Table 7.
Experimental results on Statlog dataset with 13 features
| Mean ± SD | RF | LR | SVM | ELM | KNN | Proposed ensemble |
|---|---|---|---|---|---|---|
| E (%) | 71.76 ± 7.44 | 77.16 ± 4.53 | 68.49 ± 6.06 | 77.31 ± 8.11 | 66.70 ± 4.26 | 86.36 ± 5.51 |
| Precision (%) | 65.19 ± 14.96 | 73.70 ± 7.77 | 68.15 ± 10.03 | 61.48 ± 29.28 | 59.26 ± 11.05 | 78.52 ± 7.37 |
| Recall (%) | 86.54 ± 10.48 | 83.13 ± 8.57 | 75.62 ± 6.28 | 82.45 ± 18.42 | 73.57 ± 13.65 | 92.56 ± 8.19 |
| G-mean | 82.18 ± 9.64 | 83.72 ± 14.18 | 76.29 ± 7.45 | 82.60 ± 14.51 | 76.35 ± 18.16 | 90.17 ± 8.08 |
| MC (%) | 75.12 ± 9.10 | 56.30 ± 7.77 | 62.69 ± 25.27 | 41.12 ± 33.75 | 85.19 ± 43.82 | 34.81 ± 24.58 |
| Specificity (%) | 78.05 ± 7.26 | 84.32 ± 8.97 | 76.96 ± 16.40 | 82.81 ± 8.72 | 79.23 ± 17.11 | 87.84 ± 5.73 |
| AUC (%) | 79.35 ± 11.28 | 83.16 ± 9.78 | 83.16 ± 9.82 | 81.27 ± 12.51 | 78.53 ± 6.94 | 87.99 ± 8.39 |
The average sd on 10-folds CV. The best result is bolded.