Table 3. Estimation of the best classifier for the first level in the two-level system with supervised classifiers: Accuracy and FPR.
| 1R | DT | 1-NN | 5-NN | NB | RBF | SVM | LR | AdaB | Bag | RF | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| K3b | Accuracy (%) | 66.9 | 71.9 | 79.6 | 84.0* | 78.3 | 80.2 | 81.8 | 73.7 | 79.7 | 78.7 | 78.4 |
| FPR (%) | 34.0 | 28.5 | 22.5 | 12.8 | 26.2 | 19.5 | 17.5 | 25.7 | 21.3 | 20.3 | 24.8 | |
| L1b | Accuracy (%) | 71.4 | 75.5 | 79.9 | 82.5 | 77.1 | 79.9 | 81.1 | 80.5 | 81.3 | 80.4 | 83.1* |
| FPR (%) | 30.8 | 24.3 | 18.8 | 17.0 | 38.8 | 31.5 | 25.5 | 18.8 | 19.5 | 22.3 | 22.8 | |
| K6b | Accuracy (%) | 76.6 | 75.5 | 76.9 | 80.0 | 79.4 | 81.5 | 82.1 | 76.5 | 81.0 | 82.4* | 81.8 |
| FPR (%) | 20.3 | 24.8 | 25.5 | 21.0 | 22.5 | 15.3 | 18.3 | 24.5 | 18.0 | 16.8 | 20.3 | |
| Aver. | Accuracy (%) | 71.0 | 74.0 | 78.9 | 82.4* | 78.3 | 80.5 | 81.7 | 76.4 | 80.5 | 80.2 | 80.7 |
| FPR (%) | 29.1 | 26.2 | 22.3 | 16.4 | 28.7 | 21.7 | 20.0 | 23.4 | 19.9 | 19.9 | 22.9 | |
The values are the average for the 5-runs using 10-fold CV, representing * the highest accuracy for each subject.