文章摘要
Long Zhenhuan,Zhang Feipeng,Zhou Xiaoying.A Continuous Piecewise Linear Quantile Regression with Multiple Change-points[J].The Journal of quantitative and technical economics,2017,(8):150-161
带多个变点的逐段连续线性分位数回归模型及应用
A Continuous Piecewise Linear Quantile Regression with Multiple Change-points
  
DOI:
中文关键词: 多个变点  逐段连续线性  分位数回归  LASSO  线性化技巧
英文关键词: Multiple Change-points  Continuous Piecewise Linear Quantile Regression  LASSO  Linearization Technique.
基金项目:本文获得国家自然科学青年基金项目(11401194)、湖南省自然科学青年基金项目(2017JJ3021)、湖南大学中央高校基本科研业务费专项资金(227201305039)的资助。
Author NameAffiliation
Long Zhenhuan School of Finance and Statistics, Hunan University 
Zhang Feipeng School of Finance and Statistics, Hunan University 
Zhou Xiaoying School of Finance and Statistics, Hunan University 
Hits:
Download times:
中文摘要:
      研究目标:建立具有多个变点的逐段连续线性分位数回归模型(Continuous Piecewise Linear Quantile Regression with Multiple Change Points,CPLQR)。研究方法:先通过LASSO和广义贝叶斯信息准则确定变点个数,再通过线性化技巧来估计变点的位置与回归系数。研究发现:新方法能够同时确定变点个数、估计变点位置和回归系数,而且具有较强的稳健性;应用该方法于年龄和身体质量指数之间关系,进一步证实了模型的实用性。研究创新:新方法能够处理多个变点的问题,通过LASSO和广义贝叶斯信息准则确定变点数目,避免了主观判断的弊端;借助线性化技巧,解决了目标函数在变点处不可导问题。研究价值:本文结果将为分析经济、金融、医药和生物等学科中存在结构变化的数据提供强有力的研究工具。
英文摘要:
      Research Objectives: To propose a new method for a continuous piecewise linear quantile regression model with multiple change points (CPLQR). Research Methods:Estimating procedure determines the number of change-points by LASSO and generalized BIC, and then the locations of change-points and regression coefficients are estimated by a linearization technique. Research Findings: The new method can not only determine the number of change-points, but also simultaneously and robustly estimate the locations of change-points and the regression coefficients. We apply the new method to analyze the relationship between the age and the BMI, which demonstrates the practical utility of the methodology. Research Innovations: The new method could deal with multiple change points problem, and the number of change points is determined by Lasso and generalized BIC. By utilizing a linearization technique, the new method overcomes the non-differentiability of the objective function due to the existence of change points. Research Value: This paper provides an effective tool for the data with change points in various research areas, such as economics, finance, biology, medicine and so on.
View Full Text   View/Add Comment  Download reader
Close