Data mining defined
Data mining is a two-fold process that encompasses the extraction of valid, meaningful information from databases and data warehouses, and the analysis of previously unknown data relationships that can help guide strategic business decisions.

Data mining can help organizations retain customers and increase marketing effectiveness, which, in turn can increase buying, cross-selling, and return on investment (ROI) in enterprise data systems.

Some of the information data mining helps companies find in their data stores include:

• Associations ­ two or more events can be correlated to each other. An example would be finding the relationship that red wine buyers are the same customers who purchase ABC brand of corkscrews.
• Classifications ­ discovering patterns that lead to a new organization of data. For example, a corporation finds patterns that allow it to develop profiles of champagne buyers.
• Clusters ­ the unearthing of previously hidden groups of information. For example, champagne buyers may be found to purchase particular brands during particular times of the year.
• Sequences ­ discerning events that lead to other events. For example, consumers who buy particular brands of white wine will purchase a wine refrigeration unit within X months.
• Forecasts ­ discoveries of patterns that one can use to predict the future. For example, the patterns of buyers of Canadian lager beer lead to a prediction that they will also purchase boxed white zinfandel wine.

There are various data mining techniques available, ranging in complexity from statistical queries to the establishment of neural networks. The common thread among these techniques is the discovery of new patterns and associations among customer data ‹ ultimately enabling companies to better understand and respond to their customers' needs.

Data mining quick study
Data mining has been around for some time, in one form or another, writes Rachel Konrad in ZDNet's article "Data mining: Digging user info for gold" (Feb. 8, 2001, (http://www.zdnet.com/zdnn/stories/news/0,4586,2683567,00.html). Statisticians have pored over databases manually, seeking clues to patterns. Utility companies were among the first to begin using the automated technique decades ago, but the corporate world didn't begin to catch on until recently. By searching for patterns to predict customer behavior, insurance and credit-card companies started using data mining as a means to prevent or reduce fraud.

Increasingly, organizations began looking at data mining as a means to leverage greater returns on investment for their data warehouses ­ the networked, subject-organized database repository of all relevant and meaningful company information ‹ and their enterprise systems.

But implementing data mining is no easy task. In a white paper ("Deciding on Storage for Business-Critical, Enterprise Data Warehousing," 1998, http://www.hp.com/solutions1/e-intelligence/bi/storage/idc_storage.pdf) IDC Research reports that companies are experiencing data growth anywhere from 20 to 120 percent annually, which puts enormous strain on data warehousing capabilities and in turn makes focused and intelligible data mining operations that much more complex.

But companies are taking on the challenge anyway. The Palo Alto Management Group (PAMG) ("Database Solutions," http://www.pamg.com/dbsolutions/white_paper.html) reports that surveyed organizations gave the highest ranking to these reasons for implementing data mining:

• Improving decision or management processes
• Improving customer service
• Staying ahead of the competition
• Refining corporate strategies
• Reducing operations costs
• Retaining customers
• Identifying new customers

PAMG also reports that the surveyed organizations' most important criteria for buying a particular data mining solution are:

• Data integrity
• System reliability
• Query capability
• Scalability
• Availability
• High performance-to-price ratio

PAMG forecasts that the entire database solutions market will grow to a $113 billion industry by 2002 ‹ of which data mining is expected to comprise a quarter. With such numbers, it's not surprising that scores of vendors have entered the data mining fray with product solutions. Currently, some of the major players include DigiMine, IBM, Microsoft, Oracle, and SAS Institute.

Here are three essential articles that provide a thorough insight to data mining technologies and issues:

"Collaborative Business" (InformationWeek, May 7, 2001) reports the results of the magazine's Information Sharing & Collaboration study. A striking new trend is for businesses to share data mining information with each other. Examples cited include Home Depot, which is sharing data with several suppliers. Other companies in the sharing mood include Cummins Inc., Briggs & Stratton, and Whirlpool Corp.
http://www.informationweek.com/836/prcollaborate.htm

"Data Mining Strategies" (DM Review, July 2000) is an excellent overview, if a bit boosterish: "You'll see you don't have to be Einstein to do data mining and that data mining can have widespread positive impact in your organization."
http://www.dmreview.com/portal_ros.cfm?NavID=91&EdID=2367&PortalID=9

"Surviving the Perfect Storm in Data Management" (DM Review, January 2001) notes that "the collection of data that analysts so calmly referred to as a "sea of data" just ten years ago has now swollen to tsunami forces."
http://www.dmreview.com/master.cfm?NavID=198&EdID=2910