In anticipation of his upcoming conference keynote presentation, Fraud Screening for 2/3rds of All Card Transactions: A Consortium and Its Data at Predictive Analytics World Financial in New York City, October 23-27, 2016, we asked Scott Zoldi, Chief Analytics Officer at FICO, a few questions about his work in predictive analytics.
Q: In your work with predictive analytics, what behavior or outcome do your models predict?
A: Falcon models generate transaction scores which determine the likelihood of the transaction being fraudulent, by incorporating multi-faceted fraud risk assessments from the cardholders’ historic spending behavior, characteristics of merchants and transaction instruments. The higher the score is, the more likely the transaction is fraudulent.
Q: How does predictive analytics deliver value at your organization – what is one specific way in which it actively drives decisions or operations?
A: Falcon models are data-driven analytical models built from the FICO Fraud data consortium where banks agree to pool their payment card transaction and fraud data for building the most predictive models. These predictive models mitigate the financial impact of credit and debit card fraud and improve customer experiences. The pursuit of the highest performing model has driven the creation of a payment card fraud consortium data asset that contains more than two decades of historical data.
Q: Can you describe a quantitative result, such as the predictive lift of your model or the ROI of an analytics initiative?
A: Since introducing Falcon models based on the data consortium, we monitor 2.5 billion payment cards world-wide, and the payment card fraud losses have been reduced by more than 70% in the US. The data consortium has enabled the creation of 88 granted fraud analytic patents and more than 44 fraud analytic patent applications pending derived from over 3,000 researcher-years working with the asset to maximize fraud detection while minimizing customer impact.
Q: What surprising discovery or insight have you unearthed in your data?
A: One is that in 2014 nearly half of the fraudulent cross-border transactions on UK debit cards took place in the US, as US had not yet implemented EMV technology. The card-not-present transactions (for example, online purchase) are seen to grow rapidly from year to year in volume and the fraud rate is much higher than card-present transactions. The data also reveal the occurrence of data compromise events associated with some merchants that lead to exposure to card data in the dark web.
Q: Sneak preview: Please tell us a take-away that you will provide during your talk at Predictive Analytics World.
A: One take away that will be discussed is how we utilize auto-encoder technologies to ensure that the data on which our models are built is similar to the data seen in production. This provides an ability to monitor data through these deep learners in order to determine when we may have models that will perform sub optimally based on transaction pattern shifts in production data environments.
Don’t miss Scott’s conference keynote presentation, Fraud Screening for 2/3rds of All Card Transactions: A Consortium and Its Data at Predictive Analytics World Financial NYC, on Wednesday, October 26, 2016 at 8:50 to 9:40 am. Click here to register to attend.
By: Eric Siegel, Founder, Predictive Analytics World
Eric Siegel is the founder of Predictive Analytics World (www.pawcon.com) — the leading cross-vendor conference series consisting of 10 annual events in Boston, Chicago, San Francisco, Toronto, Washington D.C., London, and Berlin — and the author of the bestselling, award-winning book, Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die.