Explain 'Data Drift' in anomaly detection.
Updated May 5, 2026
Short answer
The distribution of input data changes over time[cite: 1].
Deep explanation
If the 'normal' behavior shifts (e.g., higher average sales during holidays), a static model will incorrectly flag these as anomalies[cite: 1].
Real-world example
Changes in consumer behavior after a major global event[cite: 1].
Common mistakes
- Not retraining models periodically[cite: 1].
Follow-up questions
- How to detect it?