Table 1.
Search range of hyperparameters for developed machine learning models.
| Hyperparameters | XGBoost | LightGBM | CatBoost |
|---|---|---|---|
| reg_alpha | 1 × 10−5, 1 × 10−4, 1 × 10−3, 1 × 10−2, 0.1, 1, 5, 10, 100 | - | |
| reg_lambda | - | ||
| colsample_bytree | 0.5~1.0 (0.1) | - | |
| min_child_weight | 250~350 (10) | - | |
| eta | 0.1~0.3 (0.1) | - | |
| l2_leaf_reg | - | 3~8 (1) | |
| border_count | - | 32~255 (10) | |
| colsample_bylevel | - | 0.3~0.8 (0.1) | |
| bagging_temperature | - | 0~10 | |
| min_data_in_leaf | - | 1, 5, 10, 20, 30 | |
| max_depth | 10~25 (1) | 3~9 (1) | |
| subsample | 0.7~0.9 (0.1) | 0.5~1.0 | |
| learning_rate | 0.01~0.05 (0.005) | ||
| n_estimators | 1000~10,000 (10) | ||
| eval_metric | Root Mean Square Error | ||