Skip to main content

Table 12 Classification of difficulty classes based on pair surface features by decision tree

From: A detailed error analysis of 13 kernel methods for protein-protein interaction extraction

 

Performance

Confusion matrix

 

Difficulty class

P

R

F1

D

N

E

Total

difficult (D)

43.5

20.8

28.2

148

543

20

711

neutral (N)

92.0

96.2

94.1

178

14 090

372

14 640

easy (E)

72.6

60.0

65.7

14

678

1 037

1 729

Total

88.0

89.4

88.5

N/A

N/A

N/A

 
  1. Classification by the Weka J48 classifier. Confusion matrix columns correspond to predicted classes.