Table 9 Experimental results on Cleveland 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 (%) | 86.78 ± 6.15 | 86.53 ± 6.75 | 86.50 ± 5.89 | 84.19 ± 7.59 | 79.44 ± 9.05 | 93.83 ± 4.93\(^{*}\) |
Precision (%) | 82.67 ± 7.28 | 83.00 ± 7.45 | 82.00 ± 6.25 | 79.00 ± 8.32 | 72.00 ± 11.88 | 88.67 ± 5.49 |
Recall (%) | 80.26 ± 14.28 | 78.02 ± 16.41 | 81.20 ± 15.12 | 77.86 ± 19.94 | 73.70 ± 14.34 | 89.68 ± 8.78 |
G-mean | 82.24 ± 8.84 | 82.24 ± 9.12 | 81.51 ± 8.03 | 78.77 ± 11.80 | 72.01 ± 11.84 | 90.77 ± 6.71 |
MC (%) | 54.67  ± 33.45 | 59.67 ± 38.12 | 54.00 ± 35.38 | 63.67 ± 42.95 | 78.67 ± 39.79 | 22.00 ± 15.61 |
Specificity (%) | 84.63 ± 7.49 | 87.49 ± 6.38 | 82.47 ± 6.54 | 80.42 ± 7.43 | 71.26 ± 14.11 | 89.31 ± 5.13 |
AUC (%) | 81.53 ± 8.75 | 81.99 ± 9.38 | 80.91 ± 8.14 | 79.99 ± 11.05 | 70.53 ± 12.65 | 89.54 ± 5.54 |