Skip to content

Y-oHr-N/OptGBM

Repository files navigation

OptGBM

Python package codecov PyPI PyPI - License Binder

OptGBM (= Optuna + LightGBM) provides a scikit-learn compatible estimator that tunes hyperparameters in LightGBM with Optuna.

Examples

import optgbm as lgb
from sklearn.datasets import load_boston

reg = lgb.LGBMRegressor(random_state=0)
X, y = load_boston(return_X_y=True)

reg.fit(X, y)

y_pred = reg.predict(X, y)

By default, the following hyperparameters will be searched.

  • bagging_fraction
  • bagging_freq
  • feature_fractrion
  • lambda_l1
  • lambda_l2
  • max_depth
  • min_data_in_leaf
  • num_leaves

Installation

pip install optgbm

Testing

tox