dataeval.detectors.drift.DriftMMDOutput¶

class dataeval.detectors.drift.DriftMMDOutput¶

Output class for DriftMMD (Maximum Mean Discrepancy) drift detector.

Extends DriftBaseOutput with MMD-specific distance threshold information. Used by MMD-based drift detectors that compare kernel embeddings between reference and test distributions.

drifted¶

Whether drift was detected based on MMD permutation test.

Type:¶

bool

threshold¶

P-value threshold used for significance of the permutation test.

Type:¶

float

p_val¶

P-value obtained from the MMD permutation test, between 0 and 1.

Type:¶

float

distance¶

Squared Maximum Mean Discrepancy between reference and test set. Always >= 0, with higher values indicating greater distributional difference.

Type:¶

float

distance_threshold¶

Squared Maximum Mean Discrepancy threshold above which drift is flagged, always >= 0. Determined from permutation test at specified significance level.

Type:¶

float

Notes

MMD uses kernel methods to compare distributions in reproducing kernel Hilbert spaces, making it effective for high-dimensional data like images.

data()¶

The output data as a dictionary.

Return type:¶

dict[str, Any]

meta()¶

Metadata about the execution of the function or method for the Output class.

Return type:¶

ExecutionMetadata