Table 2.

Comparison of GTB with other typical classifiers on heterogenous network-derived features

Method	Precision	Recall	F-Measure	MCC	AUC
GTB	0.897	0.872	0.884	0.772	0.949
kNN	0.738	0.833	0.783	0.542	0.768
SVM	0.882	0.779	0.840	0.728	0.859
Logistic	0.499	0.527	0.510	0.014	0.520
Naive Bayes	0.504	0.988	0.770	0.086	0.508
Random forest	0.880	0.841	0.862	0.733	0.866
AdaBoost	0.878	0.854	0.863	0.732	0.866
LogitBoost	0.803	0.820	0.811	0.617	0.808

The boldface figures indicate that GTB achieves the best performance compared with other typical classifiers on heterogenous network-derived features