Table 2.
The best evaluation scores from jackknife cross-validation and independent testing of different models with various feature types and feature numbers.
| Classifier | Features(D) | Jackknife cross-validation | Independent testing | ||||||
|---|---|---|---|---|---|---|---|---|---|
| ACC | MCC | Sn | Sp | ACC | MCC | Sn | Sp | ||
| rfGPT_1 | 2-gapDC(59) | 93.2% | 0.864 | 94.7% | 91.6% | 84.4% | 0.466 | 94.1% | 46.2% |
| rfGPT_2 | 2-gapDC(55) | 91.1% | 0.823 | 94.7% | 87.4% | 89.1% | 0.631 | 98.0% | 53.8% |
| rfGPT_3 | 2-gapDC+SAAC(43) | 93.7% | 0.874 | 93.7% | 93.7% | 82.8% | 0.484 | 88.2% | 61.5% |
| rfGPT_4 | 2-gapDC+SAAC(93) | 90.5% | 0.811 | 92.6% | 88.4% | 90.6% | 0.696 | 96.1% | 69.2% |
| rfGPT_5 | 2-gapDC+SAAC(94) | 93.2% | 0.864 | 93.7% | 92.7% | 84.4% | 0.546 | 88.2% | 69.2% |
| rfGPT_6 | 2-gapDC+SAAC(66) | 90.0% | 0.800 | 89.5% | 90.5% | 89.1% | 0.695 | 90.2% | 84.6% |