基于函数性视角的经济数据分析——以主微分分析方法为例

Economic Data Analysis Based on Functional Viewpoint By the Principal Differential Analysis

基于经济数据的函数性特征,引入函数型数据分析方法,研究发现经济数据中的面板数据可作为函数型数据的特例,函数型数据分析方法在处理高维数据、缺失数据以及样本观测点不规则分布等特殊的数据类型有独特的优势。着重介绍和拓展了主微分分析方法,在集合了主成分分析方法优势的同时从微分方程的解出发探讨数据的特征。通过对全国银行间同业拆借利率进行主微分分析,显示出主微分分析方法能够揭示其它方法所不能反映的数据特征。

Based on the functional characteristic of economic data,this paper introduce the functional data analysis method,we find that panel data in economic data can be regard as special case of functional data,and the functional data analysis has unique advantages in deal with high dimensional data,missing data,sample observation point with irregular distribution and so on.The paper emphasizes introduce and extend the principal differential analysis method,through analyzing the nationwide Interbank Offered Rate by the principal differential analysis,we find that principal differential analysis can reveal some characteristics of the data but others method do not.