quantlab.model.predefined.lgbm#
Qlib’s LightGBM benchmark model.
LGBMRegressor reproduces Qlib’s LGBModel (qlib/contrib/model/gbdt.py)
with its Alpha158 benchmark parameters (workflow_config_lightgbm_Alpha158.yaml):
lgb.train on the training rows, the training and validation sets both
evaluated every round, and LightGBM’s native early stopping on the validation
set. LightGBM ships in the gbdt extra.
The module is named lgbm.py rather than lightgbm.py so it does not
shadow the lightgbm package.
Classes
Qlib's LightGBM on the rows of the cross-sectional rank target. |