Strongest Association Rules Mining for Efficient Applications
Jie Li, Yong Dong Xu, Yunfeng Wang, Chao‐Hsien Chu · 2007
Rule explosion has become an important problem of association rules mining, as conventional mining algorithms often produce too many rules for decision makers to digest. In this paper, the notion of strongest association rules (SAR) is proposed for representing all association information with fewer rules, and a matrix-based algorithm is developed for mining SAR set. Our experiments show that the number of SAR is about 26% of the number of all rules in average, and the number does not monotonously increase with a smaller minimal confidence.