An Approach to Finding Optimized Correlated Association Rules
Huang Hou · Chinese Journal of Computers · 2006
This paper proposes a framework based on correlation.In this framework,residual(analysis) is used to determine whether two itemsets are independent of each other.The measure can help us find really correlated rules instead of concurrent ones.In addition,negatively correlated rules also can be found with residual analysis.Moreover,the support and confidence thresholds not have to be specified in the framework.To improve the quality of rules,the authors employ genetic algorithm to find optimized rules.By running the algorithm on synthetic and real datasets,the author argue that the algorithm outperform over Apriori-like algorithm on the interestingness of rules.