Design a fraud mitigation strategy under constraints

You are preparing a one-page briefing for a hiring-manager interview focused on fraud data science. The scenario is as follows: Present situation: The production fraud model is underperforming: approximately $$40\%$$ of genuine fraud is stopped before loss. Here, “stopped” means the transaction is blocked or otherwise prevented from causing financial damage. A disproportionate amount of the remaining fraud seems to come from newly active or less-mature regions. You must work…