BoostCB
- class lightautoml.ml_algo.boost_cb.BoostCB(default_params=None, freeze_defaults=True, timer=None, optimization_search_space={})[source]
Bases:
TabularMLAlgo
,ImportanceEstimator
Gradient boosting on decision trees from catboost library.
All available parameters listed in CatBoost documentation:
freeze_defaults:
True
: params may be rewritten depending on dataset.False
: params may be changed only manually or with tuning.
timer
:Timer
instance orNone
.- init_params_on_input(train_valid_iterator)[source]
Get model parameters depending on input dataset parameters.
- Parameters:
train_valid_iterator (
TrainValidIterator
) – Classic cv-iterator.- Return type:
- Returns:
Parameters of model.
- fit_predict_single_fold(train, valid)[source]
Implements training and prediction on single fold.
- Parameters:
train (
Union
[NumpyDataset
,CSRSparseDataset
,PandasDataset
]) – Train Dataset.valid (
Union
[NumpyDataset
,CSRSparseDataset
,PandasDataset
]) – Validation Dataset.
- Return type:
- Returns:
Tuple (model, predicted_values).
- predict_single_fold(model, dataset)[source]
Predict of target values for dataset.
- Parameters:
model (
CatBoost
) – CatBoost object.dataset (
Union
[NumpyDataset
,CSRSparseDataset
,PandasDataset
]) – Test dataset.
- Return type:
- Returns:
Predicted target values.
- get_features_score()[source]
Computes feature importance.
Computes as mean values of feature importance, provided by CatBoost (PredictionValuesChange), per all models.
- Return type:
- Returns:
Series with feature importances.
- fit(train_valid)[source]
Just to be compatible with
ImportanceEstimator
.- Parameters:
train_valid (
TrainValidIterator
) – Classic cv-iterator.