Discovering Fuzzy Association Rules with Interest and Conviction Measures

摘要

Association rule mining forms an important research area in the field of data mining. The theory of fuzzy sets can be used over relational databases to discover useful, meaningful patterns. In this paper, we propose an algorithm to mine fuzzy association rules over relational databases using Interest and Conviction measures. In the present work, we introduce fuzzy interest and fuzzy conviction measures and eliminate the rules, which have negative correlation. The experiments are conducted on an insurance database using our approach. The presented approach is very useful and efficient when there are more infrequent itemsets in a database.