Skip to main content

Table 2 Performance of classifiers on test data

From: PeerShark: flow-clustering and conversation-generation for malicious peer-to-peer traffic identification

 

Decision trees

Random forests

Bayesian network

Class

Precision

Recall

FP rate

Precision

Recall

FP rate

Precision

Recall

FP rate

Malicious

95.3%

93.4%

0.2%

95.3%

94.9%

0.2%

91.9%

88.4%

0.3%

Benign

99.7%

99.8%

6.6%

99.8%

99.8%

5.1%

99.5%

99.7%

11.6%