一种基于EFD的混合属性聚类算法  

Clustering algorithm for mixed attribute based on EFD

作　　者：王文庆 向孜瑞 WANG Wenqing;XIANG Zirui(School of Automation,Xi’an University of Posts and Telecommunications,Xi’an 710121,China;Internet of Things Application Technology Joint Demonstration Laboratory,Xi’an 710121,China)

机构地区：[1]西安邮电大学自动化学院,陕西西安710121 [2]物联网应用技术联合示范实验室,陕西西安710121

出　　处：《西安邮电大学学报》2024年第1期103-110,共8页Journal of Xi’an University of Posts and Telecommunications

基　　金：陕西省重点研发计划项目(2018ZDXM-GY-039)。

摘　　要：为了提高混合属性聚类效率,提出一种基于扩张翻转距离(Expand Flip Distance, EFD)的混合属性聚类算法。以信息熵及熵权法为基础,通过定义扩张属性和属性扩张量得到EFD,将其作为待聚类对象属性区分的依据,进行聚类对象的属性约简,最终对约简后的属性构建混合属性聚类模型,实现混合属性聚类。实验结果表明,所提算法获得的聚类谱系图和聚类结果均优于对比算法,验证了该算法的合理性和有效性。In order to improve the clustering efficiency of mixed attributes,a mixed-attribute clustering algorithm based on the expand flip distance(EFD)is proposed.Based on the information entropy and entropy weight method,the EFD is obtained by defining the expansion attribute and the attribute expansion amount,which is used as the basis for the attribute differentiation of the object to be clustered,and the attribute reduction of the clustered object is carried out,and finally the mixed attribute clustering model is constructed for the reduced attribute to realize the mixed attribute clustering.Experiment results show that the clustering pedigree map and the clustering results obtained by the proposed algorithm are better than those of the comparison algorithms,which verifies its rationality and effectiveness.

关 键 词：混合属性聚类 扩张属性 属性扩张量 扩张翻转距离 属性差异化 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]