Given a dataset on user transactions, how would you identify patterns of fraudulent activity at Chime?
