Table 2.
Evaluation of performance per class in feature subset of BE in four algorithms.
| CRC | HIV1 | JIA | ME/CFS | MS | Stroke | Average | |
|---|---|---|---|---|---|---|---|
| Accuracy | |||||||
| LogitBoost | 96.84 ± 0.43 | 99.71 ± 0.14 | 98.52 ± 0.22 | 96.93 ± 0.46 | 98.28 ± 0.29 | 98.32 ± 0.46 | 98.1 ± 0.33 |
| LMT | 95.93 ± 0.3 | 98.66 ± 0.22 | 98.95 ± 0.22 | 96.26 ± 0.57 | 98.18 ± 0.44 | 98.8 ± 0.22 | 97.8 ± 0.33 |
| SVM | 95.59 ± 0.5 | 98.85 ± 0.25 | 98.28 ± 0.38 | 96.46 ± 0.08 | 98.08 ± 0.22 | 98.75 ± 0.22 | 97.67 ± 0.28 |
| KNN | 90.28 ± 0.3 | 97.27 ± 0.43 | 97.27 ± 0 | 94.73 ± 0.36 | 96.41 ± 0.14 | 96.55 ± 0.5 | 95.42 ± 0.29 |
| FPR | |||||||
| CRC | HIV1 | JIA | ME/CFS | MS | Stroke | Average | |
| LogitBoost | 3.7 ± 0.83 | 0.26 ± 0.11 | 0.85 ± 0.09 | 1.18 ± 0.24 | 0.4 ± 0.09 | 1.14 ± 0.28 | 1.26 ± 0.27 |
| LMT | 3.93 ± 0.4 | 0.85 ± 0.3 | 0.6 ± 0 | 1.7 ± 0.41 | 0.7 ± 0.43 | 0.9 ± 0.18 | 1.45 ± 0.29 |
| SVM | 4.77 ± 0.48 | 0.59 ± 0.2 | 0.8 ± 0.09 | 1.59 ± 0.09 | 0.9 ± 0.15 | 0.6 ± 0.1 | 1.54 ± 0.19 |
| KNN | 12.93 ± 0.23 | 1.83 ± 0.49 | 1.35 ± 0.15 | 0.87 ± 0.32 | 0.4 ± 0.17 | 2.34 ± 0.18 | 3.29 ± 0.26 |
| FNR | |||||||
| CRC | HIV1 | JIA | ME/CFS | MS | Stroke | Average | |
| LogitBoost | 2.28 ± 0.38 | 0.36 ± 0.31 | 16.09 ± 3.98 | 28.47 ± 9.62 | 32.18 ± 7.18 | 3.78 ± 1.48 | 13.86 ± 3.82 |
| LMT | 4.31 ± 0.22 | 2.69 ± 0 | 11.49 ± 5.27 | 31.25 ± 3.61 | 27.59 ± 3.45 | 2.36 ± 1.64 | 13.28 ± 2.37 |
| SVM | 3.8 ± 0.66 | 2.69 ± 1.08 | 22.99 ± 7.18 | 29.86 ± 1.2 | 25.29 ± 1.99 | 3.78 ± 0.82 | 14.74 ± 2.16 |
| KNN | 4.44 ± 0.44 | 5.2 ± 0.31 | 34.48 ± 3.45 | 64.58 ± 2.08 | 77.01 ± 5.27 | 7.8 ± 2.56 | 32.25 ± 2.35 |
The model was validated by 10-fold cross-validation and repeated three times. Values represent the mean of accuracy ± variance.