PayPal antifraud system
Botha develops a bottom-up financial model to help understand PayPal’s business, and finds that identity fraud is costing the company millions of dollars each month:
I was trying to work out how long it takes on average for fraud to be notified. I realised that our losses were much larger than we had believed. I quantified the size of the problem and spurred engineers to build frameworks for fraud control. We caught the problem early and saved the company from bankruptcy…After identifying the fraud problems, we used logistic regression analysis to build the initial fraud models to predict whether a new user was a good or bad user.