cinnamon.drift.abstract_drift_explainer.AbstractDriftExplainer

class cinnamon.drift.abstract_drift_explainer.AbstractDriftExplainer

__init__()

get_feature_drift(feature: Union[int, str]) → AbstractDriftMetrics

Compute drift measures for a given feature in X.

For numerical feature: - Difference of means - Wasserstein distance - Result of Kolmogorov 2 sample test

For categorial feature (not supported currently. No categorical feature allowed): - Wasserstein distance - Result of Chi2 two sample test

See the documentation in README for explanations about how it is computed, especially the slide presentation.

Parameters

featureUnion[int, str]

Either the column index of the name of the feature.

Returns

feature_drift: Union[DriftMetricsCat, DriftMetricsNum]

Drift measures of the input feature.

get_feature_drifts() → List[AbstractDriftMetrics]

Compute drift measures for all features in X.

For numerical features: - Difference of means - Wasserstein distance - Result of Kolmogorov 2 sample test

For categorial features (not supported currently. No categorical feature allowed): - Wasserstein distance - Result of Chi2 two sample test

See the documentation in README for explanations about how it is computed, especially the slide presentation.

Returns

feature_drift: list of Union[DriftMetricsCat, DriftMetricsNum].

Drift measures for each input feature in X.

get_target_drift() → AbstractDriftMetrics

Compute drift measures for the labels y.

For classification : - Wasserstein distance - Result of Chi2 2 sample test

For regression: - Difference of means - Wasserstein distance - Result of Kolmogorov 2 sample test

See the documentation in README for explanations about how it is computed, especially the slide presentation.

Returns

target_driftUnion[DriftMetricsCat, DriftMetricsNum]

Drift measures for the labels y.