Covariate Shift
A type of data drift in which the distribution P(X) of the input features changes while the conditional distribution P(Y|X) remains unchanged. It can be mitigated through resampling or reweighting.
Timeline (last 90 days)
PSI detection is mainly applicable to covariate drift and label drift, since it only focuses on feature distributions and does not involve labels.
Covariate shift and model drift can trigger negative transfer, causing additional data to drag down model performance.
Covariate shift refers to the inconsistency of the feature distribution P(X) between source and target, while model shift involves the shift of the conditional distribution P(Y|X).
All Facts (3)
PSI detection is mainly applicable to covariate drift and label drift, since it only focuses on feature distributions and does not involve labels.
50%UnverifiedCovariate shift and model drift can trigger negative transfer, causing additional data to drag down model performance.
50%UnverifiedCovariate shift refers to the inconsistency of the feature distribution P(X) between source and target, while model shift involves the shift of the conditional distribution P(Y|X).
50%