1.3Standard
Create fraud features from order history.
Only use information available at prediction time.
Choose two designs.
× Incorrect
Compute features from prediction-time history only, manage definitions, units, missing values, and versions, and use the same processing for training and inference.
Detailed explanation
CorrectAggregate only history available at prediction time
Future information cannot leak.
IncorrectUse a full-period average including future orders
Unavailable future data leaks.
IncorrectUse a different unit only in production
Training and inference meanings diverge.
CorrectManage feature definitions, units, missingness, and versions consistently
Training and inference remain aligned.
Check official documentation
Compare the following points with the official documentation.
AWS公式AIF-C01 Domain 1.3の特徴量、データリーケージ、ライフサイクルを確認する。
Official references
What to confirm
特徴量の再現性と未来情報の混入防止を説明できる。
Key points
- Feature
- Time
- Leakage
Notes
- Environment: AWS公式AIF-C01試験ガイドとAWS公式ドキュメントの確認
- Compare the definition with the AIF-C01 objective and the service documentation before applying it.
- Keep the concept separate from a particular answer choice or implementation step.
Foundation review
Feature
A numeric, categorical, or derived value supplied to a model as input.
Feature store
A system that manages feature definitions and values consistently for training and inference.
Question IDAWS-AIF-275
EnvironmentAWS公式AIF-C01試験ガイドとAWS公式ドキュメントの確認
Last technical review2026-08-20
Report an error or rights issue