关联规则挖掘算法的改进

An Improved Data Mining Algorithm of Association Rule

为了提供一种更加准确高效的关联规则算法,在传统的Apriori算法的基础上引入分而治之的理念和加权的思想。先把数据库分成互不相交的块,根据需求分析从每一个块中产生用户感兴趣的子集,把所有的子集合并成挖掘对象,再利用普通的关联规则算法产生频繁项集,最后在该项集的基础上产生加权频繁项集。该算法基本上克服了传统Apriori算法的缺点,从而大大地提高了运算效率,最大限度解决了“项集生成瓶颈”问题,并且使得生成的关联规则更加科学、准确。

The aim is to provide a kind of the association rule that is more accurate and more effective. This algorithm builds based on the traditional Apriori algorithm. Firstly, the algorithm divides DB into mutually disjoint blocks, produces sub- set that users ore interested with accoding to requirement analysis from blocks, combines all the sub- set into dig object. Secondly, this algorithm produces frequent items using previous common association rules mining algorithm,then prnduces weighted itema from the previous items. The algorithm basically overcomes the shortcoming of the traditional Apriori algorithm, thus, can greatly improve operational efficiency,solve“the bottleneck of producing item- set”problem with superior limit, and make the association rule more scientific, more accuratc.