Business value of anomaly detection

3. Applications: Data Quality Checking

Outlier detection is essential for detecting errors (e.g., typos, sensor failures, strange business events).
But: Catching every outlier is not always worth it — it needs to be cost-effective.
Business Cost:
- False positives → wasted time investigating non-issues.
- False negatives → missing important problems.

Thus, anomaly detection should optimize business value, not just technical accuracy.