基于CMAR算法的水平加权多分类关联规则挖掘  

MINING HORIZONTAL WEIGHTED MULTIPLE CLASS-ASSOCIATION RULES BASED ON ALGORITHM CMAR

作　　者：尹辉[1] 李振军[2] 尹政[3] 

机构地区：[1]山东司法警官职业学院信息中心,济南250014 [2]济南市商业银行科技部,济南250001 [3]山东省泰安市医疗器械维修管理站,山东泰安271000

出　　处：《山东师范大学学报（自然科学版）》2008年第2期18-21,共4页Journal of Shandong Normal University(Natural Science)

摘　　要：关联分类是数据挖掘中一种新的分类方法,它将关联规则挖掘和分类进行了算法集成.但在实际应用中,用户会比较倾向于自己最感兴趣或认为最重要的那部分属性,因此有必要加强这些属性对规则的影响,同时减弱另一些用户兴趣不大或认为不重要的属性对规则的影响.为此,本文提出了水平加权关联规则的问题,并结合CMAR算法,加以改进给出了关于该问题的解决方案及有效算法NewCMAR,实验表明,修改后的NewCMAR算法正确并符合实际,有实用价值.Classification based on Multiple Association Rules is a new method. It integrated association roles algorithm with classification algorithm. However, in real applications, users are more inclined to such items as they are most interested in or feel most important about. So, it＇ s necessary to emphasize the affection of such items and reduce the affection of other items to rules. Therefore, this paper proposes the problem of mining horizontal weighted association rules, advances the algorithm CMAR, and presents an effective algorithm NEW- CMAR to handle the problem of mining horizontal weighted association ntles.

关 键 词：数据挖掘 关联规则 关联分类 水平加权 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]