Table 2.
Comparison of different classifiers and the expert algorithm (baseline), measured by their average performance (and standard deviation) in cross-validation.
| Classifiers | Feature Sets | Accuracy | Sensitivity | Specificity | Precision | AUC |
|---|---|---|---|---|---|---|
| Expert Algorithm | - | 0.84 | 0.78 | 1.00 | 1.00 | 0.71 |
| LR | #107 | 0.86 (0.06) | 0.90 (0.09) | 0.84 (0.10) | 0.70 (0.11) | 0.88 (0.07) |
| #33 | 0.91 (0.04) | 0.98 (0.03) | 0.88 (0.06) | 0.77 (0.07) | 0.92 (0.03) | |
| #5 | 0.99 (0.01) | 1.00 (0) | 0.98 (0.01) | 0.95 (0.03) | 0.99 (0.01) | |
| NB | #107 | 0.94 (0.05) | 0.98 (0.03) | 0.93 (0.07) | 0.85 (0.11) | 0.98 (0.02) |
| #33 | 0.91 (0.07) | 1.00 (0) | 0.88 (0.10) | 0.79 (0.15) | 1.00 (0) | |
| #5 | 0.96 (0.03) | 1.00 (0) | 0.94 (0.05) | 0.87 (0.09) | 1.00 (0) | |
| RF | #107 | 0.98 (0.01) | 1.00 (0) | 0.97 (0.02) | 0.94 (0.05) | 1.00 (0) |
| #33 | 0.98 (0.01) | 1.00 (0) | 0.97 (0.02) | 0.94 (0.05) | 1.00 (0) | |
| #5 | 0.98 (0) | 0.98 (0.03) | 0.98 (0.01) | 0.95 (0.03) | 1.00 (0) | |
| kNN | #107 | 0.83 (0.06) | 0.87 (0.05) | 0.81 (0.08) | 0.65 (0.09) | 0.91 (0.01) |
| #33 | 0.94 (0.05) | 0.98 (0.03) | 0.92 (0.08) | 0.84 (0.12) | 0.98 (0.02) | |
| #5 | 0.97 (0.03) | 1.00 (0) | 0.96 (0.04) | 0.90 (0.08) | 0.99 (0.01) | |
| SVM | #107 | 0.96 (0.04) | 0.95 (0.03) | 0.96 (0.04) | 0.91 (0.10) | 0.96 (0.03) |
| #33 | 0.97 (0.02) | 0.97 (0.04) | 0.97 (0.02) | 0.93 (0.06) | 0.97 (0.02) | |
| #5 | 0.98 (0.01) | 0.95 (0.03) | 0.99 (0.01) | 0.98 (0.03) | 0.97 (0.02) | |
| J48 | #107 | 0.98 (0.02) | 1.00 (0) | 0.97 (0.02) | 0.93 (0.05) | 0.98 (0.01) |
| #33 | 0.97 (0.02) | 0.97 (0.04) | 0.97 (0.02) | 0.94 (0.05) | 0.99 (0.01) | |
| #5 | 0.97 (0.03) | 0.95 (0.03) | 0.97 (0.03) | 0.94 (0.07) | 0.98 (0.03) |
The bold values indicate the best models in terms of accuracy, sensitity, specificity, precision and AUC. The significance values are inappropriate for them.