MOAFeatureImportanceClassifier#

class capymoa.feature_selection.MOAFeatureImportanceClassifier[source]#

Bases: FeatureImportanceClassifier, MOAClassifier

MOA-backed feature-importance classifier.

Instantiate this class when the underlying learner is a MOA classifier. Pure Python implementations should subclass FeatureImportanceClassifier instead.

This wrapper is currently restricted to:

HoeffdingTree learners and subclasses, which are wrapped with FeatureImportanceHoeffdingTree
MOA ensembles built from HoeffdingTree learners, which are wrapped with FeatureImportanceHoeffdingTreeEnsemble

If MOA adds other feature-importance learner families in the future, they will not automatically work through this class. In that case this wrapper should be refactored to support those learners explicitly.

Accepted base_learner inputs: - CapyMOA MOAClassifier instance - CapyMOA MOAClassifier class - raw MOA learner instance - raw MOA learner class

__init__( schema: Schema | None = None, base_learner: Any = None, random_seed: int = 1, window_size: int | None = None, )[source]#

cli_help()[source]#

get_feature_importances( normalize: bool = True, ) → list[float][source]#: Return the current feature importance scores.

get_top_k_features( k: int, normalize: bool = True, ) → list[int][source]#

get_windowed_feature_importances() → list[dict[str, Any]] | None[source]#

predict( instance: Instance, ) → int | None[source]#

Predict the label of an instance.

The base implementation calls predict_proba() and returns the label with the highest probability.

Parameters:: instance – The instance to predict the label for.
Returns:: The predicted label or None if the classifier is unable to make a prediction.

predict_proba( instance, ) → ndarray[tuple[Any, ...], dtype[float64]] | None[source]#

Return probability estimates for each label.

Parameters:: instance – The instance to estimate the probabilities for.
Returns:: An array of probabilities for each label or None if the classifier is unable to make a prediction.

train(instance: Any) → None[source]#

Train the classifier with a labeled instance.

Parameters:: instance – The labeled instance to train the classifier with.

feature_importances_per_window: list[dict[str, Any]] | None#

random_seed: int#

The random seed for reproducibility.

When implementing a classifier ensure random number generators are seeded.

schema: Schema#: The schema representing the instances.