一种改进的加权关联规则挖掘算法

摘要

The weighted association rule mining algorithm-New-Apriori which is based on the classic Apriori algorithm has three problems : the repeat scanning on database is needed, the weight definition is not logical and the basic properties of Apriori no longer come into existence due to the introduction of weight. In order to solve these problems, a method that combines matrix and weighted association rules is adopted. The feature of this improved algorithm is that the only one scanning on database is needed. The algorithm does not ignore the items which appears with low frequency but high weight value, and what's more, K support expectation is introduced as the reference of pruning step.%基于经典Apriori算法的加权关联规则挖掘算法New-Apriori存在3个问题,需要多次扫描数据库,权值定义不合理和权值的引入导致Apriori基本性质不再成立.为了解决这些问题,采用将矩阵和加权关联规则算法相结合的方法,该改进算法具有只需要扫描一次数据库,权位的定义既考虑到了权重大的项目,也没有忽略频繁出现而权重小的项目,并引入K-项支持期望作为减枝的依据的特点.