Table 3.
Performance of the proposed Mr2DNM for five datasets.
| Dataset | NF | # | Reduction (%) | Optimal feature sequence | Accuracy (%) | Time (×103 s) | AUC |
|---|---|---|---|---|---|---|---|
| WBCD | 9 | 7 | 22.22 | F2, F6, F1, F7, F5, F3, F8 | 96.80 | 54.4 | 0.9942 |
| BUPA | 7 | 5 | 28.57 | F5, F6, F1, F4, F3 | 72.66 | 7.1 | 0.7458 |
| IONO | 34 | 8 | 76.47 | F5, F1, F8, F4, F3, F28, F7, F14 | 90.73 | 24.6 | 0.9227 |
| PIMA | 8 | 7 | 12.5 | F2, F5, F8, F6, F4, F1, F3 | 76.80 | 33.2 | 0.8198 |
| VOTE | 16 | 6 | 62.5 | F4, F5, F12, F3, F14, F8 | 96.57 | 10.2 | 0.9779 |