Skip to main content

Table 2 Predicting performance with different classifiers

From: Predicting combinative drug pairs towards realistic screening via integrating heterogeneous features

 

S1

S2

S3

AUC

AUPR

AUC

AUPR

AUC

AUPR

LR

0.954

0.821

0.909

0.635

0.809

0.592

SVM_Linear

0.904

0.639

0.856

0.470

0.720

0.373

SVM_RBF

0.938

0.821

0.904

0.638

0.833

0.609

  1. LR is logistic regression, SVM_Linear and SVM_RBF are the SVMs with linear kernel and RBF kernel respectively. The cost parameter is fixed with 100 and the sharp parameter γ of RBF are assigned with 0.02, 0.05 and 0.001 in S1, S2 and S3 respectively when training SVM. The bold entries highlight the best results