聚类系数无显著性差异下的灰色综合聚类方法研究

发布时间：2018-06-18 14:37

本文选题：灰色聚类 + 聚类系数　；参考：《中国管理科学》2005年04期

【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。
[Abstract]:In the grey clustering evaluation analysis, when there is no significant difference in grey clustering coefficient, the existing grey clustering methods can not cluster the clustering objects accurately. However, there is no significant difference in clustering coefficient in practical research. Therefore, a new grey comprehensive clustering method is proposed in this paper. The concrete steps are as follows: firstly, the clustering coefficients of each clustering object are calculated and normalized, and then the comprehensive clustering coefficients of each grey class are calculated according to the function of the grey clustering coefficients in the clustering process. Finally, the clustering object is clustered according to the comprehensive clustering coefficient, and the grey class of the clustering object is determined. It is also proved that when the difference of clustering coefficients of clustering objects is greater than 1-2 / S, the clustering results obtained by the general grey clustering method and the grey comprehensive clustering method are exactly the same. Finally, taking the choice of the leading industry in the second industry of Jiangsu Province as an example, the paper makes an empirical analysis.
【作者单位】：南京航空航天大学经济与管理学院南京航空航天大学经济与管理学院南京航空航天大学经济与管理学院南京航空航天大学经济与管理学院
【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)
【分类号】：C931

【参考文献】