Table 5.
Classifier accuracy for features selected by the ReliefF algorithm.
| Classifier Accuracy (%) | |||||
|---|---|---|---|---|---|
| Feature Type * |
RBF SVM |
LDA | Naive Bayes |
kNN | D3 |
| Trbp (5) | 66.15 | 79.60 | 80.00 | 72.25 | 55.85 |
| Arbp (2) | 81.20 | 78.70 | 75.95 | 90.00 | 85.30 |
| Brbp (2) | 90.00 | 80.85 | 79.90 | 90.00 | 90.00 |
| Grbp (1) | 75.00 | 69.00 | 75.00 | 70.00 | 63.25 |
| APV (1) | 62.85 | 37.50 | 62.80 | 71.20 | 66.40 |
| SASI (2) | 63.35 | 72.35 | 70.70 | 55.00 | 72.55 |
| HFD (2) | 77.45 | 53.65 | 66.20 | 81.00 | 85.70 |
| LZC (2) | 81.25 | 78.25 | 72.75 | 81.95 | 69.55 |
| DFA (3) | 78.80 | 56.40 | 72.55 | 86.00 | 72.90 |
(*) Number of features used (see Table 3).