Table 4 Holm post-hoc test results based on the MAUCs
From: Boosting methods for multi-class imbalanced data classification: an experimental review
Control method (CatBoost) | ||||
|---|---|---|---|---|
Algorithm | Z | p-value | Holm | Hypothesis (α = 0.05) |
AdaBoost.MH | 3.3387 | 0.0008 | 0.0038 | Rejected |
SAMME | 2.7277 | 0.0045 | 0.0055 | Rejected |
LogitBoost | 1.3311 | 0.1831 | 0.0008 | Not rejected |
GradientBoost | 2.2258 | 0.026 | 0.01 | Not rejected |
XGBoost | 1.1783 | 0.2386 | 0.025 | Not rejected |
MEBoost | 2.9022 | 0.0037 | 0.0041 | Rejected |
SMOTEBoost | 1.0038 | 0.3154 | 0.05 | Not rejected |
RUSBoost | 2.8804 | 0.0039 | 0.0045 | Rejected |
LightGBM | 2.6186 | 0.0088 | 0.0062 | Not rejected |
AdaC1 | 3.3387 | 0.0008 | 0.0038 | Rejected |
AdaC2 | 2.8368 | 0.0045 | 0.005 | Rejected |
AdaC3 | 2.4876 | 0.0128 | 0.0083 | Not rejected |
AdaCost | 2.1167 | 0.0342 | 0.0125 | Not rejected |