一种新的改进的Apriori算法

A New Improved Apriori Algorithm

本文通过对关联规则挖掘算法Apriori算法的分析和研究,指出了其在具体应用中存在的主要问题。提出与以往不同的改进策略:在约简数据库事务的同时,生成频繁项目集和保存具有非频繁子集候选项目集的项集,在提高频繁项目集即关联规则生成效率的同时,进一步减少了对候选项目集的重复验证。最后将改进的Apriori算法应用到一个Web交叉销售系统,并和经典的Apriori算法进行了比较,取得了较好的效果。

The paper analysed the problems of classic Apriori algorithm of association rules in practical applications by research on the algorithm , and proposed a new method which improves the classic Apriori algorithm： the new method reduces the transactions of database, as well as it forms the frequent itemsets and saves candidate itemsets that are not frequent.The result showed that it not only promotes the efficiency in generation of frequent itemsets,but also reduces the verification that is repeated in generation of candidate itemsets.At last the improved algorithm was applied in a web cross-selling system and compared with the classic Apriori algorithm,and a good effect was achieved.