This form of data mining is when patterns emerge that really have no connection but a connection is “conceived” when one doesn’t really exist. Randomness will eventually exhibit patterns, given a large enough sample. However, randomness is still random. Just because a woman from Indianapolis buys an apple tree, doesn’t mean that all women in Indianapolis are suddenly struck with an urge to have apple trees. Even if fifty women in Indianapolis buy a certain kind of apple tree, that pattern doesn’t mean that women in Indianapolis must have apple trees to be fulfilled. The explanation for this seeming pattern of data could be that a women’s group asked its members to all go out a purchase one apple tree to help replace those in an orchard of a farmer down the road.
This very simplistic example of how an apparent pattern may not mean what it appears to mean shows that large amounts of data may seem to tell advertisers one thing, when the truth is very different from what they data may seem to suggest. This form of data mining is called by a less neutral term, which is “data dredging.” In this instance, the data is misleading and of no value to the potential advertiser or consumer. For data mining to be helpful to all, it must be used responsibly and with further research which confirms that the connections made in the sample are indeed patterns, and not random occurrences of the data.