Data Mining

Data mining is the process of extracting information from a data warehouse for decision-making purposes. The information comes from different sources and affiliates of the company, and even from outside sources. The data mining process sorts through all of the information and turns it into something useful. The idea is to draw conclusions that you can use to guide future actions.

Data mining has been around a lot longer than computers -- it just wasn't always called data mining. For example, during the War of Independence, locals pieced together information about the movement of British troops from snippets of information gleaned from taverns and by watching the troops, who stood out in their bright red jackets. During the World War II, people were warned by posters that appeared in public places that "loose lips sink ships." And it was important that soldiers didn't write home about their movements or location because, given enough little bits of information, the enemy can put it all together and find out something substantial.

This is the idea behind data mining.

Data mining is used extensively for as a customer relationship management (CRM) tool. Data mining is used to build predictive models and score customers based on relevant criteria. That means that companies can take the information they collect on customers and combine it with demographics information. To this huge data warehouse full of information they can apply data mining and behavioral principles to the information to generate pretty good gestimates on how to better serve their customers. Here are some examples:

• Huntington Bank keeps track of 2,000 cost centers or points of contact within the organization and million of accounts. The bank uses this information to determine what services to offer to what customers and to track customer activity and preferences.

• Fingerhut, the online and catalog retailer in Minnetonka, MN, uses data mining tools to determine what customers to send what catalogs to. This way they save a lot of money and time not sending material to people who will through it in the trash without even looking though it.

• Giga Information Group in Santa Clara, CA, uses data mining techniques to generate business intelligence which requires the integration of many sources of information.

• Cypress Semiconductor Corp. uses data mining techniques to track its customer satisfaction rate. Customers fill out customer satisfaction surveys to find out what customers are not happy and why. The company's philosophy is that nothing distinguishes one customer from another as customers' feelings about the company.

• The New York Times tried to bring new customers based on demographics and it didn't work well. So trying again, the paper discovered that their readers shared certain values. They identified these values (such as an interest in lifelong learning) from mailing lists obtained from outside sources like The History Channel.

• FleetBoston Financial Corp., uses data mining techniques to assess the profitability of each customer. It uses this information to target its marketing efforts to people based on their current and past contributions to the financial company's bottom line.

For more information on these data-mining applications see

• Krill, Paul. "Analytics Redraw CRM Lines," InfoWorld, December 3, 2001, pp. 17-18.
• Heresniak, E.J. "Data Mining," Across the Board, January/February 2002, pp. 65-66.
• Anthes, Gary H. "Picking Winners and Losers," Computerworld, February 18, 2002, p.34.