Skip to main content

Table 5 Gradient Boost parameters

From: Forecasting the carsharing service demand using uni and multivariable models

Model Parameter List
XGBoost { ‘colsample_bytree’: 1, ‘eval_metric’: ‘rmse’, ‘learning_rate’: 0.005, ‘max_depth’: 10, ‘min_child_weight’: 8, ‘n_estimators’: 2000, ‘objective’: ‘reg:squarederror’, ‘subsample’: 0.5 }
CatBoost { ‘iterations’: 15000, ‘learning_rate’: 0.001, ‘max_depth’: 3, ‘num_leaves’: 31, ‘loss_function’: ‘MAE’, ‘eval_metric’: ‘MAE’}
LightGBM { ‘bagging_fraction’: 0.7, ‘bagging_freq’: 10, ‘feature_fraction’: 0.9, ‘learning_rate’: 0.001, ‘max_depth’: 8, ‘n_estimators’: 40, ‘num_iterations’: 15000, ‘num_leaves’: 32 }