基于模糊分类的多维定性数据的统计分析

Statistical Analysis of Multidimensional qualitative data on Fuzzy Classification

在多维定性数据处理的问题中,不仅要研究变量之间的关系、样本之间的关系,还需研究样本与变量之间的关系,从而对潜在样本群子结构水平及其类型特征进行分析。本文采用对偶标度统计方法利用设计矩阵变换将原始多维定性数据进行类型特征分析,并在此基础上建立了基于模糊分类的直方图,表现出了潜在样本群的特征及其子结构之间的差异与联系。文章以分析儿童主体品质发展规律为例,说明此方法的有效性。

In processing multidimensional qualitative data,people not only need study the relations between variables and between samples,but also the relation between variables and samples,sequentially the substructure level and the characters of the types of the latent sample group are analyzed. In this paper we apply the dual scaling and design matrix to analyze the characters of multidimensional qualitative data,and construct a histogram based on fuzzy classification,thereby clarify the characters and the relations of the substructure level of the latent sample group. An example about development of children’s principalqualities is given to illustrate efficiency of the method.