Table 2. Results after applying stacking ensemble.
| ML | Accuracy (%) | Precision (%) | Recall (%) | F1-score (%) | AUC (%) | p-value (vs Stacking) | 95% CI of F1 difference |
|---|---|---|---|---|---|---|---|
| DT | 99.93 | 89.89 | 81.63 | 85.56 | 90.80 | 0.2027 | [−0.0161 to 0.0550] |
| RF | 99.96 | 97.40 | 76.53 | 85.71 | 97.25 | 0.0199 | [−0.0450 to −0.0067] |
| SVM | 99.94 | 97.02 | 66.33 | 78.79 | 95.13 | 0.0225 | [0.0083–0.0635] |
| XGBoost | 99.95 | 95.00 | 77.55 | 85.39 | 97.83 | 0.0628 | [−0.0261 to 0.0011] |
| CatBoost | 99.96 | 97.44 | 77.55 | 86.36 | 98.37 | 0.0315 | [−0.0423 to −0.0033] |
| LR | 99.92 | 88.06 | 60.20 | 71.52 | 97.01 | 0.0047 | [0.0593–0.1727] |
| Stacking hybrid | 99.96 | 98.73 | 79.59 | 88.14 | 89.80 | ---- | ---- |
| Voting | 99.95 | 97.37 | 75.51 | 85.06 | 97.45 | ---- | ---- |
| Subspace Random | 99.95 | 95.95 | 72.45 | 82.56 | 86.22 | ---- | ---- |
| Bagging | 99.95 | 97.30 | 73.47 | 83.72 | 86.73 | ---- | ---- |