The Fuzzy Frequent Pattern Tree

A significant data mining issue is the effective discovery of association rules. The extraction of association rules faces the problem of the combinatorial explosion of the search space, and the loss of information by the discretization of values. The first problem is confronted effectively by the Frequent Pattern Tree approach of [10]. This approach avoids the candidate generation phase of Apriori like algorithms. But, the discretization of the values of the attributes (i.e. the "items") at the basic Frequent Pattern Tree approach implies a loss of information. This loss usually either deteriorates significantly the results, or constitues them completely intolerable. This work extends appropriately the Frequent Pattern Tree approach in the fuzzy domain. The presented Fuzzy Frequent Pattern Tree retains the efficiency of the crisp Frequent Pattern Tree, while at the same time the careful updating of the fuzzy sets at all the phases of the algorithm tries to preserve most of the original information at the data set. The paper presents an application of the Fuzzy Frequent Pattern Tree approach to the difficult problem of the discovery of fuzzy association rules between genes from massive gene expression measurements. keywords: Association Rules, Fuzzy Association Rules, Depth-First Search, Frequent Pattern Tree, Data Mining, Gene Expression Analysis