首页 | 本学科首页   官方微博 | 高级检索  
     检索      

时间序列多个突变点的贝叶斯推断——对我国GDP序列的实证分析
引用本文:王维国,王霞,颜敏.时间序列多个突变点的贝叶斯推断——对我国GDP序列的实证分析[J].数学的实践与认识,2010,40(9).
作者姓名:王维国  王霞  颜敏
基金项目:教育部"用信息技术改造计量经济学课程"专项课题,辽宁省2008年度优秀人才项目,辽宁省高等学校创新团队项目
摘    要:突变点的存在对经济分析与建模会产生重要影响."邹检验"仅仅在序列存在一个突变点时有效.为了对序列中可能存在的多个突变点进行判断,引入了基于贝叶斯推断的多个突变点判断理论,并将该理论应用于我国GDP序列中.笔者检测出该序列存在三个突变点,分别位于1961年,1976年,1989年.此外,发现加入合理的突变点后,模型的预测精度得到显著的提高.

关 键 词:贝叶斯推断  多个突变点  GDP

Bayesian Inference for Multiple Change Points in Time Series——Demonstration Based on Chinese GDP Series
WANG Wei-guo,WANG Xia,YAN Min.Bayesian Inference for Multiple Change Points in Time Series——Demonstration Based on Chinese GDP Series[J].Mathematics in Practice and Theory,2010,40(9).
Authors:WANG Wei-guo  WANG Xia  YAN Min
Abstract:There is great impact for economical analysis and modeling due to the existence of change points."Chow test"is effective only in the condition which there is single change point in a series.In order to estimate multiple change points which may exist in series,we introduced Bayesian analysis and studied change points in Chinese GDP series by that.There are three change points in GDP series,which is 1961,1976,1989 respectively.Besides,it is familiar that the model with proper change points is better in forecast.
Keywords:bayesian inference  multiple change points  GDP
本文献已被 CNKI 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号