分位回归基于最优去相关得分的子抽样算法  

Subsampling Algorithm for Quantile Regression Based on Optimal Decorrelation Score

作　　者：黄小峰 邹雨浩 袁晓惠 HUANG Xiaofeng;ZOU Yuhao;YUAN Xiaohui(School of Mathematics and Statistics,Changchun University of Technology,Changchun 130012,China)

机构地区：[1]长春工业大学数学与统计学院,长春130012

出　　处：《吉林大学学报（理学版）》2024年第5期1102-1112,共11页Journal of Jilin University:Science Edition

基　　金：国家社会科学基金(批准号:22BTJ019);吉林省教育厅科学研究项目(批准号:JJKH20230749KJ)。

摘　　要：针对海量数据下高维分位回归模型,首先,构造基于去相关得分函数的子抽样算法,以估计感兴趣的低维参数;其次,推导所提估计的极限分布,并根据渐近协方差矩阵求出L-最优准则下的子抽样概率,给出高效的两步算法.模拟和实证分析结果表明,最优子抽样方法显著优于均匀子抽样方法.For the high-dimensional quantile regression model with massive data,firstly,a subsampling algorithm based on the decorrelation score function was constructed to estimate the low-dimensional parameters of interest.Secondly,we derived the limit distribution of the proposed estimates and calculated the subsampl ing probability under the L-optimal criterion according to the asymptotic covariance matrix,giving an efficient two-step algorithm.The simulation and empir ical analysis results show that the optimal subsampling method is significantly superior to the uniform subsampling method.

关 键 词：去相关得分 高维 海量数据 分位回归 子抽样 

分 类 号：O212.2[理学—概率论与数理统计]