Skip to main content

Table 5 Experimental results on Statlog dataset with the best feature subset

From: A hybrid cost-sensitive ensemble for heart disease prediction

Mean ± SD

RF

LR

SVM

ELM

KNN

Proposed ensemble

E (%)

87.53 ± 5.39

87.87 ± 6.82

88.67 ± 5.02

82.81 ± 5.54

76.94 ± 11.33

94.44 ± 3.78

Precision (%)

83.70 ± 6.58

84.07 ± 8.01

84.81 ± 6.40

78.15 ± 6.64

70 ± 15.37

92.59 ± 4.62

Recall (%)

80.64 ± 11.80

82.08 ± 13.07

83.85 ± 10.98

70.65 ± 13.77

62.85 ± 17.51

92.15 ± 7.10

G-mean

83.14 ± 7.54

83.79 ± 8.19

84.41 ± 7.10

76.65 ± 8.05

68.40 ± 15.63

92.56 ± 4.79

MC (%)

51.85  ± 26.07

50 ± 34.67

44.81 ± 23.78

75.19 ± 29.63

96.67 ± 44.56

22.22 ± 19.36

Specificity (%)

86.13 ± 6.17

86 ± 6.58

85.45 ± 7.83

84.29 ± 8.43

75.18 ± 16.55

93.21 ± 5.43

AUC (%)

83.75 ± 8.26

83.92 ± 9.44

85.07 ± 7.72

80.17 ± 6.96

68.42 ± 13.73

92.08 ± 5.51

  1. The average \(+-\) sd on 10-folds CV. The best result is bolded