聚类系数无显著性差异下的灰色综合聚类方法研究  

Study on the Integrated Grey Clustering Method under the Clustering Coefficient with Non-Distinguished Difference

作　　者：党耀国[1] 刘思峰[1] 刘斌[1] 翟振杰[1] 

机构地区：[1]南京航空航天大学经济与管理学院,江苏省南京市210016

出　　处：《中国管理科学》2005年第4期69-73,共5页Chinese Journal of Management Science

基　　金：国家自然科学基金资助项目(70473037);国家教育部博士点基金资助(20020287001);江苏省自然科学基金重点项目(BK2003211);南京航空航天大学博士创新基金资助项目(019004)

摘　　要：在灰色聚类评估分析中,当灰色聚类系数无显著性差异时,按照已有的灰色聚类方法无法对聚类对象进行准确的聚类,而在实际研究中经常会遇到聚类系数无显著性差异这类问题。因此本文提出了一种新的灰色综合聚类方法。具体步骤是:首先计算各聚类对象的聚类系数,并对其进行归一化处理;再根据对象中每一灰类的灰色聚类系数在聚类过程中的作用,计算聚类对象的综合聚类系数;最后根据综合聚类系数对聚类对象进行聚类,确定聚类对象应属的灰类。并且证明了当聚类对象的聚类系数差异大于1-2/S时,一般灰色聚类方法与灰色综合聚类方法所得聚类结果完全相同。最后,以江苏省第二产业内部主导产业选择为例进行了实证分析。Through analyzing in depth the present grey cluster model, we think they can be used to exact cluster with respect to cluster objectives when there are not distinguished differences between grey cluster coefficients, which often come forth in practical economic phenomena. Based on the present theories on grey cluster, a new grey cluster evaluation method is developed. Idiographic processes are the following：First, compute cluster coefficients of every cluster objectives and transform then into non- dimensional sequences;Second, compute the integrated cluster coefficients of every cluster objective based on the action of every cluster coefficient in the cluster process; Third, determine grey class of every cluster objective based on the integrated cluster coefficients. Furthermore, we prove that the result of our model and present model are equivalent when the distinguished difference coefficients are larger than 1 -2/S. Last, we illustrate the integrated cluster method with the selecting dominant industries of Jiangsu＇s second industry.

关 键 词：灰色聚类 聚类系数 显著性差异 

分 类 号：N94[自然科学总论—系统科学]