Table 3.
Predictive performance of classifiers constructed by combining the best single features.
| Features | AC (%) |
SN (%) |
SP (%) |
ST (%) |
MCC | ROC AUC |
|---|---|---|---|---|---|---|
| S3 | 82.53 | 72.19 | 87.15 | 79.67 | 0.5922 | 0.8835 |
| S3, Bu | 83.41 | 68.00 | 90.30 | 79.15 | 0.6019 | 0.8821 |
| S3, Bu, Tt | 82.88 | 61.90 | 92.26 | 77.08 | 0.5822 | 0.8725 |
| S3, Bu, Tt, B | 83.65 | 62.10 | 93.28 | 77.69 | 0.6009 | 0.8768 |
| S3, Bu, Tt, B, Aa | 83.65 | 61.90 | 93.36 | 77.63 | 0.6009 | 0.8743 |
| S3, Bu, Tt, B, Aa, Mc | 83.59 | 61.71 | 93.36 | 77.54 | 0.5993 | 0.8737 |
| All 20 features | 82.88 | 56.00 | 94.89 | 75.45 | 0.5791 | 0.8690 |