MLA-C01 Question 203
Single answerYou are training a classification model using Amazon SageMaker, and after the initial evaluation, the model has an F1 score of 0.65. Your business requirement specifies a high precision model because false positives can lead to customer dissatisfaction. Which approach would most effectively improve model performance for this requirement?
- A
Adjust the classification threshold to favor higher precision.
- B
Increase the size of the training dataset by adding more labeled data.
- C
Focus on improving recall by tuning the model hyperparameters.
- D
Switch to a different model architecture that is optimized for recall.
Show answer and explanation
Correct answer: A
Explanation
The business requirement emphasizes high precision to minimize false positives. Adjusting the classification threshold is a targeted approach to prioritize precision over recall. Other options, while potentially improving overall model performance, do not specifically address the precision requirement.
- A. Correct.
Correct. Adjusting the classification threshold can directly impact precision and recall trade-offs. By increasing the threshold, you can reduce false positives, improving precision at the expense of recall.
- B. Incorrect.
Incorrect. Increasing the dataset size may improve both precision and recall, but it does not directly address the need for higher precision in the short term.
- C. Incorrect.
Incorrect. Improving recall would likely increase the number of false positives, which contradicts the business requirement for high precision.
- D. Incorrect.
Incorrect. Switching to a model optimized for recall does not align with the business requirement to prioritize precision.