TMCnet Feature
April 24, 2013

5 Data Mining Tips

Sifting through vast reams of customer information and consumer statistics is a daunting prospect, but data mining like that is the future of any business that wants to stay ahead of the game and its competitors. But what is the best way to sort and categorize all those facts and figures, in such a way that is accurate, accessible to all who need to read it, and presentable to business leaders, shareholders and employees? Here are five data mining tips.

1. Mine More Data

Traditionally, data miners would sample a company's data to answer questions (such as what demographic accounted for the most sales, the least business, etc). With so much data to go through, sampling was the only conceivable way to get the job done - mining the entirety of the data was so unfeasible. Now, however, with computer software to make the data mining process faster and more accurate, businesses should increase the scope of their mining, covering more data to reveal more accurate trends, gleam more perceptive insights, and make stronger decisions based on those trends and insights.

2. Make The Data Mining Results More Available

Reserving the end reports and results of data mining for business owners and shareholders is of little use to the employees who have to liaise with customers, when those employees would benefit from knowing what the data mining says about those customers. The best data mining software compiles readable, accessible reports from mountains of data, so even those not directly involved in the data mining process itself will be able to understand the conclusions and findings of the mining reports. With this information, dealing with customers' concerns or reaching out to new customers in either familiar or unfamiliar demographics, becomes not only easier for the employees who have to do this job, but likely more successful.A business leader should ask him or herself, can their employees on the front lines (as it were) benefit from knowing what the data mining results are? In most cases, the answer will be yes.

3. Use The Data Cautiously

It's always good to keep an open mind with regards the implications and suggested strategies of the data mining, but a business leader should already have a plan in mind for what they want to do with their business. Misusing the results of data mining will be a classic case of thinking everything is a nail when you've got a brand new hammer. The results should be looked at in context and perspective, and should not be blindly or enthusiastically incorporated into a business plan if they run contrary to expectations or against predicted trends from other sources.

It is, in effect, possible to mine big data, only to be surprised by the results. A business owner has to make the decision of adapting a strategy to these unexpected results, or adapting the results to the strategy.

4. Choose the Right Data to Mine

When embarking on a data mining process, it's tempting to throw everything into the slow cooker and see what comes out. Doing this runs the risk of skewing results, adding variables that have no bearing on the business or the expected and/or desired results. Before starting to mine data, a business leader should talk with his or her teams to decide what data to include in the process, and, equally as important, what data to exclude. A different mining operation can be conducted on that second set of data as time and resources permit, but there is much more to data mining than enthusiastically loading every single one of a businesses' net assets into the computer software and clicking the "Start" button.

5. Aggregate Old Data

Customers change over time, and therefore, so will your data. But don't simply junk old data mining results, even if they are out of date and no longer valid. Archive them for future consultation, so that you always have a frame of reference in which to view your current data. A business owner may not be called on to look up this archived data on a daily basis, but when the time comes to analyze past trends and predictions, they'll be very glad that data wasn't simply thrown away after it was used.

Edited by Stefania Viscusi
» More TMCnet Feature Articles


» More TMCnet Feature Articles