Order data can contain missing values, duplicates, and outliers.
Do not publish a dashboard dataset until defined quality criteria are met.
Which approach is most appropriate?
Integrate Glue Data Quality rules into the pipeline and quarantine or notify on a failed quality gate.
Detailed explanation
Completeness, uniqueness, and range checks can stop output that fails the required criteria.
Completeness, uniqueness, and range checks can stop output that fails the required criteria.
Post-publication observation does not prevent bad data from reaching consumers.
Post-publication observation does not prevent bad data from reaching consumers.
Deletion removes evidence needed for investigation and replay.
Deletion removes evidence needed for investigation and replay.
Silent replacement hides the cause and changes the meaning of the analysis.
Silent replacement hides the cause and changes the meaning of the analysis.
Check official documentation
Compare the following points with the official documentation.
AWS Glue Data Quality、rules、evaluation、ETL integrationの公式説明を確認する。
What to confirm
可視化後の発見と、処理前の品質ゲートの違いを説明できる。
Key points
- Data Quality
- Completeness
- Uniqueness
- Quality gate
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
Rule specificity
Translate business tolerances into measurable quality rules.
Quarantine
Keep failing rows or partitions for investigation and replay instead of silently changing them.