基于多指标面板数据的灰色矩阵关联模型及其应用

Grey Matrix Relational Modeling and Its Application Based on Multivariate Panel Data

多指标面板数据包含研究对象的诸多信息,但复杂的数据结构也给分析与测度截面个体之间的相似性带来了一定的困难,针对这一问题,探讨了多指标面板数据的矩阵表示,分析了多指标面板数据的时空特征,基于多指标面板数据的时空特征构建了可用于多指标面板数据截面相似性测度的灰色矩阵关联分析模型,该模型将能够表征面板数据时空特征的"水平"距离、"增量"距离、"变异"距离引入灰色关联度计算模型,并将灰色关联度计算由一般向量空间拓展到矩阵空间。并以我国区域道路交通事故成因分析为例,对我国区域道路交通事故的成因进行了分析,检验了模型的有效性。

Though containing plenty of information of the research objects, the multivariate panel data, due to the complex data structure, still bring about difficulty in analyzing and measuring the similarity among the cross-section individuals. To solve the problem, the possibility for the multivariate panel data to be represented as the matrix has been discussed, and the spatial and temporal characteristics of them also has been analyzed. Based on the spatial and temporal characteristics of the multivariate panel data, the grey matrix relation analysis model, which can be applied to measure the similarity of the multivariate panel cross-section data, is built. The model is capable of leading the ＂horizon＂ distance, the ＂increment＂ distance and the ＂variation＂ distance, which of the spatial and temporal characteristics of the panel data, to the grey relational degree model. And it furthers the calculation from the general vector space to the matrix space. In this article, we take traffic accidents in some area in our country as an example. We have studied and analyzed the phenomenon, and we conclude that this model is of applicability.