计算机工程与应用 ›› 2009, Vol. 45 ›› Issue (8): 141-142.DOI: 10.3778/j.issn.1002-8331.2009.08.043

• 数据库、信号与信息处理 • 上一篇    下一篇

引入影响度的关联规则衡量标准

陈建军,高玉斌   

  1. 中北大学 现代教育技术与信息中心,太原 030051
  • 收稿日期:2008-01-23 修回日期:2008-04-23 出版日期:2009-03-11 发布日期:2009-03-11
  • 通讯作者: 陈建军

Evaluation criterion for association rules with influence degree

CHEN Jian-jun,GAO Yu-bin   

  1. Department of Information Technology,North University of China,Taiyuan 030051,China
  • Received:2008-01-23 Revised:2008-04-23 Online:2009-03-11 Published:2009-03-11
  • Contact: CHEN Jian-jun

摘要: 关联规则挖掘是数据挖掘中重要的研究课题,而其中一个重要问题是对挖掘出的规则的感兴趣程度的评估。分析并讨论了传统支持度-置信度框架、相关度和有效度等衡量标准的不足,并将T检验思想引入到关联规则的衡量中,提出了一种新的关联规则衡量标准-影响度。实验结果表明,在传统挖掘方法的基础上引入影响度,可以有效克服现有衡量标准的一些不足,减少冗余规则的产生。

关键词: 关联规则, T检验, 影响度, 衡量标准

Abstract: Discovering association rules is one of more important tasks in data mining.One of the important problems is the evaluation of interestingness for the discovered rules.Authors analyze and discuss the shortage of the support-confidence framework,the correlativity and the validity.With introducing T-Testing,authors propose the Effect that is a new evaluation criterion for association rules.The experimental result shows that introducing effect based on common approach to association rules mining can effectively overcome the shortage of the existing evaluation criterion for association rules,and reduce the creation of redundant rules.

Key words: association rule, T-Testing, influence degree, evaluation criterion