lightgbm_quickstart.py

# Code snippet to train lightgbm models quickly, using the number of iterations based on CV

import lightgbm as lgb
lgb_dftrain = lgb.Dataset(features_train, target_train)

params = {'boosting_type': 'gbdt',
          'objective': 'binary',
          'enable_bundle': True,
          'max_conflict_rate': 0,
          'max_depth': 20,
          'min_split_gain': 1,
          'lambda_l2': 1,
          'lambda_l1': 1,
          'subsample': 0.5,
          'subsample_freq': 1,
          'feature_fration': 0.5,
          'metric': ['binary_logloss', 'binary_error'],
          'learning_rate': 0.03}

cv_hist = lgb.cv(params,
                 lgb_dftrain,
                 verbose_eval=10,
                 num_boost_round=20000,
                 nfold=3,
                 stratified=True,
                 early_stopping_rounds=50,
                 show_stdv=False)
                   
best_iter = np.argmin(cv_hist['binary_error-mean'])
gbm_model = lgb.train(params,
                      lgb_dftrain,
                      verbose_eval=50000,
                      num_boost_round=best_iter,
                      valid_sets=lgb_dftrain)