MLS-C01 Question 228
Select 3You are building a binary classification model to predict whether a customer will purchase a product or not. The dataset is highly imbalanced, with only 5% of the customers making a purchase. After training the model, you evaluate its performance using various metrics. Which of the following metrics would be most appropriate to assess the performance of your model in this scenario?
- A
Accuracy
- B
Precision
- C
Recall
- D
F1 Score
- E
Root Mean Square Error (RMSE)
Show answer and explanation
Correct answers: B, C, D
Explanation
In highly imbalanced datasets, metrics like accuracy can be misleading because they do not account for the class imbalance. Metrics such as precision, recall, and F1 score are more appropriate because they focus on the correct identification of the minority class (positive cases) and handle trade-offs between false positives and false negatives. RMSE is irrelevant in this scenario as it is used for regression models, not classification.
- A. Incorrect.
Accuracy is not a reliable metric in highly imbalanced datasets because it can be misleading. For example, a model that predicts 'no purchase' for all customers would achieve 95% accuracy without actually being effective.
- B. Correct.
Precision is a useful metric in this scenario because it measures the proportion of correctly predicted positive cases out of all predicted positive cases. This is particularly important when false positives are costly or need to be minimized.
- C. Correct.
Recall is another important metric here because it measures the proportion of actual positive cases that were correctly identified. This is important in scenarios where missing actual positive cases (false negatives) is critical.
- D. Correct.
F1 Score is a balanced metric that combines both precision and recall, making it particularly suitable for imbalanced datasets. It provides a single score that considers both false positives and false negatives.
- E. Incorrect.
Root Mean Square Error (RMSE) is generally used for regression problems, not classification problems. It is not suitable for evaluating the performance of a binary classification model.