An Alternative Method for Traffic Accident Severity Prediction: Using Deep Forests Algorithm

Table 3

Average predictive performance of different models.

	Accuracy (%)	Recall	False alarm rates	F1 score	Roc
Deep forests	90.69	0.92	0.09	0.91	0.93
RFs	88.98	0.90	0.10	0.90	0.92
XGboost	83.49	0.83	0.16	0.83	0.87
LightGBM	83.01	0.83	0.17	0.83	0.87
Decision tree	81.04	0.81	0.19	0.81	0.85
KNN	77.26	0.77	0.23	0.77	0.82
DNN	53.52	0.54	0.47	0.47	0.52