Research Article
Application of Improved Three-Dimensional Kernel Approach to Prediction of Protein Structural Class
Table 2
LOOCV success rates by covariant discriminant, neural network, SVM, bagging, and improved 3D kernel approach.
| | Algorithm | Rate of correct prediction for each class | Overall rate of correct prediction | | Type A | Type B | Type C | Type D | Type E |
| | Covariant discriminant | 74.0% | 52.0% | 83.7% | 49% | 45.4% | 76.4% | | Neural network | 75.63% | 30.92% | 88.86% | 50.98% | 30.91% | 77.76% | | SVMs | 77.7% | 28.3% | 92.5% | 52.9% | 35.5% | 80.9% | | Bagging | 79.80% | 48.68% | 93.21% | 49.02% | 60.91% | 84.18% | | 3D kernel | 78.11% | 31.02% | 94.36% | 52.63% | 45.46% | 84.50% |
|
|
