Advanced Data Mining

Marketing secrets from the financial sector show how data mining charts a profitable course to customer management

1 2 Page 2
Page 2 of 2

As data mining vendors improve their products, they've begun to assert that their tools can be used by business managers without extensive training in statistics. Beware of such pitches, data mining practitioners say. To get the best results, you need experts who understand the advantages and limitations of the half-dozen or so algorithms used in data mining products. "There are some products where you can try a canned approach and that might work sometimes," says Hoffman. Those tools are far more valuable, though, in the hands of people who understand what's going on under the hood, he says.

Bank of Montreal's Mrazek says sometimes his staff applies several different statistical models to one problem. The most significant results might lie in the intersections of each model's results, and you need a thorough understanding of the algorithms to interpret them, he says. Banks that have made serious investments in data mining typically employ several people trained in statistical analysis or applied mathematics. "If you don't have the statistical expertise, you're using the tools blindly," Mrazek says.

So while data mining specialists extol the virtues of the technology in one breath, they caution about its complexity in the next. "Data mining is a powerful technology, but it's not magic," says Herb Edelstein, cofounder and president of Two Crows Corp., a data mining consulting firm based in Potomac, Md. "It doesn't sit in your computer, find things in your databases and tell you what to do with them." Data mining can be downright tedious, Mrazek says. Some models can take several months to complete.

But, like gazing at one of those 3-D posters, persistence pays off. And banks are now seeing revealing pictures of their customers within the "dots" housed on their databases. Those images reflect higher profits.

Related:

Copyright © 1998 IDG Communications, Inc.

1 2 Page 2
Page 2 of 2
7 secrets of successful remote IT teams