Table 8.
The accuracy of binary-level classification (%) of the DT, RF, and QDA classifiers that obtained the highest accuracy in phase I compared to after using the three search strategies of the hybrid FS approach (phase II of the proposed diagnostic tool).
Classifier | Before FS | Forward | Backward | Bidirectional |
---|---|---|---|---|
DT | 97.62 | 98.2 | 98.2 | 98.2 |
RF | 97.78 | 98.0 | 98.0 | 98.0 |
QDA | 97.6 | 97.8 | 97.6 | 97.8 |