Significant rule power factor: An algorithm for new interest measure

Abstract

In the process of knowledge discovery, a large number of patterns are generated. It is therefore infeasible for an expert to provide his opinion considering this vast amount of patterns. One of the methods that are useful to overcome this problem is association rule interestingness measures. Interest measure essentially helps to fetch the data of interest in data mining. The data of interest completely depends on the competitive business requirements. Therefore, many interest measures exist; a few are support, confidence, lift, leverage. In this paper, we are proposing a new interest measure which is more informative and accurate when compared with many existing interest measures. We have used WEKA, JEDIT, ANT open source tools to conduct experiments. Through experiments, it has been observed that our proposed interest measure can fetch more important rules than many of the existing interest measures.