I have trained a lgbm model in sklearn API format just like this:
cb_classifier = LGBMClassifier(**params)
cb_classifier.fit(X_train[features],
y_train,
eval_set = (X_validation[features], y_validation),
categorical_feature = cat_dims
)
After the model was generated, I obtained the predictions for X_test with that model, but now I would like to obtain the same model for predicting the results in the native LGBM format, like:
native_lgbm_model.predict(X_test)
I tried saving the cb_classifier with booster like this:
cb_classifier.booster_.save_model("path/booster_lgbm.txt")
native_lgbm_model = lightgbm.Booster(model_file='path/booster_lgbm.txt')
But when trying to predict the results they were different from the results obtained when predicting with the sklearn API.
Is there any way to convert cb_classifier into native LGBM model?
