基于知识粗糙度的混合变量决策树生成方法  

Construction of Hybrid Decision Tree Based on Knowledge Roughness

作　　者：路红梅[1] 胡学钢[1] 

机构地区：[1]合肥工业大学计算机与信息学院

出　　处：《计算机技术与发展》2008年第1期56-58,62,共4页Computer Technology and Development

基　　金：安徽省自然科学基金项目(2006kj091B)

摘　　要：单变量决策树难以反映信息系统属性间的关联作用,构造的决策树往往规模较大。多变量决策树能较好地反映属性间的关系,得到非常简单的决策树,但使构造的决策树难以理解。针对以上两种决策树特点,提出了基于知识粗糙度的混合变量决策树的构造方法,选择知识粗糙度较小的分类属性来构造决策树。实验结果表明,这是一种操作简单、效率很高的决策树生成方法。It is difficult for tmivariate decision tree to reflect the relationship of attributes, multivariate decision tree can resolve this problem preferably, the former produces big tree, the latter gains simple tree but difficult to explain. Aim to upwards points, in this paper, advance a knowledge roughness based approach to hybrid decision tree, select less knowledge roughness as tested attribute to construct decision tree. As a resuit, find this is a good approach with simple operation and higher efficiency.

关 键 词：粗糙集 知识粗糙度 单变量决策树 多变量决策树 混合变量决策树 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]