You are working on a supervised classification problem where one class is much rarer than the other. The model will be used in production, so false negatives and false positives both matter.
How would you handle class imbalance in a classification problem?
