dataeval.detectors.drift

Drift detectors identify if the statistical properties of the data has changed.

Submodules

updates

Update strategies inform how the drift detector classes update the reference data when monitoring.

Classes

DriftCVM

Drift detector employing the Cramér-von Mises (CVM) Drift Detection test.

DriftKS

Drift detector employing the Kolmogorov-Smirnov (KS) distribution test.

DriftMMD

Maximum Mean Discrepancy (MMD) Drift Detection algorithm using a permutation test.

DriftUncertainty

Test for a change in the number of instances falling into regions on which the model is uncertain.

Output Classes

DriftMMDOutput

Output class for DriftMMD drift detector.

DriftOutput

Output class for DriftCVM, DriftKS, and DriftUncertainty drift detectors.

Functions

preprocess_drift(x, model[, device, ...])

Prediction function used for preprocessing step of drift detector.