Some sensor values are physically impossible outliers.
Continue aggregating normal data while investigating the anomalies later.
Which quality approach is most appropriate?
Flag outliers with a rule and retain them in a quarantine table with the original value, reason, and input version.
Detailed explanation
Deletion prevents investigation and replay.
Deletion prevents investigation and replay.
A single low-impact violation may not justify stopping all consumers.
A single low-impact violation may not justify stopping all consumers.
Normal and anomalous rows remain separate while evidence is preserved.
Normal and anomalous rows remain separate while evidence is preserved.
The original measurement and its cause are lost without a documented correction basis.
The original measurement and its cause are lost without a documented correction basis.
Check official documentation
Compare the following points with the official documentation.
AWS Glue Data Quality custom rules、anomaly detection、quarantineの公式説明を確認する。
What to confirm
補正・全停止・削除と、隔離して継続する設計を区別できる。
Key points
- Outlier
- Quarantine
- Quality gate
- Replay
Notes
- Environment: AWS公式DEA-C01試験ガイドと各サービスの公式ドキュメントの確認
- Confirm current AWS service behavior in the official DEA-C01 documentation.
Foundation review
Stop criteria
Distinguish isolated sensor errors from broad input failure using impact and violation rate.
Evidence retention
Store the original value, rule version, reason, and processing time for later review.