Table 3 Hyperparameters tuning of XGBoost
From: XGBoost-SHAP-based interpretable diagnostic framework for alzheimer’s disease
Range of hyperparameters | Hyperparameters on ADNI dataset | Hyperparameters on NACC dataset |
|---|---|---|
Booster = gbtree | Booster = gbtree | Booster = gbtree |
eta = (0.01: 0.3) | eta = 0.01 | eta = 0.01 |
gamma = (0.5: 1) | gamma = 0.5 | gamma = 0.5 |
max_depth = (5: 8) | max_depth = 5 | max_depth = 5 |
min_child_weight = 1 (default) | min_child_weight = 1 | min_child_weight = 1 |
subsample = 1 (default) | subsample = 1 | subsample = 1 |
colsample_bytree = 1 (default) | colsample_bytree = 1 | colsample_bytree = 1 |
lambda = (1–3) | lambda = 3 | lambda = 3 |
nrounds = (75: 2000) | nrounds = 482 | nrounds = 482 |
scale_pos_weight | scale_pos_weight = (1, 0.62, 3.4) | scale_pos_weight = (1, 2.5, 2) |