MLA-C01 Question 199
Select 3You have trained a binary classification model to predict whether customers will purchase a product (1) or not (0). After deploying the model, you evaluate its performance using a test dataset and calculate the following metrics: accuracy = 92%, precision = 70%, recall = 50%. Which of the following conclusions can you draw about the model's performance?
- A
The model performs well overall because the accuracy is high.
- B
The model struggles to identify true positives as indicated by its low recall.
- C
The model is likely making a significant number of false positives given its precision.
- D
The model is suitable for use in a scenario where false negatives are more costly than false positives.
- E
The accuracy metric alone is not sufficient to assess the model’s effectiveness in this scenario.
Show answer and explanation
Correct answers: B, C, E
Explanation
The accuracy metric is not sufficient to evaluate the model in a binary classification scenario, particularly when precision and recall provide more insight into the model's ability to correctly predict positive and negative outcomes. The low recall indicates issues with identifying true positives, while the precision suggests a significant number of false positives. These factors must be considered to determine the suitability of the model for specific business requirements.
- A. Incorrect.
High accuracy alone does not necessarily mean the model performs well, especially in imbalanced datasets. In this case, the low precision and recall suggest issues with the model’s performance.
- B. Correct.
Recall measures the model's ability to correctly identify actual positives. A recall of 50% indicates that the model is missing many true positives, so this option is correct.
- C. Correct.
Precision measures the percentage of true positives among all predicted positives. With a precision of 70%, the model has a notable number of false positives, so this option is correct.
- D. Incorrect.
If false negatives are more costly than false positives, a low recall (50%) would make this model unsuitable for such scenarios, so this option is incorrect.
- E. Correct.
Accuracy alone does not provide a complete picture of the model's performance, especially in imbalanced datasets. Precision and recall should also be evaluated, making this option correct.