Performance Drift

Performance Drift is in a sense the “most straightforward” kind of drift analysis. It analyses whether the actual performance of the model changes.

However, having ground truth / labels is naturally required for Performance Drift, which is not always possible. See Automating model evaluations and drift analysis for a discussion on this.

When Ground truth is not available, Input Data Drift and Prediction Drift can provide insights into whether you have a concept drift.