摘要
连续属性的处理是当前分类规则中一个热点研究问题。以往的算法往往是建立在离散化过程的基础上进行的,然而,该类方法不但会破坏数据中信息的精度,同时也使得概念的转换十分困难。文章在分析了以往算法的基础上,提出了利用包含度和蕴含度的方法进行连续属性的分类规则学习,并对该种方法的属性约简问题进行了讨论。可以看出,通过该文的研究较好地解决了数据精度和动态概念挖掘的问题,利用包含度和蕴含度的方法是一个十分有价值的研究方向。
Mining classification rules in database with continuous attributes is a focus of recent research,and the past algorithms usually are based on discretization.However,these algorithms not only decrease the information accuracy of the original data but also make the change of concept become difficult.On the basis of analyzing past algorithms ,this paper proposes a method that uses inclusion and implication degree to learn classification rules from database with con-tinuous attributes and discusses the reduction in our method.It could get that the research gives a satisfying solution about the problems of the accuracy of information and dynamic conception data mining.The use of inclusion and impli-cation degree is profitable.
出处
《计算机工程与应用》
CSCD
北大核心
2003年第32期18-21,共4页
Computer Engineering and Applications
基金
国家自然基金资助(编号:60172037)
关键词
连续属性
分类规则
动态概念
包含度
蕴含度
属性约简
Continuous Attributes,Classification,Dynamic concept ,Inclusion Degree,Implication Degree,Relative Reduct