Your question is Debugging a Failing ML Model. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
You're reviewing a supervised learning model that is underperforming in validation or production, and you need a structured way to find the root cause.
How do you approach debugging a machine learning model?