Data Mining

I have found data mining of interest for quite some time. My first exposure was at an IBM seminar where they discussed data mining as applied to the supermarket. The one thing I remember from all those years ago was combined purchasing habits. The idea is to place combined purchases in a close location, such that if one goes in to buy item A, more than likely they will be buying item B. As a customer service issue, it saves them time, due to convenience. As a revenue enhancer, the person that is going in to buy A, may not have even thought of their need for B. As such, by making them close to each other.... more than likely B will be purchased, even if that was not the original intent.

Its a win win for all. The customer saves time, and the store does additional business. For example, milk and orange juice are always within line of sight. It makes sense. The same can be said about having baby wipes next to diapers. What data mining does, is capture the less than obvious purchasing habits of a consumer. The data is all there in even the most simplistic cash register systems. Its just a matter or sorting it out. If we add in the consumer discount purchases, we can also capture age, sex, and long term buying habits. Its way more than simply getting a free bag of cat foof for every ten at Petco. They can target coupons, mailers, and email to what your needs are. The result, applicable ads eg if you send me a coupon or deal on dog products I'm guaranteed not to buy. I don't have a dog. If you send me kitten ads, I just might do so, as we have a bunch of them. Its a win for both, I get less junk mail, the store gets greater purchases. However, there is an insideous priavcy issue. Do you really want the store to know that much about you????

Getting back to the IBM data mining example. Lets say item A is diapers... so we would expect item B to be baby powder, baby wipes, etc. In reality, its not even close. Apparently the consumption and packaging of the ancilliary items does not make them correlate evenly with diapers. What does correlate is BEER. This caused much laughter and crude comments by the engineers. Yet the data was right there in front of us. The combination is so odd, it would not be acceptable to the consumer to have them side by side. However, since the seminar all those years ago, I have noticed that beer is often within line of sight, or within an isle of baby products. Next time you are in a Super Walmart, note where the baby items are relative to the beer.... you will find it of interest. It is probably not by accident.