Significant Support (SISU): A New Interest Measure in Association Rule Mining

Abstract

In machine learning, association rule mining is a field with immense opportunity to explore relationships among various attributes and item-sets. However, in Association rule mining, statistically it is the interest measure which play the crucial role to decide these relationships. There exist various types of interest measures based upon the business needs and problem statements. In this paper, a novel interest measure has been proposed to decide the overall importance of an association rule. Statistical comparisons and experimental results have also been embedded to support its potential.