Table 4.

Classification results of the training set (cross-validation approach) and test set with new unseen data.

Classifier	Training set (cross-validation approach)							Test set (new unseen data)
	Acc.	BAcc	Sens.	Spec.	Prec.	F1	#F	Acc.	BAcc	Sens.	Spec.	Prec.	F1
Random Forests	0.78 (0.02)	0.74 (0.03)	0.64 (0.09)	0.84 (0.05)	0.66 (0.06)	0.65 (0.05)	2(N), 1(C)	0.75	0.71	0.58	0.83	0.62	0.60
Extra Trees	0.76 (0.04)	0.72 (0.03)	0.61 (0.07)	0.83 (0.08)	0.66 (0.10)	0.63 (0.04)	3(N), 1(C)	0.74	0.70	0.59	0.81	0.61	0.59
AdaBoost	0.73 (0.06)	0.69 (0.05)	0.50 (0.06)	0.88 (0.05)	0.68 (0.09)	0.57 (0.06)	3(N), 1(C)	0.73	0.64	0.48	0.85	0.60	0.54
Gradient Boosting	0.76 (0.04)	0.71 (0.05)	0.59 (0.11)	0.84 (0.04)	0.63 (0.04)	0.60 (0.07)	3(N), 1(C)	0.72	0.64	0.41	0.87	0.60	0.49
XGBoost	0.77 (0.04)	0.73 (0.05)	0.65 (0.11)	0.82 (0.07)	0.65 (0.09)	0.64 (0.06)	3(N), 2(C)	0.74	0.69	0.55	0.83	0.61	0.58

Acc, accuracy; Bacc, balanced accuracy; Sens, sensitivity; Spec, specificity; Prec, precision; F1, F1-score; #F, number of features; N, numerical; C, categorical. The results indicates mean and standard deviation in parenthesis.