Research Article
A Hybrid Feature Selection Method Based on Rough Conditional Mutual Information and Naive Bayesian Classifier
Table 3
Number and accuracy of selected feature with different algorithms tested by CART.
| Number | Unselect | CFS | Consistency | mRMR | RCMI | Accuracy | Accuracy | Feature number | Accuracy | Feature number | Accuracy | Feature number | Accuracy | Feature number |
| 1 | 72.35% | 72.57% | 22 | 71.90% | 24 | 73.45% | 21 | 75.00% | 16 | 2 | 79.35% | 81.94% | 8 | 80.00% | 13 | 83.23% | 7 | 85.16% | 8 | 3 | 89.74% | 90.31% | 13 | 89.46% | 7 | 86.89% | 7 | 91.74% | 6 | 4 | 96.15% | 96.10% | 8 | 95.28% | 9 | 95.76% | 5 | 95.54% | 3 | 5 | 74.52% | 74.04% | 19 | 77.88% | 14 | 76.44% | 10 | 77.40% | 15 | 6 | 92.53% | 91.65% | 24 | 85.94% | 13 | 92.24% | 21 | 91.36% | 24 | 7 | 95.63% | 95.63% | 5 | 95.63% | 12 | 95.63% | 1 | 96.09% | 2 | 8 | 93.50% | 94.55% | 11 | 94.73% | 8 | 95.43% | 9 | 94.90% | 4 | 9 | 94.94% | 94.94% | 10 | 96.07% | 5 | 93.26% | 8 | 94.94% | 5 | 10 | 92.08% | 93.07% | 10 | 92.08% | 5 | 92.08% | 4 | 93.07% | 10 |
| Average | 88.08% | 88.48% | 13 | 87.90% | 11 | 88.44% | 9.3 | 89.52% | 9.3 |
|
|