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.
Anomaly detection should optimize business value, not just technical accuracy.