GUEST COLUMN: Myths about data mining

- Thursday, June 9, 2011

GUEST COLUMN: Myths about data mining

By Patricio Llaona, head of marketing for the Caribbean and Latin America, Teradata

A lot of successful companies have discovered that many of the stories surrounding data mining are just myths. Rather than falling victim to them, visionaries have gained enormous competitive advantages through the use of data mining to solve complex business problems and achieve profitability.

In fact, it was sophisticated data mining technology that convinced major retail companies such as Walmart to stock up, for example, on a special version of canned meat for the hunting season. Go ahead, laugh. Much more than a proposal, the idea helped Walmart generate additional revenue during the season when this product was consumed most, according to the analyses, and it is emblematic of how well this supermarket chain understands its customers.

Start your 15 day free trial now!

cta-arrow

Already a subscriber? Please, login

So what is data mining?

It's is a powerful analysis tool that allows business executives to use historical client behavior to predict future conduct. This technology finds patterns that reveal the mysteries of consumer behavior. Its findings can be used to increase revenues, decrease costs and identify business opportunities, offering new competitive advantages.

One of the reasons for the myths related to data mining is that people are confused about what it really is. In essence, data mining is a group of complex mathematical techniques to discover and interpret previously unknown patterns within detailed sets of data. Since the mid-1980s, the use of data mining has expanded from academic, medical and scientific research to its effective application in retail, banking, telecommunications, insurance, tourism and the hotel industry.

Since data mining is considered an analytical tool, it is often confused with online analytic processing (OLAP). OLAP is a valuable analytical technique when used to examine business operations to gain a historical perspective of something that happened. For example, when a marketing head wants to understand why sales fell in a particular region. OLAP tools allow you to ask questions across multiple dimensions, such as sales per store, sales per product and sales over time. By looking at historical data from different perspectives, it enables you to analyze the factors (store, product or time) that had an impact on sales.

Data mining is about solving other problems. It can be used to predict future events, such as the next month's sales in relation to advertising or what kind of customer may respond to a particular advertisement.

The increasing use of this technology dispels the main myths about data mining. First, data mining does not provide immediate results like a crystal But with background data that accurately reflect the business, the process is effective and efficient. In this regard, it's all about the data.

Second, data mining does not require a specific and independent database. Generally, suppliers claim that this kind of database is necessary for efficient processing.

Nowadays, advances in database technology mean that data mining is no longer done in separate data marts. In fact, effective data mining requires a data warehouse for the entire company, which costs considerably less than hiring independent data marts.

Finally, some people believe that data mining is so complex that you need at least three PhDs to make it work - one in statistics or quantitative methods, one in business and one in computer science.

The truth is that successful projects have been carried out without the need for any advanced degree. Data mining is a collaborative effort between people with training in these three areas. The business people have to guide the project, creating a set of specific questions from their area and then interpreting the emerging patterns. The analytical modelers, who know techniques, statistics and data mining tools, have to construct a reliable model. The people who work in systems services contribute their knowledge of data treatment and interpretation, in addition to supplying key technical support.

The end result: Data mining is no longer a slow, expensive or very complicated process to carry out. The technology and business know-how is available to implement efficient processes that are feasible in terms of costs. Companies of all sizes are challenging the old myths and are demonstrating that data mining is essential to succeed in today's highly competitive and customer-centered business world.

DISCLAIMER: The content of this piece is entirely the responsibility of the author and does not necessarily reflect the views of Business News Americas. We encourage Guest Column pieces, and those interested in submitting one for possible publication should contact the editor at cmolinari@bnamericas.com