Home / AWS Certified AI Practitioner (AIF-C01) / 4 Practice / AWS-AIF-085
One classification threshold produces very different precision and recall across groups.
Consider fairness and business requirements together.
Choose two considerations.
Hint 1: Focus on the requirement Compare group-level outcomes before and after a threshold change and document the trade-offs that result.
Hint 2: Key terms Review the fairness and threshold concepts used for group comparisons.
Fairness metricA defined measure for comparing treatment, error, or impact across relevant groups.
Group-level errorA prediction error measure calculated separately for a particular group.
Decision thresholdA score boundary that determines whether a model output triggers a class or action.
Trade-offA change that improves one objective while affecting another objective, group, or risk.
Re-evaluationRepeating relevant measurements after a change to determine its actual effect. × Incorrect Measure group-level errors and relevant legal or policy requirements, then reassess all groups after any threshold change.
Detailed explanation A Correct Measure group-level errors and review legal and policy requirements
A single metric cannot describe every impact.
B Incorrect Declare legal compliance from accuracy alone
Legal and ethical compliance cannot be concluded from accuracy alone.
C Correct Re-evaluate overall and group performance after changing thresholds
An improvement for one metric can create another disparity.
D Incorrect Exclude groups from evaluation to hide the difference
Exclusion does not improve fairness.
Check official documentation Compare the following points with the official documentation.
AWS公式AIF-C01ガイドDomain 4.1の公平性・バイアス評価を確認する。
Official references Official documentation 1 Official documentation 2 What to confirm 指標、影響、方針、変更後評価を組み合わせた公平性検討を説明できる。
Key points Group metrics Trade-offs Re-evaluation 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 Aggregate masking An overall score can hide unequal errors or impact between groups.
Policy context A metric result must be interpreted with the intended use, law, and organizational policy.
Path & Pass is an independent learning site and is not affiliated with any certification body. Questions are original practice content based on published objectives.
