Home / AWS Certified AI Practitioner (AIF-C01) / 1 Practice / AWS-AIF-315
Split customer data into training, validation, and test sets.
Do not inflate the final result.
Choose two practices.
Hint 1: Focus on the requirement Keep final evaluation independent from model choices, and split by time or entity so information cannot cross the prediction boundary.
Hint 2: Key terms Review test isolation and data leakage.
Data leakageInformation unavailable at prediction time that accidentally enters training or evaluation.
Holdout testData kept separate from model selection for a final performance estimate.
Entity splitSeparating records by customer, device, or other entity to prevent cross-entity leakage.
Temporal splitUsing earlier periods for development and later periods for evaluation. × Incorrect Keep test data out of model selection and split by customer or time so related records and future information do not cross the boundary.
Detailed explanation A Correct Keep test data out of model and threshold selection until final evaluation
Generalization remains measurable.
B Incorrect Use future values as current prediction features
Unavailable information leaks.
C Incorrect Tune features repeatedly after viewing the test score
The test set becomes a training signal.
D Correct Split by customer or time and prevent related or future records from crossing
Dependence and future leakage are reduced.
Check official documentation Compare the following points with the official documentation.
AWS公式AIF-C01 Domain 1.2のデータ分割、リーケージ、評価を確認する。
Official references Official documentation 1 Official documentation 2 What to confirm テストデータと未来情報の漏えいを避けた評価手順を説明できる。
Key points Test isolation Leakage Time split 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 Data leakage Information unavailable at prediction time that accidentally enters training or evaluation.
Holdout test Data kept separate from model selection for a final performance estimate.
Path & Pass is an independent learning site and is not affiliated with any certification body. Questions are original practice content based on published objectives.