Release 10-23: Drift Detection
Fraud.net's data science team has a new and improved way to detect if a custom model has drifted in performance. Usually, models drift when a new demographic or product has been introduced by a client. As the behaviours of our client's customers change, we want to make sure our models adjust accordingly as well. With our new method of Drift Detection, our models stay up to date and continue to perform well, even as the data received changes.