ML functional performance metrics · Question 41 of 60
Of 1,000 transactions, 10 are fraudulent. A model classifies all transactions as "no fraud" (fraud = positive). Which statements are correct?
3 answers are correct. Hard
Show answer
Answer
- A Accuracy alone is misleading as an acceptance criterion here
- C The recall is 0%
- D The accuracy is 99%
Explanation
There are TP = 0, FN = 10, FP = 0 and TN = 990. Accuracy = 990 / 1000 = 99%, recall = 0 / 10 = 0%: not a single fraud case is detected. Precision is undefined because there are no positive predictions (0 / 0). Although accuracy is high, the model does not serve its purpose, which is why metrics such as recall, precision or F1-score are needed.