团粒结构分析法:一种复杂数据分析方法  

Granule Structure Analysis for Complex Data

作　　者：姜懋 张永光[2] 刘卓军[2] JIANG Mao;ZHANG Yong-guang;LIU Zhuo-jun(University of Chinese Academy of Science,Academy of Mathematics and Systems Science,Chinese Academy of Sciences,Beijing 100190,China;Academy of Mathematics and Systems Science,Chinese Academy of Sciences,Beijing 100190,China)

机构地区：[1]中国科学院大学,中国科学院数学与系统科学研究院,北京100190 [2]中国科学院数学与系统科学研究院,北京100190

出　　处：《数学的实践与认识》2024年第2期143-151,共9页Mathematics in Practice and Theory

摘　　要：数据里变量之间存在复杂联系,传统的数理统计方法已经不能解决问题,很多实际问题对数据处理提出了更高的要求.针对具有维度高,变量之间关联复杂,群组效应显著等特征的复杂数据提出了一个新的复杂数据处理方案:通过分析数据各变量之间的关联关系,找出具有群组效应的若干变量构成的变量簇,称其为团粒.为了有效地发现团粒,还提出了GC算法,用以获取若干具有群组效应的变量组.在发现团粒以后,通过分析团粒内部变量之间的相互关联,得到了反映团粒特征的内核变量.并通过实例分析说明该方法能有效地分析复杂数据变量之间的关联性.Due to the complex relationship between indicators in data,traditional mathematical statistical methods can no longer meet the demand,and many practical problems put forward higher requirements for data processing.In order to analyze those characteristics of complex data such as its high dimensions and indicators correlation between complex,group effect,this paper put forward its own remarkable complex data processing scheme:through the analysis of the correlation of data between various indicators,find out the several variables which have effect of group composition variables cluster,we call it for granule,in order to effectively find the granules,this paper proposed GC,and find several indicators which have effect of group team.After the discovery of the granules,we obtained the kernel variables refecting the characteristics of the aggregates by analyzing the correlation between the internal indicators of the granules.Examples are given to show that the discovery of granules and the structural analysis of granules can effectively analyze the correlation between complex data indicators.

关 键 词：复杂数据 团粒 群组效应 GC算法 团粒结构 内核变量 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论] O212.1[自动化与计算机技术—计算机科学与技术]