TABLE III. Performance and Hyper-Parameters of Four Classifiers Based on Handcrafted Features and DL Score.

Classifiers	Hyper-parameters	RSD (%)	Training AUC (mean±SD)	Test AUC (mean±SD)
Logistics Regression	C = 1000 Penalty = ‘L1’	8.092	0.854±0.010	0.848±0.097
SVM	C = 0.125 Gamma = ‘10.0’ Kernel = ‘rbf’	8.203	0.806±0.010	0.804±0.093
Decision Tree	Criterion = ‘gini’ Min samples leaf = 19 Min samples split = 2	9.541	0.900±0.011	0.805±0.092
Random Forest	Criterion = ‘gini’ Min samples leaf = 5 Min samples split = 2 Max features = ‘auto’ N estimators = 10	10.20	0.981±0.004	0.839±0.085