TABLE 5. Performance Metrics Achieved by the Unpruned Models Through Different Ensemble Strategies for the Multiclass Classification Task.
| Method | Acc. | AUC | Sens. | Prec. | F | MCC |
|---|---|---|---|---|---|---|
| Majority Voting | 0.9742 | 0.9807 [0.9686 0.9928 | 0.9742 | 0.9748 | 0.9742 | 0.9537 |
| Averaging | 0.9782 | 0.9969 [0.992 1.0 | 0.9782 | 0.9786 | 0.9782 | 0.9607 |
| Weighted Averaging | 0.9762 | 0.9968 [0.9918 1.0] | 0.9762 | 0.9767 | 0.9762 | 0.9572 |
| Stacking | 0.9663 | 0.9865 [0.9764 0.9966 | 0.9663 | 0.96 | 0.9662 | 0.9402 |
* Bold values stand for the method with a statistically significant better performance than the other ensemble methods.