基于分辨关系的决策树构建算法  

Algorithm of Constructing Decision Tree Based on Definite Relation

作　　者：刘军[1] 

机构地区：[1]南京工业大学电子与信息工程学院,江苏南京210009

出　　处：《微电子学与计算机》2012年第1期109-112,116,共5页Microelectronics & Computer

基　　金：国家自然科学基金项目(60673185);教育部留学回国人员科研启动基金项目(200711108)

摘　　要：针对完全以核属性为首选构建决策树算法的不足,本文以粗糙集的分辨关系为基础,提出以决策属性的分辨量和分辨类别两个参数确定条件属性的重要性,择优自顶向下构造决策树.该算法的优点在于算法简洁有效和时空复杂度低.实例分析的结果表明,该方法能获得更为简洁有效的决策树.The algorithm of decision tree is achieved by the relative core of attributes,which is incomplete.The algorithm is proposed in this paper on the definite relation of rough set,which chooses important attribute by the definite count and the classifying value,and makes decision tree from top to bottom with the attribute.The benefit of the algorithm is simple in calculation and is effectively used in the classification.The experiments and comparison show that the algorithm provides more precise and simple decision tree.

关 键 词：粗糙集 决策树 分辨关系 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]