Table 1.
Jackknife cross-validation and independent testing results after training on the benchmark data set D1 without feature selection.
| Feature(D) | SMOTE (Y/N) | Jackknife cross-validation | Independent testing | ||||||
|---|---|---|---|---|---|---|---|---|---|
| ACC | MCC | Sn | Sp | ACC | MCC | Sn | Sp | ||
| 2-gapDC(400) | N | 74.5% | 0.326 | 94.7% | 28.6% | 79.7% | 0.318 | 90.2% | 38.5% |
| SAAC(60) | N | 69.3% | 0.073 | 97.9% | 4.8% | 78.1% | −0.07 | 98.0% | 0.0% |
| 2-gapDC+SAAC(460) | N | 75.2% | 0.351 | 94.7% | 31.0% | 79.7% | 0.237 | 94.1% | 23.1% |
| 2-gapDC(400) | Y | 86.3% | 0.743 | 96.8% | 75.8% | 82.8% | 0.351 | 98.0% | 23.1% |
| SAAC(60) | Y | 87.9% | 0.763 | 93.7% | 82.1% | 81.2% | 0.388 | 90.2% | 46.2% |
| SAAC+2-gapDC(460) | Y | 90.5% | 0.817 | 96.8% | 84.2% | 81.2% | 0.287 | 96.1% | 23.1% |