Table 2.
The performance of the McTWO feature selection algorithm in comparison with four other feature selection algorithms.
| Method | ACC | SENS | SPEC | PRE | F1 | MCC | AUC |
|---|---|---|---|---|---|---|---|
| Boruta | 0.65 | 0.603 | 0.733 | 0.733 | 0.634 | 0.337 | 0.730 |
| mRMR | 0.667 | 0.661 | 0.663 | 0.726 | 0.662 | 0.347 | 0.760 |
| RFE | 0.692 | 0.671 | 0.702 | 0.725 | 0.678 | 0.366 | 0.768 |
| RF | 0.708 | 0.698 | 0.727 | 0.767 | 0.711 | 0.435 | 0.821 |
| Two-step | 0.733 | 0.732 | 0.770 | 0.797 | 0.743 | 0.505 | 0.889 |
mRMR: Minimum redundancy maximum relevance, RFE: Recursive feature elimination, RF: Random forest, ACC: Accuracy, SENS: Sensitivity, SPEC: Specificity, PRE: Precision, F1: F1-score, MCC: Matthew’s correlation coefficient, AUC: Area under the ROC curve.