Table 4 Comparison of Leave-one-out cross-validation accuracy obtained for protein structural classification problem on the two datasets by our approach and existing approaches.
From: A Time-Series-Based Feature Extraction Approach for Prediction of Protein Structural Class
Dataset | Method | Prediction accuracy for each structural class (%) | Overall accuracy (%) | |||
|---|---|---|---|---|---|---|
| Â | Â |
|
|
|
| Â |
Dataset1 | Our approach | 86.96 | 88.52 | 88.89 | 66.15 | 82.97 |
| Â | Component coupled [6] | 84.3 | 82.0 | 81.5 | 67.7 | 79.1 |
| Â | Neural network [7] | 68.6 | 85.2 | 86.4 | 56.9 | 74.7 |
| Â | SVM [10] | 74.3 | 82.0 | 87.7 | 72.3 | 79.4 |
| Â | Rough sets [13] | 77.1 | 77.0 | 93.8 | 66.2 | 79.4 |
Dataset2 | Our approach | 93.33 | 94.44 | 97.04 | 90.7 | 93.94 |
| Â | Component coupled [6] | 93.5 | 88.9 | 90.4 | 84.5 | 89.2 |
| Â | Neural network [7] | 86.0 | 96.0 | 88.2 | 86.0 | 89.2 |
| Â | SVM [10] | 88.8 | 95.2 | 96.3 | 91.5 | 93.2 |
| Â | Rough sets [13] | 87.9 | 91.3 | 97.1 | 86.0 | 90.8 |
