Model Drift

This occurs when the performance of a deployed machine learning model degrades over time because the real-world data it encounters changes or 'drifts' from the data it was originally trained on. This necessitates retraining the model with updated data to maintain accuracy.