Ranking interesting subgroups

Abstract

Subgroup discovery is the task of identifying the top k patterns in a database with most significant deviation in the distribution of a target attribute Y. Subgroup discovery is a popular approach for identifying interesting patterns in data, because it combines statistical significance with an understandable representation of patterns as a logical formula. However, it is often a problem that some subgroups, even if they are statistically highly significant, are not interesting to the user. We present an approach based on the work on ranking Support Vector Machines that ranks subgroups with respect to the user's concept of interestingness, and finds more interesting subgroups. This approach can significantly increase the quality of the subgroups.