3.3Standard
A classifier returns confidence 0.8.
Actual accuracy differs by department and input type.
Choose two appropriate evaluations.
× Incorrect
Compare predicted and actual accuracy by confidence band and evaluate precision, recall, and business cost at different thresholds.
Detailed explanation
IncorrectUse only the overall average to declare calibration
Subgroups and input differences can be hidden.
IncorrectDisplay the raw score as an exact probability
An uncalibrated score is not a probability.
CorrectCompare precision, recall, and business cost at different thresholds
Decision policy impact becomes measurable.
CorrectCompare predictions with actual accuracy in each confidence band
The score can be checked as a probability-like value.
Check official documentation
Compare the following points with the official documentation.
AWS公式AIF-C01 Domain 3.3の評価指標、信頼度、分類しきい値を確認する。
Official references
What to confirm
内部スコアと校正された確率を区別し、閾値の影響を評価できる。
Key points
- Calibration
- Precision
- Recall
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
Calibration
The degree to which predicted confidence matches the observed frequency of correct predictions.
Threshold
A boundary that turns a score into an action such as accept, reject, or review.
Question IDAWS-AIF-212
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