MLS-C01 Question 216
Select 2You are building a machine learning model to classify images into 10 categories. After training the model, you are tasked with evaluating its performance. The evaluation metrics you choose must align with the business requirements: minimizing false positives for critical categories while maintaining overall accuracy. Which combination of evaluation metrics would be most appropriate for this scenario?
- A
Accuracy
- B
Precision
- C
Recall
- D
F1 Score
- E
Mean Squared Error
Show answer and explanation
Correct answers: B, D
Explanation
To evaluate a classification model while focusing on minimizing false positives for critical categories, precision is the most appropriate metric since it directly addresses the false positive rate. However, to ensure a balance between precision and recall, the F1 score should also be considered. Accuracy alone is insufficient in this case because it does not address the trade-offs specific to false positives and false negatives. Mean Squared Error is not applicable as it is used for regression tasks, not classification.
- A. Incorrect.
Accuracy measures the overall correctness of the model’s predictions but does not consider the trade-off between false positives and false negatives. In this case, accuracy alone may not align with the requirement to minimize false positives for critical categories.
- B. Correct.
Precision is the ratio of true positives to the total predicted positives. It is useful for minimizing false positives, which is a key business requirement in this scenario.
- C. Incorrect.
Recall measures the ratio of true positives to the total actual positives. While useful, recall alone does not focus on minimizing false positives as required by the business.
- D. Correct.
F1 Score is the harmonic mean of precision and recall. It helps balance the trade-off between precision and recall, making it suitable for scenarios where both false positives and false negatives need to be considered.
- E. Incorrect.
Mean Squared Error (MSE) is typically used for regression problems, not classification, and is irrelevant to the scenario described.