基于相容粗糙集技术的连续值属性决策树归纳  

Induction of Decision Tree for Continuous-valued Attributes Based on Tolerance Rough Sets Technique

作　　者：翟俊海[1,2] 翟梦尧[3] 李胜杰[1,2] 

机构地区：[1]河北大学数学与计算机学院,保定071002 [2]河北省机器学习与计算智能重点实验室,保定071002 [3]河北大学工商学院,保定071002

出　　处：《计算机科学》2012年第11期183-186,共4页Computer Science

基　　金：国家自然科学基金项目(61170040);河北省自然科学基金项目(F2010000323;F2011201063;F2012201023);河北大学自然科学基金项目(2011-228)资助

摘　　要：决策树是常用的数据挖掘方法,扩展属性的选择是决策树归纳的核心问题。基于离散化方法的连续值决策树归纳在选择扩展属性时,需要度量每一个条件属性的每一个割点的分类不确定性,并通过这些割点的不确定性选择扩展属性,其计算时间复杂度高。针对这一问题,提出了一种基于相容粗糙集技术的连续值属性决策树归纳方法。该方法首先利用相容粗糙集技术选择扩展属性,然后找出该属性的最优割点,分割样例集并递归地构建决策树。从理论上分析了该算法的计算时间复杂度,并在多个数据集上进行了实验。实验结果及对实验结果的统计分析均表明,提出的方法在计算复杂度和分类精度方面均优于其他相关方法。Decision tree is a popular data mining method,and it is a crucial problem to select expanded attributes in the induction of decision tree.The uncertainty of each cut of each continuous-valued attributes needs to be measured during the selection of expanded attributes for induction of decision tree based on discretion method,and the computational time complexity is very high.In order to deal with this problem,a method of induction of decision tree for continuous-valued attributes based on tolerance rough sets technique was proposed.The method consists of three stages.First expanded attributes are selected with tolerance rough sets technique,and then the optimal cut of the expanded attribute is found,and the sample set is partitioned by the optimal cut,finally the decision tree can be generated recursively.We analysed the computational time complexity of the algorithm in theory and conducted some experiments on multiple database.The experimental results and the statistical analysis of the results demonstrate that the proposed method outperforms other related methods in terms of computational complexity and classification accuracy.

关 键 词：相容粗糙集 决策树 扩展属性 割点 统计分析 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]