Demographic Segmentation of Cosmetics Product Customers Based on CHAID Method

Abstract

Chi-square Automatic Interaction Detection (CHAID) is one of decision tree method with chi-square criteria that will build non-binary trees based on a relatively simple algorithm. In this method, nominal, ordinal, and continuous data type can be used. However, the continuous predictors were splitted into categories with approximately equal number of observations. This research used CHAID method on building SHP Inc demographic segmentation. The sampling was performed to obtain relatively balance categories to produce a model that predict each category. There were three explanatory variables having structural linkage toward number of transaction variable. With depth equals to three, there were nine segments related to gender, age, and transaction period variable. These segmentation provided an overview about customers shopping pattern. Therefore, a more targetted campaign could be implemented perfectly according to the segmet characteristics.