Skip to main content

Table 5 Classification performance of naïve Bayes algorithm on genes from van't Veer breast cancer dataset as training and independent test set of 234 samples for testing.

From: Refining gene signatures: a Bayesian approach

 

Boulle

Wrapper

Corr1

Corr10

Amsterdam Signature

Average # of genes

6

8

10

11

70

ACC (%)

66.24

60.68

61.54

63.25

61.54

SENS (%)

69.57

65.22

73.91

81.16

86.96

SPEC (%)

64.85

58.79

56.36

55.76

50.91

PPV (%)

45.28

39.82

41.46

43.41

42.55

NPV (%)

83.59

80.17

83.78

87.62

90.32

AUC

0.6721

0.6200

0.6514

0.6846

0.6893

  1. The average number of genes represents the average over the nested 10 fold CV.