Research Article
JPPRED: Prediction of Types of J-Proteins from Imbalanced Data Using an Ensemble Learning Method
Table 3
Performance of JPPRED with or without feature selection by 10-fold cross validation.
| Subfamily | Measure | No feature selection | Feature selection |
| Type I J-protein | Sn | 0.905 | 0.921 | Sp | 0.849 | 0.859 |
| Type II J-protein | Sn | 0.745 | 0.782 | Sp | 0.857 | 0.866 |
| Type III J-protein | Sn | 0.851 | 0.861 | Sp | 0.855 | 0.877 |
| Type IV J-protein | Sn | 1 | 1 | Sp | 0.849 | 0.860 |
| Dimension | 256 | 224 |
| AvgSn | 0.875 | 0.891 |
| Acc | 0.852 | 0.862 |
|
|