LGBtrainer helps you find hyper params for LGBM and simplifies the process of training the model and finding hyperparams.
pip install LGBtrainer
- Parameters:-
- train = it should be your train dataset(which is fit for training purpose)
- test = it should be your test dataset(which is fit for testing purpose)
- y_train = it should be your target column or values(same rows as train)
- cv = the number of splits or folds(it is used for both finding hyperparams + training the model)
- num_rounds = number of training rounds(it is used for both finding hyperparams + training the model)
- metric = only 'auc' and 'rmse' can be used(For now only these two are supported)
- objective = 'binary' or 'regression' or any other can be provided
- max_eval = number of evaluations performed for finding params(note:- larger number might take more time depending on size of dataset)
- Example:-
-from LGBtrainer import Model
-model = Model(train, test, y_train, metric='auc', objective='binary', max_eval=3, cv=5)
-params = model.get_params()
-predictions = model.lgb_model(params)