Provide A New Approach for Mining Fuzzy Association Rules using Apriori Algorithm

Abstract

Association rules mining is one of the most popular data mining models. Minimum-support is used in association rules mining algorithms, like Apriori, FP-Growth, Eclat and etc. One problem Apriori algorithm and other algorithms in the field association rules mining, this is user must determine the threshold minimum-support. Suppose that the user wants to apply Apriori algorithm on a database with millions of transactions, definitely user can not have the necessary knowledge about all the transactions in the database, and therefore would not be able to determine an appropriate threshold. In this paper, using averaging techniques, we propose a method in which Apriori algorithm would specify the minimum support in a fully automated manner. Our goal in this paper improved algorithm Apriori, to achieve it, initially will try to use fuzzy logic to distribute data in different clusters, and then we try to introduce the user the most appropriate threshold automatically. The results show that this approach causes the any rule which can be interesting will not be lost and also any rule that is useless cannot be extracted. The simulation results on a real example show that our approach works better than the classic algorithms.Keywords: Apriori Algorithm, Association Rules, Data Mining, Frequent Patterns, Fuzzy Logic & C-Means Clustering Support