拓展粗糙集理论及其在工业项目风险评估中的应用

发布时间：2019-03-21 12:29

【摘要】：在偏好决策问题中,由于不完备信息普遍存在于应用环境中,相关学者提出了多种优势关系来应对这一问题。同时,与之相对应的粗糙集模型也得到了拓展。但在实际应用环境中,仍然存在许多改进之处。本文通过对影响误分类因素的分析,寻找不完备信息下粗糙集模型的改进点,从而提高粗糙集模型的分类性能,并将改进后的粗糙集模型应用到A公司数据分析中。实际应用表明,新的粗糙集模型分类性能更优。以下是本文的主要工作:权重优势关系的定义。在广义扩展优势关系的基础之上,考虑了不同属性重要度不同对分类的影响。提出了不完备信息系统中基于粗糙集的权重设定方法,引入一个阈值来对优势关系进行限定,并给出了阈值的选取方法。在此基础上提出了权重优势关系及其粗糙集模型。通过实例的运算验证,相对于广义扩展优势关系而言权重优势关系使得粗糙集模型具有更好的分类性能。灰度优势关系的定义。通过分析造成优势二元关系不确定性的影响因素,以可比较信息的多少、条件属性的权重大小、以及属性之间差异程度为三个维度,构建两对象对比时确定性程度的灰度度量。定义了优势关系中差异系数以及灰度的概念,并以此为基础建立灰度优势关系,并提出相应的粗糙集模型。通过理论证明以及实例运算,证明了灰度优势粗糙集模型的实用性。在工业项目数据中的应用分析。对现有的A公司工业项目数据进行分析、量化,应用前面提出的各属性权重的设定方法和灰度优势关系的粗糙集模型对工业项目数据进行建模。并对其决策规则进行挖掘,为工业项目的选择以及投资提供决策依据。
[Abstract]:In the problem of preference decision-making, because incomplete information exists widely in the application environment, relevant scholars have proposed a variety of advantages to deal with this problem. At the same time, the corresponding rough set model is extended. However, in the practical application environment, there are still many improvements. In this paper, through the analysis of the factors affecting the misclassification, the improved points of rough set model under incomplete information are found to improve the classification performance of the rough set model, and the improved rough set model is applied to the data analysis of Company A. The practical application shows that the classification performance of the new rough set model is better. The following is the main work of this paper: the definition of weight superiority relationship. On the basis of generalized extended dominance relation, the influence of different attribute importance on classification is considered. The weight setting method based on rough set in incomplete information system is proposed. A threshold value is introduced to limit the superiority relation, and the selection method of threshold is given. On this basis, the weight dominance relation and its rough set model are proposed. The calculation results show that the rough set model has better classification performance than the generalized extended dominance relation. Definition of gray-scale dominance relationship. By analyzing the influencing factors of the uncertainty of the dominant binary relationship, the gray scale measurement of the certainty degree when the two objects are compared is constructed according to the number of comparable information, the weight size of the conditional attribute and the degree of difference between the attributes in order to construct the gray scale measure of the degree of certainty when the two objects are compared. In this paper, the concept of difference coefficient and gray scale in superiority relation is defined, and based on it, the grey superiority relation is established, and the corresponding rough set model is put forward. The practicability of the gray-scale dominant rough set model is proved by theoretical proof and practical operation. Application analysis in industrial project data. This paper analyzes and quantifies the existing industrial project data of Company A, and models the industrial project data by using the method of setting the weight of each attribute and the rough set model of gray-level superiority relation. And mining its decision-making rules to provide the decision-making basis for the choice of industrial projects and investment.
【学位授予单位】：南京航空航天大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F224;F272

【参考文献】