摘要
将MCMC算法融合到主成分回归分析模型中,提出MCMC主成分回归分析方法.新方法既具有有效避免解释变量之间的多重共线性问题以及简化回归方程结构的主成分回归分析方法的优势,又能够充分利用MCMC算法的融合先验信息、模型信息及样本似然函数的长处.将方法应用于对嘉兴市1997年至201.0年的经济发展指标的数据建模分析,结果表明,方法能有效克服现有分析方法的不足,建立预测精度更高的模型.
This paper integrates Markov Chain Monte Carlo (MCMC) algorithm into the principal component regression analysis and proposes the MCMC principal component re- gression algorithm. The new method can not only achieve to avoid multi-collinearity of explanatory variables and simplify the structure of the regression equation provided by prin- cipal component regression, but also be able to take full advantage of the prior information, model information, and sample likelihood function brought by Bayesian MCMC algorithm. We apply the method in analyzing the economic data of Jinxing from 1997 to 2010. The results show that it can overcome the shortcomings of the exist method and set the model with higher accuracy prediction.
出处
《数学的实践与认识》
北大核心
2015年第1期56-62,共7页
Mathematics in Practice and Theory
基金
嘉兴学院教改重点课题(85151316)
全国教育信息技术研究"十二五"规划2013年度课题(136231128)
关键词
主成分回归
MCMC算法
经济指标
Principal component regression
MCMC algorithm
economic index