Skip to main content

Table 3 Sensitivity and specificity of various classifiers trained on the hepatobiliary data set for difference percentages of under-sampling

From: Improving sensitivity of machine learning methods for automated case identification from free-text electronic medical records

Under-sampling

SVM

MyC

RIPPER

C4.5

Imbalance

(%)

Sens.

Spec.

Sens.

Spec.

Sens.

Spec.

Sens.

Spec.

ratio

0

0.89

0.77

0.92

0.68

0.91

0.71

0.90

0.79

42

10

0.89

0.76

0.93

0.65

0.91

0.75

0.90

0.80

38

20

0.89

0.75

0.93

0.63

0.91

0.73

0.91

0.79

34

30

0.89

0.76

0.94

0.61

0.93

0.72

0.90

0.78

30

40

0.89

0.73

0.93

0.60

0.92

0.69

0.91

0.77

25

50

0.90

0.70

0.93

0.58

0.92

0.71

0.91

0.76

21

60

0.90

0.71

0.94

0.56

0.92

0.72

0.92

0.73

17

70

0.91

0.67

0.95

0.55

0.91

0.72

0.92

0.70

13

80

0.92

0.64

0.94

0.49

0.92

0.73

0.92

0.68

9

90

0.94

0.52

0.91

0.60

0.93

0.67

0.93

0.59

5

100

0.99

0.12

0.99

0.07

0.99

0.03

0.99

0.14

0.5