多指标面板数据的聚类分析及其应用

The Clustering Analysis of Multivariable Panel Data and Its Application

多指标面板数据的多元统计分析在国内研究中尚属空白.本文分析了面板数据的数据格式和数字特征,根据聚类分析原理,重新构造了多指标面板数据的距离函数和离差平方和函数,在此基础上,说明了多指标面板数据的聚类分析过程.最后对我国各地区工业企业生产效率进行了聚类实证分析,显示了良好的效果。

It is well known that finance data tend to heavy4ailed. In this paper, on a basis of an exponential regression model for log-spacings we propose an extreme quantile estimator of heavy-tailed distribution and attain an estimation of value-at-risk. As an empirical example we consider a vaiue-at-risk calculations for China stock index.