Caesar AI Atlas

Model Drift

Caesar AI Atlas Definition

Model drift is the degradation or change in model performance over time as real-world data, user behavior, or operating conditions diverge from the data used during training or evaluation. Detecting drift is important for deciding when to retrain, recalibrate, or replace a model.

Also Referenced In

Concept Comparisons

Related Terms