Welcome to your interview.
The question is on your right: Detect Rare Payment Fraud. Take a moment with it first.
Talk your thinking through with me if you like - when you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes). Discussion and graded submissions share your five interviewer interactions, so spend them well.
You are building a binary classifier for a digital payments product to detect fraudulent transactions before they are approved. Fraud cases are rare, but missing them is costly, while too many false positives create friction for legitimate users.
How would you approach training and evaluating a model on a highly imbalanced classification dataset in this setting? Explain how you would handle the class imbalance, choose metrics, and make tradeoffs between catching more fraud and limiting unnecessary declines.