How to Buy Data Mining: A Framework for Avoiding Costly Project Pitfalls in Predictive Analytics

Information Management Magazine, October 2005

Eric A. King

How does someone purchase an intangible, cryptic, seemingly immeasurable technology? Beyond the inherent up-front risks of engaging in what is essentially a discovery process, just identifying a starting point can be intimidating and mystifying. Despite its elusive nature, data mining technology has surpassed the flash-in-the-pan "miracle tool" stigma with widespread and sustained success stories highlighted in mainstream publications, along with recurring case studies of improved operational efficiencies, enhanced business intelligence and residual payback. For any organization with annual revenues more than $50 million, employing data mining technology is not a matter of whether, but when.

Data mining has been seeping into mainstream business applications for more than two decades. Numerous case studies may be quickly referenced via a simple Internet or publication search. Its progress is unstoppable, propelled by sustained value justifications- yet stinted by the complexities of development, interpretation, integration and adoption. This article will suggest how to properly approach the starting line and how to implement a purposefully flexible framework for establishing an efficient and effective organizational data mining process.

Cutting Through The Buzz

Let's first make sure that we're on the same page when talking about data mining. It is not wholly incorrect to label data mining as retrospective searches on a large database for specific criteria, otherwise known as online analytical processing

(OLAP) or SQL queries. An example of OLAP or SQL queries would entail mining a large repository to identify females between the ages of 28 and 45 from New York, New Jersey and Delaware with incomes between $65,000 to $90,000 who purchased blue slacks between July 1 and August 15. For this query, we know the exact question to ask of the database. This practice typically explores just 5 to 15 percent of a large database.

For the purposes of this article, data mining shall refer to computer-aided pattern discovery of previously unknown interrelationships and recurrences across seemingly unrelated attributes in order to predict actions, behaviors and outcomes. Simply put, when referring to data mining in this article, we are looking at prediction derived from information hidden within large volumes of data rather than retrospection drawn from an OLAP or SQL query.

It is important to recognize and relate much of the popular terminology that is thrown about in order to provide context going forward. Data mining technology is not new. Methods for automating pattern discovery and prediction have existed for decades. Despite a considerable level of hype and strategic misuse, data mining has not only persevered but also matured and adapted for practical use in the business world. How could a community that is so data-rich, yet information-poor and profit-driven abandon a tool that can validate its own ability to predict customer behavior?

Alongside the technology, terminology has evolved over the last four decades. Names from 40 years ago are still recognizable as common phrases today. In the '70s and '80s, names such as artificial intelligence and machine learning that implied the computer had its own consciousness were somewhat oversold (perhaps even "over-souled"). The names of various data mining building blocks such as neural networks, genetic algorithms and evolutionary computing deservedly carry Darwinist tones of natural selection, as the underlying mathematics emulates biological processes. From a mathematician's perspective, these processes may be viewed as statistics on steroids.

In the '90s through the early '00s, the technology has been commonly referred to as data mining and knowledge discovery. However, due to the duality of the term data mining often referring to both OLAP and pattern discovery, a shift is rapidly moving toward far more descriptive and accurate nomenclature such as predictive modeling and predictive analytics. In fact, this will probably be one of the last articles I write using the label data mining - which is too mainstream to abandon just yet.

Just What is Data Mining?

Is data mining considered a service? Is it hardware? Software? A scored file? A system or a process? A customized solution? There does not seem to be a consensus, which makes data mining all the harder to visualize, define, manage ... and purchase. Two people may discuss data mining and have entirely different concepts in mind. Of course, all of the previously mentioned descriptions are technically correct. While the business community may appropriately view data mining as a productive, value-driven solution, that perspective focuses on the destination, not the journey. If credit were given to the best definition of data mining, process would score the point.

Viewing data mining as a process encompasses all the hard and soft resources, and implies a structured yet ongoing approach to an evolving optimization problem. When viewed as a process, data mining projects may be planned and implemented in a procedural way that all but ensures success. As well, expectations should be inherently leveled to never expect a "final answer" nor anticipate a single pass. When implemented properly, productive results should be expected early and continually improved.

How Not to Buy Data Mining

It is far too common for organizations to adapt their data mining project design to a blend of their perception of what data mining is with a standard corporate practice for evaluating and purchasing products and services. The result is a popular yet doomed approach:

1. Collect product literature from data mining tool vendors at industry events or as advertised in journals.
2. Invite vendors whose retail price of their flagship product fits within available discretionary budgets to visit on site.
3. Gain a free education in data mining through subjective presentations at the vendor's expense (too many are anxious to chase any sales bait, qualified or otherwise).
4. Purchase a data mining tool from the vendor who presented last.
5. Throw some data at the tool and await magical results.
6. Stare at the numbers or even visualizations thereof, wondering why an angelic chorus did not accompany the results.
7. Without knowing whether the results are useless or phenomenal, data mining is dismissed as hyped and/or pie-in-the-sky technology.

Page 1 of 4.