摘要
属性重要度与排序方法是多粒度粗糙集研究的一个重要内容。在现有属性重要度的基础上,本文对属性重要度进行改进,提出了多粒度下信息粒的权重与条件属性的权重的概念,并给出了一种数据预处理的方法。与传统方法相比,它能有效克服数据差别不大的数值被划分到不同类中去的问题。同时,基于证据理论提出利用类概率函数的加权排序方法,此方法也可用于多元对象的排序问题。并用一些实例说明了此方法的实用性和有效性。
Attribute importance and sorting method is an important part of the study of multi granularity rough sets. Based on the existing attribute importance, this paper improves it and puts forward the concept of the weight of information granules and the weight of conditional attributes under multiple granularity. And a method of data preprocessing is given. Compared with traditional methods, this method can effectively overcome the problem of data with little difference being divided into different classes. Meanwhile, a weighted ranking method using class probability function based on evidence theory is proposed. This method can also be used for the sorting of multiple objects. Some examples are given to illustrate the practicability and effectiveness of this method.
作者
蔡明礼
冯涛
王荣欣
CAI Mingli;FENG Tao;WANG Rongxin(College of Science of Hebei University of Science and Technology, University, Hebei Shijiazhuang, 050018, China)
出处
《数码设计》
2018年第4期69-76,共8页
Peak Data Science
基金
国家自然科学基金(61573127).河北省“三三三人才工程”培养研究项目(A2017002112)。
关键词
属性重要度
证据理论
条件属性权重
信息粒权重
Attribute importance degree
Evidence theory
Condition attribute weight
Information granule weight
