A cluster-based genetic-fuzzy mining approach for items with multiple minimum supports

Abstract

In the past, we proposed an algorithm for extracting appropriate multiple minimum support values, membership functions and fuzzy association rules form quantitative transactions. The evaluation process might take a lot of time, especially when the database to be scanned could not totally fed into main memory. In this paper, an enhanced approach, called the Cluster-based Genetic-Fuzzy mining approach for items with Multiple Minimum Supports (CGFMMS), is thus proposed to speed up the evaluation process and keep nearly the same quality of solutions as the previous one. Experimental results also show the effectiveness and the efficiency of the proposed approach. © 2008 Springer-Verlag Berlin Heidelberg.