The more (variables), the merrier the results will be for users of business intelligence tools. But that"s difficult to achieve with BI models that can take weeks to create and still include only a handful of variables. And most users find themselves having to rely on such models now, claims Joerg Rathenberg, vice president of marketing and communications at San Francisco-based KXEN Inc. In contrast, he says, his company"s KXEN Analytic Framework tools use advanced modeling techniques based on the statistical learning theories of Vladimir Vapnik, a Russian-born mathematician who is now a research fellow at NEC Laboratories America Inc. and a professor of computer science at Columbia University in New York. KXEN"s software can automatically build predictive models with hundreds or even thousands of variables, according to Rathenberg. He says it can take up to three weeks for a statistician using conventional tools to create a working model that business analysts can use to find "rules" among the information in a data warehouse. "We can do three or four models a day," he says. Release 3.11 is due next month with new association capabilities that rank the most relevant rules divined from BI queries.