
You are reviewing a classifier on an imbalanced dataset where the positive class is rare. The team says the model looks strong on accuracy, but they are not sure that reflects real performance.
How would you evaluate a model on an imbalanced dataset?