Model Drift

The gradual degradation of a deployed model's performance over time as the real world changes and diverges from the conditions under which the model was trained. Model drift encompasses both data drift and concept drift, and is a key reason why deployed models require ongoing monitoring.