外关联规则挖掘

根据项集内项的关联性,现有关联规则挖掘算法可分为正关联规则挖掘和负关联规则挖掘两大类,它们反映的是项集内频繁项之间的关联性。通过对实际数据集的分析发现,一个项集可以划分成若干子项集,子项集内的项有较高的相关性,而不同的子项集相关性则较低,这意味着每个子项集与其外部的一个潜在因子间存在着关联,由此本文提出了外关联规则的概念,并基于因子分析和主成分分析方法,提出了外关联规则挖掘算法(FAAR),从而将项集内的关联规则挖掘外推到子项集和潜在因子集之间的外关联规则挖掘,扩展和丰富了关联规则挖掘的应用,在此基础上还可进一步发现它所蕴涵的正、负关联规则。

Mining of Extra Association Rule

According to the correlation of items in a item set,current association rules can be divided into positive and negative association rules.It denotes the correlation among frequent items.By analyzing actual data sets a item set can be partitioned into several subsets.And items within a subset have a high correlation in comparison to one another.It means that every subset is correlated to a latent factor,so the concept of the extra association rule is presented.The algorithm of FAAR based on factor analysis(FA) and principal component analysis(PCA) is proposed.It can discover the extra association rule,positive and negative association rules.So the mining among item sets is extended to the mining of the extra association rule between the item set and the latent factor set,thus extending the application of the mining of the extra association rule.