Segmentation Tree – Data Segmentation

Advertising campaigns, direct marketing actions, and promotional plans are some of the activities that require marketing managers to identify groups of people which are homogeneous in terms of given characteristics. When the need arises, managers can use one of the segmentation techniques that in the past 40 years have blossomed from William Belson's original work on matching and prediction, such as AID (Automatic Interaction Detection) and CHAID (Chi Square Automatic Interaction Detection), THAID (Theta AID), and CART (Classification and Regression Trees).

MM4XL's tool called Segmentation Tree tool applies Belson's original algorithm to a set of data that can be gathered, for instance, but not only, by means of a sample survey. Say, we were interested in identifying the demographic factors that most highly relate to the purchase of vitamin tablets, and a marketing research agency ran a sample survey for us. We can use the dataset for drawing a segmentation tree like the one below, which shows in pictorial form the number of people belonging to one segment (tree leaf) and which variables relate most to the use of vitamin tablets in that group. In our example it is the social class that generates the first split, followed by the geographic location where respondents reside. The size of the identified groups can be then inferred to the overall target population.