mlrl.boosting package

Submodules

mlrl.boosting.boosting_learners module

Author: Michael Rapp (mrapp@ke.tu-darmstadt.de)

Provides scikit-learn implementations of boosting algorithms.

class mlrl.boosting.boosting_learners.Boomer(random_state: int = 1, feature_format: str = 'auto', label_format: str = 'auto', prediction_format: str = 'auto', max_rules: int = 1000, default_rule: str = 'true', time_limit: int = 0, early_stopping: Optional[str] = None, head_type: str = 'auto', loss: str = 'logistic-label-wise', predictor: str = 'auto', label_sampling: Optional[str] = None, instance_sampling: Optional[str] = None, recalculate_predictions: str = 'true', feature_sampling: str = 'without-replacement', holdout: Optional[str] = None, feature_binning: Optional[str] = None, label_binning: str = 'auto', pruning: Optional[str] = None, shrinkage: float = 0.3, l2_regularization_weight: float = 1.0, min_coverage: int = 1, max_conditions: int = 0, max_head_refinements: int = 1, parallel_rule_refinement: str = 'auto', parallel_statistic_update: str = 'auto', parallel_prediction: str = 'true')

Bases: mlrl.common.rule_learners.MLRuleLearner, sklearn.base.ClassifierMixin

A scikit-multilearn implementation of “BOOMER”, an algorithm for learning gradient boosted multi-label classification rules.

get_name() → str

Returns a human-readable name that allows to identify the configuration used by the classifier or ranker.

Returns

The name of the classifier or ranker

Module contents