一个基于粗集的决策树规则提取算法  

A Rule Abstracting Algorithm of Decision Tree Based on Rough Set

作　　者：丁春荣[1] 李龙澍[1] 

机构地区：[1]安徽大学计算机科学与技术学院

出　　处：《计算机技术与发展》2007年第11期110-113,共4页Computer Technology and Development

基　　金：国家自然科学基金项目(60273043);安徽省自然科学基金项目(050420204);安徽省高校拔尖人才基金项目(05025102)

摘　　要：决策树是数据挖掘任务中分类的常用方法。在构造决策树的过程中,分离属性的选择标准直接影响到分类的效果,传统的决策树算法往往是基于信息论度量的。基于粗糙集的理论提出了一种基于属性重要度和依赖度为属性选择标准的决策树规则提取算法。使用该算法,能提取出明确的分类规则,比传统的ID3算法结构简单,并且能提高分类效率。The decision tree is a usual method of classification in data mining. In the process of constructing a decision tree, the criteria of selecting attributes to split will influence the efficiency of classification directly. The decision tree algorithm traditionally is based on information theory measure. Presented a new algorithm for classification rules extraction by choosing attributes of importance of attributes and dependance based on rough set. Using this algorithm,can extract crisp rules from clarification information system. Compared with the traditional ID3 algorithm, it＇s simpler in the structure, and can improve the efficiency of classification.

关 键 词：数据挖掘 粗糙集 决策树 属性约简 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]