Fraud is only 0.1% of transactions.
Accuracy alone does not show missed fraud.
Choose two appropriate practices.
Use positive-class metrics such as precision and recall, and state how evaluation sampling relates to the production rate.
Detailed explanation
The important positive-class behavior becomes invisible.
It detects no fraud and fails the business goal.
A sampled test set must not be mistaken for production performance.
A majority-class model can have high accuracy while missing all fraud.
Check official documentation
Compare the following points with the official documentation.
AWS公式AIF-C01ガイドDomain 1.3の分類評価と不均衡データを確認する。
Official references
What to confirm
多数派だけで高く見える正解率の限界を説明できる。
Key points
- Imbalance
- 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
Class imbalance
A data distribution in which some target classes have far fewer examples than others.
F1 score
The harmonic mean of precision and recall for judging a balance between missed and false positives.