Data mining profile: IBM
Comments of Dan Graham, business intelligence solutions executive, IBM
”Sometimes a lot of things are mushed together as ‘data mining.’ Lately, data mining has crystallized. It’s a facet of statistics. But it’s a specific kind. SAS Institute and SPSS have done a great job with math packs. Data mining today is quite different than classic statistics. It’s pattern detection software. We’re locked into neural nets, advanced regressions. Those are popular with customers.”
He chides that most of that terrifies people who were glad to leave math in college. To simplify things, he likens data mining to math word problems. Data mining is useful in CRM. You can make offers to customers, to which they respond, basing your predictions on analysis of accumulated data.
“In data mining, most of the activity is in preparing the data to work on. Those are big labor costs. The work is almost identical to what you need to do a data mart or warehouse. So there’s a downstream benefit to the warehouse work.”
Back to Data mining comes of age