Subgroup discovery

The discovery of (interesting) subgroups has a high practical relevance in all domains of science or business. For example, consider statements such as: ”the unemployment rate is above average for young men with a low educational level”, ”smokers with a positive family history are at a significantly higher risk for coronary heart disease”, or ”single males living in rural areas do rarely take out a life policy”. Subgroup discovery is well suited for finding such dependencies, i.e., discovering relations between a dependent and (several) independent variables, for inductive and explorative data analysis tasks. Then, the discovered subgroups can be applied by the analyst, e.g., for decision support. This article aims to give a first idea of subgroup discovery. We introduce the subgroup discovery task and discuss the setting and the issues concerning the search process. Then, we give an outlook on further research directions.