Data Drift

When the statistical properties of real-world data change over time, causing a deployed model's predictions to degrade. Drift can stem from changing user behavior, seasonal patterns, or external events. Monitoring for drift is essential to maintain reliability.