This post on boingboing describes an effective way Target is using data mining -- to predict who's pregnant or not based on their purchases. The original article
appeared in the New York Times and can be found here. The boingboing article describes the data mining as "creepy", while
this post at the abbot analytics blog is a little more on the side of the informatics. In either case, Target has mined the data about customers buying habits to figure out with good confidence which customers are pregnant.
Clustering is one of the core parts of data mining. In clustering, a set of points are split into clusters where each cluster presumably contains related data. The
article for the bottom picture is a little bit technical, but it describes how to cluster users based on musical preferences.
In
classification, given some data, you want to classify that data. So based on its attributes, you want to know which group it goes in. Clustering works with all points and finds groups. Classification works when you know all the groups and want to add a point. They go hand in hand.
