【作者】 彭玉兵；

【导师】 吴根秀；

【作者基本信息】 江西师范大学 ， 基础数学， 2005， 硕士

【摘要】 自Z.Pawlak从1982年提出粗糙集理论以来,该理论的应用已引起了人们的广泛关注。在不一致信息系统研究中,Ziarko提出了一种归类质量,进而在这种归类质量的基础上提出了近似约简,而由这种近似约简形成的决策规则对一部分个体的归类会产生较大的偏差。本文的工作之一是先将不一致信息系统进行约简,再形成近似决策规则,这样能一定程度上缓解偏差。众所周知,无论是决策树分类器,还是粗糙集中的信息系统,都有一个共同的缺陷:都会由于训练样本分布不均匀可能引起分类规则的不完全,导致分类出现“盲区”。北师大研究生李凤通过引入决策树模糊化及分支激活度的概念解决了决策树的“盲区”问题,本文的另一工作就是将这一思想引入到粗糙集中来,使落入信息系统“盲区”的个体不但能归类,而且有其合理性。更多还原

【Abstract】 Since rough set theory was put forward by Z.Pawlak in 1982,it have attracted a considerable deal of attention. In the research of the inconsistent information system, Ziarko put forward the measure of quality of classification and introduced approximate reducts. Such decision rules would produce a big deviation on a part of objects. One main aim of this paper is to reduce the information system, and form approximate decision rules then. It can ameliorate the deviation. Not only decision tree classifications,but also information systems are incomplete due to the distribution of samples and the tree pruning,which can produce "blind region" during classification.Li Feng introduced the fuzzification of decision tree and proposed an adaptive classification algorithm.Another main aim of this paper is to introduce the idea into information systems.更多还原