摘要
在信息技术快速发展的今天,对问题和现象的研究已不再局限于单纯利用截面数据或时间序列数据进行分析。与之相比,纵向数据包含更多的信息,广泛而深入地挖掘其数据中的信息对于人类认识自然、认识社会有着重要意义。通过对纵向数据结构特征的研究,提出了纵向数据变量的局部积差相关系数矩阵。该相关阵不仅可以对变量的内部结构进行分析,更为重要的是,以此为基础来计算变量间相关系数,可以从一个新的视角,更为深入地揭示变量间的相关关系,进而可以对变量施以新型聚类分析。
Along with the rapid development of information and technology,problems and the phenomenon is not confined to the research of Cross-sectional data,or time series data.Compared with them,longitudinal data contained more information.Widely and deeply mining the information contained in the longitudinal data has great significance to the human nature and society.Based on the study of the longitudinal data structure,Local Product-Moment Correlation Coefficient(PMCC) matrix is put forward in the paper.This local PMCC matrix could be used not only to the internal structure analysis;more importantly,it could be the basis for calculating correlation coefficient between variables from a new perspective,and moreover could reveal the relationship between variables through a new style of cluster analysis.
出处
《统计与信息论坛》
CSSCI
2011年第3期27-31,共5页
Journal of Statistics and Information
关键词
纵向数据
积差相关
聚类分析
longitudinal data
product-moment correlation coefficient
clustering analysis