Exploring the Data Using Extended Association Rule Network

Abstract

In this paper, we presented the Extended Association Rule Network (ExARN) to structure, prune and analyze a set of association rules, aiming to build hypothesis candidates. The ExARN extends the ARN, proposed by [2], allowing a more complete exploration. We validate the ExARN using two databases: contact lenses and hayes-roth, both available online for download. The results were validated by comparing the ExARN to the conventional ARN and also by comparing the results with a decision tree algorithms. The approach presented promising results, showing its capability to explain a set of objective items, aiding the user on the hypothesis building.