Table 5. Summary of tree optimizations for the 5 best models.

Feature Selector	Trees	Accuracy (%)	Sensitivity (%)	Specificity (%)	MCC	F1
T-CFS	500	82.0	56.3	96.5	0.61	0.69
S&T-CFS	500	80.9	56.3	94.7	0.58	0.69
S-ETC10	250	71.9	46.9	86.0	0.36	0.55
S-RelF10	50	71.9	50.0	84.2	0.37	0.56
AS-RelF30	150	70.8	50.0	82.5	0.34	0.55

T = Turn, S = Straight, S&T = Straight and Turn, AS = All steps, MCC = Matthews Correlation Coefficient, F1 = F1 score.