—	Sigmoid	0.228	0.477	—