Model Drift

The degradation of a model's predictive performance over time as real-world data distributions shift away from training data.