Problem is, most organizations don’t have people who know how to find those nuggets—predictors that could help them target the right customers for their products, find the most likely donors for their program, or figure out the right chemical combination for the next wonder drug. But two University professors are solving that problem. Elaine Crable and Candace Gunnarsson are team-teaching a course on data mining, the newest and most exciting development in the field of high-tech research.
“We’re really on the cutting edge,” says information systems department chair Crable. “A lot of people are doing courses on data warehousing where you learn how to store data, but not a lot of people are doing mining. The idea is we have all this data and it’s like mining for gold. You have to go through all this wasted stuff to find the nuggets. It’s used in the medical area to predict who gets diseases and in the financial area to predict risky investors. It’s just starting in the industries. They want to get it onto campuses and for students to learn it for their industry.”
Crable and Mary Walker, associate vice president for academic affairs, secured a grant from the General Electric Fund. The grant covers the cost of training and developing the course for business majors, as well as paying for a future research assessment of the program.
The University supported the grant by buying a software program called Enterprise Miner.
Gunnarsson, a statistics instructor recently appointed vice president of corporate training for a local data mining firm, stepped in to teach the course with Crable.
To learn what data mining is, students must first learn what it isn’t. Collecting data and warehousing it is not data mining, Gunnarsson says. Extracting specific information from the data that leads to creation of a desired model is.
It falls under the category of: Now that we have all these computers generating all this information, what are we to do with it?
The reason businesses are so suddenly interested in it is easy: money. Companies can save millions of dollars by mining their abundance of data and pulling out highly select, specifically targeted audiences. The more specific their target, the better the return on their marketing efforts, Crable says.
If officials from the grocery store chain Kroger, for instance, know the names and addresses of everyone who used their Kroger Plus card when purchasing a bottle of Rosemont Merlot, why bother advertising to a mass audience that might not be interested? A coupon to those specific people is less expensive and more effective.
Other industries have uses for data mining as well.
Models can also be used by law enforcement agencies, for example, to track down snipers or terrorists. Auto manufacturers and dealers can use it to detect fraud.
“This is a very hot field,” says Gunnarsson, who gets calls from other universities interested in adopting the course. “We plan to do a pedagogical document to tell others how to put together a course. Big companies are starting to use it in a big way, but schools aren’t really teaching it yet.”
Crable believes students who take the course will have enhanced opportunities to find jobs after graduation because they’ll be more valuable employees.
First semester students practiced real data mining by using a Veterans Administration database of people who had been solicited previously for donations. The assignment was to create a model that would predict those most likely to donate based on actual results.
The practice isn’t void of potential problems, however. Gunnarsson says they discuss the ethical issues that could arise such as privacy of data and misuse of information.
“Like anything, there can be abuses of the system,” she says. But there’s so much potential, too.