AWS Certified Machine Learning – Specialty — Question 354

A machine learning (ML) engineer is creating a binary classification model. The ML engineer will use the model in a highly sensitive environment.

There is no cost associated with missing a positive label. However, the cost of making a false positive inference is extremely high.

What is the most important metric to optimize the model for in this scenario?

Answer options

• A. Accuracy
• B. Precision
• C. Recall
• D. F1

Correct answer: B

Explanation

Precision is the ideal metric to optimize when the cost of false positives is extremely high, as it measures the proportion of positive identifications that were actually correct. Recall focuses on minimizing false negatives, which have no cost in this scenario, while Accuracy and F1-score do not isolate and minimize false positives as effectively as Precision.