基于事务权重与布尔矩阵的Apriori改进算法

An Improved Apriori Algorithm Based on Item Sets weights and Boolean Matrix

关联规则可在大量数据中找出事务的隐藏联系,其中,Apriori算法是关联规则分析中较为有效的方法。然而,经典Apriori算法需频繁扫描数据库和事务集,使在生成频繁项集的过程中产生大量候选项集。针对该问题,利用事务集对应权重和初始数据库映射形成的布尔矩阵,在经典Apriori算法的基础上,提出一种改进算法。数值算例结果表明,改进后的算法能较为明显地减少计算时间,从而提升经典Apriori算法效率。

Association rule mining can find the hidden relationship from a huge number of data,among which the Apriori algorithm is very effective.However,the classical Apriori algorithm needs to scan the database repeatedly,which leads to a large candidate sets when generating the frequent item sets.Aiming at this problem,based on the classical Apriori algorithm,we proposed an improved algorithm via Boolean matrix constructed by the weights of the related item sets and initial mapping of the database.Numerical experiments show that the proposed algorithm is more efficient than the classical Apriori algorithm.