An Approach to Classification Based on Fuzzy Association Rules

Abstract

Classification based on association rules is considered to be effective and advantageous in many cases. However, the sharp boundary problem in association rules mining with numerical data may lead to semantics retortion of discovered rules, which may further disturb the understandability, even the accuracy of classification. This paper aims at proposing an associative classification approach, namely Fuzzy Association Rules Classification (FARC), where fuzzy logic is used in partitioning the domains of numerical data items, giving rise to fuzzy association rules for classification. In doing so, two measures, pseudo support and pseudo confidence, as well as the notion of minimal equivalence set (MESet), are introduced, along with extensions to the corresponding mining algorithms. The experimental results revealed that FARC generated fewer rules than the traditional CBA approach without loss of accuracy.