フリー問題

問題文

A confusion matrix for a fraud detector shows 40 true positives, 15 false positives, 60 false negatives and 9 885 true negatives. Which reading is correct?

選択肢

1. The detector finds 40 of the 100 actual fraud cases, so its recall is 40 percent, and 15 of the 55 alerts it raises are wrong.
2. Its overall accuracy is below 90 percent, so the 60 missed cases and the 15 false alerts together outweigh the correct verdicts in the table.
3. The detector misses 15 fraud cases and raises 60 false alerts, so the priority should be reducing the alert volume before anything else, because the reviewers are being flooded with cases that turn out to be legitimate.
4. Its precision is 40 percent.