Here's a quick update from the Analytics 2012 conference.
I've seen lots of interesting talks. Frank Travisano and Hongsen Zhou from Chubb presented some of their work incorporating association rules into predictive models. The approach was fresh, using the confidence statistics from a market basket analysis as inputs to a logistic regression analysis to predict whether an individual purchased a product. I wondered about how this could be extended to predict alternative outcomes (less related to the market basket rules) such as customer lifetime value or time until churn, for example. Something to play with on the plane coming home, perhaps? If you attended that presentation and had some thoughts, please leave me a comment!
I've also had a chance to catch up with some of my favorite people. Here's Herb Kirk (Consultant and former Vice President of SAS Education), fielding questions about NC State University’s Institute for Advanced Analytics. If you're in the Exhibit Hall, stop by his booth and learn what they have to offer.