I am attending the 2010 Predictive Analytics World event in San Francisco. Parallel to an overwhelming interest and awareness in analytics in general, the conference has been sold out.
Day 0 (Feb. 14th) kicked off with a pre-conference workshop featuring SAS Enterprise Miner with Dean Abbott. 24 attendees showed up ready to take a deep dive into data mining using KDD Cup 1998 data. The attendees came from variety of industry segments, with only a few having SAS experience. The exercises followed the steps laid out in the analytic lifecycle to develop and score a predictive model that identified donors with a high likelihood of responding, to make fundraising cost effective. We received invaluable feedback on the key role analytic data preparation plays in building better, effective models. One of the attendees commented, “SAS Enterprise Miner is so easy to use that it has set the bar for me for any other data mining package.” Day 1 of the conference was kicked off with Eric Siegel’s keynote focusing on new and innovative uses of analytics (text analytics being one of them), and how analytics can be used to minimize operational risks and solve critical operational decisions. Eric added a fresh perspective that collective experience of your organization also plays a key role in building predictive models.
The day quickly turned into a call out for text mining and text analytics. John Elder described specific text mining case studies in government and public sector. He called text mining and text analytics the “wild west” of analytics with lots of innovation coming out in next few years. Manya Mayes followed it up with a SAS keynote on best practices around text analytics initiatives, including data quality, categorization/classification, and sentiment analysis.
Dean Abbott’s session described how a Fortune 500 company applied text mining to help desk calls related to repairs of supported devices. The end results were used to develop business rules that exceed specifications of parts needed for repairs.
The day also featured two customer case studies where SAS is being used for predictive analytics. Michael Thomas from Group RCI discussed membership lifetime value modeling using standard concepts like attrition and revenue forecasts with soft components like value received from a member’s vacation ownership for exchange. The second presentation was from Walmart Financial Services on market mix modeling to calculate effect of individual campaigns, accounting for seasonality, on sales at the store level. Marketing mix optimization uses weekly sales, local store traits, demographics, and macro-economic data measure the effectiveness of specific campaigns and make sure that the investment is targeted and productive.
The trade show floor and SAS booth area are staying busy with steady stream of customers and prospects inquiring about SAS products and capabilities. Of course there was a lot of interest in our new SAS Text Analytics offerings launched at the event. The conference organizers opened up the evening reception for Bay Area SAS users also and hence we suddenly got busy.
I’m looking forward to tomorrow’s lineup with more case studies from Charles Schwab, Deutsche Postbank and Paypal. We will also have more sessions on Text Analytics and In-database analytics.
