dataeval.shift.DriftUnivariate.Config¶

class dataeval.shift.DriftUnivariate.Config¶

Configuration for DriftUnivariate detector.

method¶

Statistical test method to use.

Type:¶

{“ks”, “cvm”, “mwu”, “anderson”, “bws”}, default “ks”

p_val¶

Significance threshold for drift detection.

Type:¶

float, default 0.05

correction¶

Multiple testing correction method.

Type:¶

{“bonferroni”, “fdr”}, default “bonferroni”

alternative¶

Alternative hypothesis for the statistical test.

Type:¶

{“two-sided”, “less”, “greater”}, default “two-sided”

n_features¶

Number of features to analyze.

Type:¶

int or None, default None

update_strategy¶

Strategy for updating reference data over time.

Type:¶

UpdateStrategy or None, default None

extractor¶

Feature extractor for transforming input data before drift detection.

Type:¶

FeatureExtractor or None, default None