基于聚类矩阵的CM-Apriori算法研究  

A Research of CM-Apriori Algorithm Based on Cluster Matrixes

作　　者：赵凤荣[1] 刘铁英[1] 蔡斐[1] 赵永安[1] 李洋洋[1] 

机构地区：[1]内蒙古大学计算机学院,呼和浩特010021

出　　处：《内蒙古大学学报（自然科学版）》2008年第5期573-575,共3页Journal of Inner Mongolia University：Natural Science Edition

摘　　要：针对关联规则中Apriori算法的不足,提出了一种基于聚类矩阵的CM-Apriori新算法.该算法只需扫描事务数据库一次,就直接按事务项数生成聚类矩阵,每次只需对部分聚类矩阵进行运算,就可以生成频繁项集.这大大减少了扫描数据库的次数和计算成本,从而有效提高了算法的运算效率,并通过实例说明了它是一种有效的关联规则挖掘方法.To overcome the shortcomings of the Apriori algorithm,a new algorithm of CM- Apriori based on cluster matrix is proposed. This algorithm scans the transaction database only once and the cluster matrixes are created according to the item count. The frequent itemsets are generated by partial cluster matrixes. The CM-Apriori algorithm reduces the times of scanning transaction database and computation cost, thus improves the efficiency of association rule algorithm greatly. The example shows that this algorithm is an effective method of association rule mining .

关 键 词：关联规则 聚类矩阵 APRIORI算法 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]