An Application of Subgroup Discovery Algorithm on the Case of Decentralization and Quality of Governance in EU

Abstract

This paper analyses a statistical relationship between the decentralization of the EU countries and the quality of their governance. The degree of decentralization is measured from a fiscal and political point of view, and the quality of governance by multiple indicators and citizen opinions. The paper presents a subgroup discovery algorithm which is capable of analysing two sets of several variables, and uses it for the analysis of EU countries. The paper is one of the first to use the data mining methods from the social sciences domain. The used algorithm has discovered some interesting patterns which show a desired relationship. We have discovered that the proportion of public sector employees is one of the most important indicators, which strongly correlates with the degree of trust in the European and national institutions, the government effectiveness and the perception of corruption.