Answer-first summary for fast verification
Answer: F-score where recall is weighed more than precision
In this scenario, the dataset is highly imbalanced with 96% of examples not containing the company's logo. Accuracy can be misleading because the model could achieve high accuracy by simply predicting 'no logo' for all images. The F1 score, which considers both precision and recall, is a more appropriate metric in such cases. However, given the high imbalance, where the minority class (images with logos) is of higher interest, it is crucial to prioritize recall over precision to ensure that the model correctly identifies as many logos as possible. Therefore, the F-score where recall is weighed more than precision (Option A) is the best metric to evaluate the model's performance in this scenario.
Author: LeetQuiz Editorial Team
Ultimate access to all questions.
You are working on a binary classification machine learning algorithm to detect whether an image of a scanned document contains a company’s logo. The dataset you're using is highly imbalanced with 96% of examples not containing the logo. Considering this skewed distribution, which metric would give you the most confidence in evaluating your model's performance?
A
F-score where recall is weighed more than precision
B
RMSE
C
F1 score
D
F-score where precision is weighed more than recall
No comments yet.