Skip to main content

Table 3 AUCs for the EBD and FI discretization methods

From: Application of an efficient Bayesian discretization method to biomedical data

Classifier

C4.5

NB

Dataset

EBD

FI

EBD

FI

1

98.00% (0.06)

98.00% (0.06)

69.32% (0.88)

66.79% (0.92)

2

73.19% (1.08)

69.37% (1.22)

78.58% (1.87)

79.96% (1.98)

3

57.24% (1.88)

55.42% (1.65)

56.08% (1.70)

54.16% (1.92)

4

68.37% (1.27)

69.43% (0.95)

58.12% (1.08)

59.72% (1.17)

5

55.21% (1.12)

54.38% (1.44)

56.87% (1.41)

53.91% (1.09)

6

61.54% (0.63)

60.11% (0.95)

88.21% (0.66)

86.38% (0.86)

7

88.45% (1.42)

88.45% (1.42)

91.35% (0.76)

91.35% (0.76)

8

54.11% (1.12)

55.49% (0.89)

58.76% (0.85)

59.61% (0.76)

9

88.34% (1.32)

86.90% (1.41)

87.65% (1.18)

84.28% (1.12)

10

76.45% (0.68)

74.30% (0.81)

85.44% (0.99)

82.59% (1.04)

11

68.25% (0.71)

66.12% (0.61)

72.38% (1.01)

70.74% (0.98)

12

56.65% (1.21)

55.14% (1.06)

57.89% (0.95)

53.72% (0.86)

13

70.45% (0.87)

73.18% (0.65)

69.89% (0.71)

71.55% (0.75)

14

56.32% (1.12)

55.16% (0.98)

54.42% (0.98)

55.12% (0.96)

15

76.12% (0.87)

73.49% (1.01)

89.45% (0.89)

91.27% (0.56)

16

82.21% (1.31)

80.06% (1.12)

82.86% (1.17)

80.11% (1.09)

17

78.65% (1.41)

80.15% (1.32)

78.14% (1.12)

75.98% (1.24)

18

94.75% (0.87)

92.31% (0.90)

96.12% (0.65)

94.19% (0.72)

19

76.31% (1.25)

74.23% (1.14)

82.42% (1.03)

81.16% (1.24)

20

94.12% (1.19)

95.43% (1.21)

100.00% (0.00)

100.00% (0.00)

21

54.24% (0.75)

52.13% (0.46)

55.09% (0.43)

54.92% (0.65)

22

64.18% (0.94)

60.65% (0.98)

64.87% (0.89)

64.25% (0.71)

23

83.24% (0.76)

81.56% (0.79)

77.23% (0.97)

76.17% (0.88)

24

80.86% (1.01)

80.21% (0.89)

84.72% (0.89)

81.21% (0.77)

Average

73.22% (1.89)

72.15% (1.77)

74.83% (1.43)

73.71% (1.24)

  1. AUCs for EBD and FI discretization methods are obtained from the application of C4.5 and NB classifiers to the discretized variables. The mean and the standard error of the mean (SEM) for the AUC for each dataset is obtained by 10 × 10 cross-validation. For each dataset, the higher AUC is shown in bold font and equal AUCs are underlined.