Table 13 Experimental results on Hungarian 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 (%) | 80.43 ± 5.37 | 82.07 ± 7.12 | 78.91 ± 5.61 | 80.40 ± 6.86 | 75.43 ± 8.64 | 89.47 ± 3.06 |
Precision (%) | 75.52 ± 5.96 | 77.93 ± 8.48 | 74.48 ± 6.54 | 75.86 ± 7.09 | 66.55 ± 14.99 | 89.31 ± 4.44 |
Recall (%) | 60.19 ± 16.84 | 62.08 ± 15.89 | 53.38 ± 17.93 | 59.42 ± 19.49 | 61.36 ± 19.71 | 82.39 ± 5.73 |
G-mean | 71.04 ± 8.34 | 73.72 ± 10.15 | 67.55 ± 9.21 | 70.97 ± 10.16 | 59.97 ± 24.07 | 82.95 ± 4.63 |
MC (%) | 87.93  ± 34.95 | 82.76 ± 37.63 | 100.00 ± 36.09 | 90.34 ± 44.33 | 94.14 ± 30.89 | 38.28 ± 12.10 |
Specificity (%) | 86.34 ± 9.83 | 88.99 ± 7.79 | 89.10 ± 11.61 | 88.13 ± 9.92 | 70.92 ± 25.22 | 92.02 ± 5.76 |
AUC (%) | 74.07 ± 9.16 | 76.31 ± 10.87 | 71.96 ± 10.98 | 74.59 ± 9.55 | 69.07 ± 9.98 | 88.38 ± 5.36 |