Table 4.

Driving behavior identification for different methods when setting $T_x$ to 60 samples and $\Delta t$ to 10 samples.

Model	${\mathit{T}}_{\mathit{x}}=60, \mathsf{\Delta }\mathit{t}=10$
	Accuracy		AUC		F1 Score
	Mean	Std	Mean	Std	Mean	Std
KNN	0.812	0	0.8986	0	0.8157	0
DecisionTree	0.7432	0.0966	0.8638	0.0502	0.7402	0.1002
RandomForest	0.7049	0.0589	0.8541	0.0275	0.7565	0.0485
DeepConvGRU-Attention	0.9701	0.0052	0.9959	0.0007	0.9702	0.0052
DeepConvLSTM-Attention	0.9524	0.0138	0.9946	0.0015	0.9526	0.0137
DeepConvGRU	0.9565	0.0091	0.9959	0.0011	0.9563	0.0093
DeepConvLSTM	0.905	0.0062	0.9887	0.0008	0.9047	0.0064
CNN	0.9081	0.0096	0.993	0.0013	0.908	0.0096
LSTM-15	0.9774	0.012	0.9982	0.0008	0.9771	0.0124
DNN-45	0.9188	0.0301	0.9731	0.0164	0.9145	0.0337