论文部分内容阅读
现实生活中多数聚类对象具有多元异构不确定性特征,表现为对象聚类指标体系异构化以及对象信息具有多元不确定性特点,而现有的不确定性多属性聚类决策方法对此类对象的聚类研究有着局限性.因此针对聚类问题,首先根据聚类对象多元不确定性信息的特点,提出广义区间灰数的概念,证明了多元不确定性信息可统一用广义区间灰数进行表征,然后结合极大熵思想,构建基于多元异构不确定性案例学习的广义区间灰数熵权配置模型,通过对对象相关的历史案例进行充分学习,测算各层指标的广义区间灰数熵权,以此确定各指标的聚类权重,再结合广义区间灰数的白化权函数对对象的新案例进行聚类分析.最后通过案例研究,证明了本文聚类模型的合理性与可行性.
Most of the clustering objects in real life have the features of multivariate heterogeneity, such as the isomerization of the object clustering index system and the multi-element uncertainty of the object information. However, the existing multi-attribute clustering decision-making method with uncertainties For the clustering problem, firstly, the concept of generalized interval gray numbers is proposed according to the characteristics of multi-uncertainty information of clustering objects, and the multi-uncertainty information can be unified by generalized interval Gray number, and then combined with the idea of maximum entropy to build a generalized interval gray number entropy weight configuration model based on case study of multiple heterogeneous uncertainties, through the full study of the historical cases related to the object, the generalized interval Gray number entropy to determine the clustering weight of each index, and then combined with the whitening weight function of generalized interval gray number clustering analysis of new cases of the object.Finally, through case studies, it is proved that the clustering model is reasonable and feasibility.