A Rough Set Based Approach for Ranking Decision Rules

In this paper we propose a new method for ranking decision rules generated from an information system. This process will reduce the overhead incurred in selecting appropriate rules for classification and hence speed up the decision making process. The algorithm proposed for rule ranking is based on discernibility matrix in Rough Set Theory. In this approach, rules generated from the given dataset using Apriori algorithm are considered as conditional attributes to construct a new decision table. From this decision table, degree of significance of each rule is calculated and rules are ranked according to this degree of significance. The algorithm is explained with the help of a test dataset. Further it is applied on a Learning Disability (LD) dataset consisting of signs and symptoms causing learning disability, which is collected from a local clinic handling learning disability in school aged children. The experiments on these datasets show that the new method is efficient and effective for ranking decision rules.