Evaluating class association rules using Genetic Relation Programming

The number of association rules generated during the data mining process is generally very large, that is, an association rule mining algorithm could generate thousands or millions of rules. However, only a small number of rules are likely to be of any interest to the domain expert analyzing the data, i.e., many of the rules are either irrelevant or obvious. Therefore, techniques for evaluating the relevance and usefulness of discovered patterns are required. The aim of this paper is to propose a new method for evaluating the relevance and usefulness of discovered association rules by reducing the number of rules extracted using an evolutionary method named genetic relation programming (GRP). The algorithm evaluates the relationships between the rules at each generation using a specific measure of distance and gives the best set of rules at the final generation. The efficiency of the proposed method is compared with other conventional methods and it is clarified that the proposed method shows comparable accuracy with others.