改进的关联规则算法  

Improved association rules algorithm

作　　者：刘扬[1] 莫红玉[1] 马垣[1] 

机构地区：[1]辽宁科技大学电子与信息工程学院,辽宁鞍山114051

出　　处：《计算机工程与设计》2010年第17期3876-3878,3906,共4页Computer Engineering and Design

摘　　要：通过对数据挖掘的经典Apriori算法和基于线性链表的关联规则挖掘算法进行研究,发现其中的不足——多次扫描数据且生成大量候选项集,增加了计数时间和内存空间。针对以上情况提出了基于候选项集分组的关联规则挖掘算法,该算法主要改进数据仅一次扫描和对候选项集进行分组计数,且动态创建候选项集的集合,有效地缩短了计数时间和占用的内存空间,使挖掘的效率更好更快。实验结果表明,该算法比文献[1]的算法效率更高。By the study on Apriori algorithm and association rules mining algorithm based on linear linker,there are many deficiencies such as multiple scanning datas and generate a large number of candidate item sets,that counting time and memory space are increased.For the above,an association rules mining algorithm based on candidate items using the packet is proposed.There are two major improvements： ① All data only needed one scan;② The method which is candidate itemsets dividing into groups and counting and which dynamically created candidate itemsets availably improved the counting time,and made the efficiency of data mining better and faster.The experiments show that the algorithm is more efficient than the one in the literature [1].

关 键 词：数据挖掘 线性链表 关联规则 候选项集分组 计数时间 内存空间 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]