Table 7 Comparing each baseline meta-model against AeKNN on the 20-benchmark datasets. Listed are the number of datasets where each meta-model produced better predictions than AeKNN (Wins), worse predictions (Losses), or more accurate predictions than all of the other 3 meta-models (Champion)
Meta-model | Wins | Losses | Champion | ||||||
|---|---|---|---|---|---|---|---|---|---|
Acc | F1-score | AUC | Acc | F1-score | AUC | Acc | F1-score | AUC | |
AeKNN | - | - | - | - | - | - | 16 | 17 | 14 |
KNN | 1 | 2 | Â | 19 | 18 | 17 | 1 | 2 | 3 |
RF | 0 | 0 | Â | 20 | 20 | 20 | 0 | 0 | 0 |
XGB | 3 | 1 | Â | 17 | 19 | 17 | 3 | 1 | 3 |