基于候选项集剪枝的Apriori算法的研究

Study of Apriori based on candidate itemsets pruning algorithm

在大数据挖掘中使用经典Apriori算法时,会产生大量的候选集,并需要扫描数据库中所有数据,使得其在实现过程中效率大大降低。为了提高Apriori算法的应用效率,通过建立线性链表记录每个事务中的项数,以达到事务压缩的目的;设置up值来提高1-项频繁集组合的门槛,发现支持度比较大的2-项候选集,以达到剪枝的目的;通过实验来设置up的取值范围,使最终产生的频繁项集的误差能够在接受的范围。实验证明提出的改进方法可以在找出绝大部分关联规则的同时,提高算法运行的效率。

Using classic Apriori algorithm in large data mining often generates a large number of candidate sets and needs to scan all of the data in the database,thus greatly reducing the implementation process efficiency.This paper presents an improved Apriori algorithm based on candidate itemsets pruning algorithm.To achieve the purpose of the transaction compression,a linear list was established to record each transaction; UP value was set to raise the threshold of 1-item frequent sets combination and 2-item candidate set that can achieve the purpose of pruning was found; the value range of UP was set through experiments and finally the error of frequent item sets is within the acceptable range.The experimental results showed that the new algorithm performance has been significantly improved.