A0381
Title: Sequential quantile regression for streaming data by least squares
Authors: Ye Fan - Capital University of Economics and Business (China) [presenting]
Abstract: Massive streaming data are common in modern economic applications, such as e-commerce and finance. They cannot be permanently stored due to storage limitations, and real-time analysis needs to be updated frequently as new data becomes available. A sequential algorithm, SQR, is developed to support efficient quantile regression (QR) analysis for streaming data. Due to the non-smoothness of the check loss, popular gradient-based methods do not directly apply. The proposed algorithm, partly motivated by the Bayesian QR, converts the non-smooth optimization into a least squares problem and is, hence, significantly faster than existing algorithms that all require solving a linear programming problem in local processing. The SQR algorithm is further extended to composite quantile regression (CQR), and it is proven that the SQR estimator is unbiased, asymptotically normal, and enjoys a linear convergence rate under mild conditions. The estimation and inferential performance of SQR are also demonstrated through simulation experiments and a real data example on a US used car price data set.
