easy5 pts · first answer countsmodel-monitoringprediction-drift

Fraud flags tripled

A fraud model normally flags 1.2% of transactions. Today it flags 3.8%, and there is no known fraud attack. What is a good first step?

A.Lower the threshold so even more are flaggedB.Assume fraud tripled and block cardsC.Check input features and recent pipeline or model changesD.Ignore it until monthly labels arrive

Discussion 0

Join free to post in the discussion.

Fraud flags tripled

Discussion 0

Related questions