Table 7.
Phase-II (nCOVID vs. Pneumonia) classification performance of different supervised models and majority voting algorithm using the validation set. (Note: best performance is highlighted with bold letters).
| Classification algorithms | Accuracy (%) | Specificity (%) | Precision (%) | Recall (%) | F1-Measure (%) | AUC | MCC |
|---|---|---|---|---|---|---|---|
| KNN | 72.093 | 76.744 | 74.359 | 67.442 | 70.732 | 0.721 | 0.444 |
| ANN | 73.256 | 53.488 | 66.667 | 93.023 | 77.670 | 0.733 | 0.506 |
| DT | 79.070 | 82.558 | 81.250 | 75.581 | 78.313 | 0.791 | 0.583 |
| NB | 80.814 | 72.093 | 76.238 | 89.535 | 82.353 | 0.808 | 0.626 |
| SVM (Linear Kernel) | 81.977 | 83.721 | 83.133 | 80.233 | 81.657 | 0.820 | 0.640 |
| SVM (Poly Kernel) | 86.047 | 79.070 | 81.633 | 93.023 | 86.957 | 0.860 | 0.728 |
| SVM (RBF Kernel) | 86.628 | 83.721 | 84.615 | 89.535 | 87.006 | 0.866 | 0.734 |
| Majority voting | 91.329 | 86.207 | 87.368 | 96.512 | 91.713 | 0.914 | 0.831 |