.png){kind=icon}
.png){kind=icon}
Return
Distributed Composite Quantile Regression for High-Dimensional Data
.png){kind=icon}
PRE
AI
DOI：10.1007/s13171-025-00430-9.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This paper investigates distributed Composite Quantile Regression (CQR) estimation for high-dimensional linear regression models, presenting a robust and computationally efficient approach. Methodologically, each iteration requires only the master machine to solve a shifted & ell;1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varvec{\ell }_{\varvec{1}}$$\end{document}-regularized least squares problem, leveraging a newly established theoretical link between composite quantile regression loss and squared loss functions. The proposed estimator achieves asymptotic equivalence to the full-data estimator after a finite number of iterations, with identical convergence properties. Empirical evaluations through simulations and real-data analyses demonstrate the method's strong finite-sample performance, validating its practical applicability and statistical efficiency in distributed computing environments.
Keywords:
Distributed system
High-dimensional data
Composite quantile regression
Journal
S
IF:
0.5
Papers:
29
Citations:
0