Communication-Efficient Feature Screening for Ultrahigh-Dimensional Data Under Quantile Regression

Feature screening is a powerful method for dealing with ultrahigh-dimensional data, but when the sample size is also increased, traditional methods encounter challenges due to constraints in data storage and computational complexity. In this article, we present a novel method for feature screening in quantile regression models implemented on the distributed system. To enhance convergence, we adopt an iterative projected gradient descent algorithm and establish the corresponding sure screening property. Extensive simulation studies confirm the effectiveness of the proposed method, and an application to a real dataset showcases its practical utility.