1.2Standard
Customer data must be split to evaluate a model.
Estimate performance on unseen data.
Choose two practices.
× Incorrect
Keep test data out of training and tuning, and check duplicates, time order, and customer boundaries for leakage.
Detailed explanation
CorrectCheck duplicates, time order, and same-customer leakage before splitting
Evaluation contamination is reduced.
IncorrectTune repeatedly on the test result and report it as final
The test becomes tuning data.
CorrectKeep test data out of model training and hyperparameter tuning
Unseen performance can be measured.
IncorrectUse training accuracy alone for production performance
Overfitting is missed.
Check official documentation
Compare the following points with the official documentation.
AWS公式AIF-C01 Domain 1.2の訓練・検証・テスト分割、データ漏洩を確認する。
Official references
What to confirm
評価データを調整に再利用せず、分割単位の漏洩も確認できる。
Key points
- Test data
- Leakage
- Unseen data
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
Holdout test
An isolated dataset used once or sparingly for an unbiased final estimate.
Data leakage
Information crossing a boundary that makes evaluation unrealistically easy.
Question IDAWS-AIF-414
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