粗糙集理论对C4．5算法的优化研究  

Research＇of an Optimized C4.5 Algorithm Based on Rough Theory

作　　者：向卓元[1] 张蕾[1] XIANG Zhuo-yuan, ZHANG Lei （Information and Safety Engineering Department, Zhongnan University of Economics and Law,Wuhan 430081 ,China）

机构地区：[1]中南财经政法大学安全科学与管理学院,湖北武汉430081

出　　处：《电脑知识与技术》2012年第6期3782-3785,共4页Computer Knowledge and Technology

摘　　要：该文提出了一种将粗糙集理论和C4．5决策树算法结合在一起的一种改进算法。该算法利用粗糙集理论中的属性的约简功能首先将初始数据进行规约，然后再将规约后的数据作为C4．5的输入进而构造出决策树。通过粗糙集的属性约简，提高了训练数据表达的清晰度，也降低了无关属性对构造决策树的影响，从而减小了决策树的大小，提高了效率，同时也提高了结果的准确率。This paper proposes an improved algorithm based on the rough set theory and C4.5 decision tree. The algorithm uses rough set theory to reduce the attributes in the decision system, and uses the reduced data as the input ofC4.5 algorithm to construct a decision tree. The new algorithm improves the clarity of training data, and also reduces the influence of irrelevant attributes, therefore, the size of decision tree can be reduced and the accuracy of the result can be improved.

关 键 词：粗糙集 属性约简 决策树 C4.5 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论;自动化与计算机技术—计算机科学与技术]