基于频繁项目集链式存储方法的关联规则算法  

Association rule algorithm based on chain storage method of frequent itemsets

作　　者：尹士闪[1] 马增强[2] 毛晚堆[3] 

机构地区：[1]石家庄铁道大学教务处,河北石家庄050043 [2]石家庄铁道大学电气与电子工程学院,河北石家庄050043 [3]石家庄铁道大学工程训练中心,河北石家庄050043

出　　处：《计算机工程与设计》2012年第3期1002-1007,共6页Computer Engineering and Design

摘　　要：为了提高经典关联规则Apriori算法的挖掘效率,针对Apriori算法的瓶颈问题,提出了一种链式结构存储频繁项目集并生成最大频繁项目集的关联规则算法。该算法采用比特向量方式存储事务,生成频繁项目集的同时,把包含此频繁项目的事务作为链表连接到频繁项目之后,生成最大频繁项目集。该算法能够减小扫描事物数据库的次数和生成候选项目集的数量,从而减少了生成最大频繁项目集的时间,实验结果表明,该算法提高了运算效率。In order to improve the Apriori algorithm mining efficiency, a mining algorithm is presented to use a chain structure to store frequent itemsets and generate maximum frequent itemsets for the bottleneck problem of the classic Apriori algorithm. In this algorithm, the transaction is stored with a bit vector, when frequent itemset is generated, the transaction including this frequent itemset as a linked list connect to frequent itemset, at end, it make maximum frequent itemsets. This algorithm reduce the number of scanning transaction database and the amount of generating candidate itemsets, thus reduce time of generated maximum frequent itemsets, experimental results show that this algorithm improve the operation efficiency.

关 键 词：数据挖掘 APRIORI算法 候选集 频繁项目集 关联规则 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]