Describe fundamental principles of machine learning on Azure →hardMultiple ChoiceObjective-mapped

AI-900 Practice Question: Describe fundamental principles of machine learning on Azure

A bank uses a machine learning model to predict credit card fraud. The model's output is a probability score. The business wants to minimize the number of false positives (legitimate transactions incorrectly flagged as fraud) because these cause customer dissatisfaction. At the same time, they must also catch most fraudulent transactions. Which metric should the bank optimize to balance these two goals?

⚠ Common exam trap

Watch out — candidates often choose precision or recall alone, not realizing that the F1 score is specifically designed to balance both metrics when the business requires minimizing false positives while still catching most true positives.

Answer choices

Why each option matters

Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.

Correct answer & explanation

✓

D: F1 score

The F1 score is the harmonic mean of precision and recall, making it the ideal metric when a balance between minimizing false positives (precision) and catching most fraudulent transactions (recall) is required. In this credit card fraud detection scenario, optimizing F1 ensures the model reduces customer dissatisfaction from false positives while still maintaining high detection of actual fraud.

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

✗
A: Accuracy
Why it's wrong here
Accuracy simply measures the proportion of correct predictions overall. In credit card fraud detection, fraudulent transactions are extremely rare—often under 0.1% of all transactions—so a model that never predicts fraud can appear 99.9% accurate while completely failing at its primary task. Because accuracy gives equal weight to false positives and false negatives, it hides the severe business cost of missed fraud and is therefore a poor performance measure for this use case.
✗
B: Precision
Why it's wrong here
Precision quantifies how many of the transactions flagged as fraudulent are actually fraudulent, so a high-precision model rarely harasses legitimate customers. However, achieving high precision usually requires the model to be very conservative and only flag cases with strong evidence, which allows many subtle or novel fraud patterns to go undetected. For fraud detection, optimizing only precision would sacrifice the model's ability to uncover true fraud, leaving the bank exposed to losses.
✗
C: Recall
Why it's wrong here
Recall measures the proportion of actual fraudulent transactions that the model successfully identifies, making it critical for catching as much fraud as possible. But a pure recall-focused approach drives the model to flag a large number of borderline legitimate transactions as suspicious, frustrating customers and flooding the bank's fraud-review queue. That excessive false-positive burden is a real operational cost, so recall in isolation is not a suitable evaluation metric.
✓
D: F1 score
Why this is correct
F1 score is the harmonic mean of precision and recall, so it provides a single balanced measure of both completeness and trustworthiness of fraud predictions. In credit card fraud modeling, both undetected fraud (false negatives) and blocked legitimate purchases (false positives) are costly, and F1 score specifically rewards models that achieve a strong trade-off between the two. It is the correct metric for this scenario because it avoids the misleading optimism of accuracy on imbalanced data.

Go deeper

AI-900 Practice Question: Describe fundamental principles of machine learning on Azure

Why each option matters

Option-by-option breakdown

Related to this question