Topic: Renewable Quantile Regression with Heterogeneous Streaming Datasets

Time: April 17th, 10:00

Location: Tencent Conference: 438-014-7163

Moderator: Prof. Jiang Rong

Rapporteur Profile:

Xuerong Chen is a Young Distinguished Professor and Doctoral Supervisor of Southwestern University of Finance and Economics (SWUFE) under the Guanghua Distinguished Scholar Program, and is a selected candidate of the National Young Talent Program and a selected candidate of the Provincial High-level Talent Program. D. from the Institute of Mathematical and Systematic Sciences, Chinese Academy of Sciences, and postdoctoral fellowship in biostatistics from the Department of Statistics, University of Missouri and Georgetown University, and visiting scholar from the University of Michigan, City University of Hong Kong, and the University of Hong Kong. His papers have been published in JASA, JCGS, Journal of Business & Economic Statistics, Biometrics, and other leading journals in statistics, biostatistics, and econometrics. He has chaired two top-level projects of National Natural Science Foundation of China (NSFC), one youth project, one sub-project of National Natural Science Foundation of China (NSFC), and one sub-project of National Key R&D Program. He was honored with the Youth Achievement Award of the Eighth Higher Education Institutions Scientific Research Excellence Award by the Ministry of Education.

Lecture Synopsis:

The renewable statistical inference have received much attention since the advent of streaming data collection techniques. However, most existing online updating methods are developed based on a homogeneity assumption and gradients; all data batches are required to be either independent and identically distributed or share the same regression parameters, and objective functions must be smooth concerning parameters. However, most existing online updating methods are developed based on a homogeneity assumption and gradients; all data batches are required to be either independent and identically distributed or share the same regression parameters, and objective functions must be smooth concerning parameters. the only existing approach that allows some regression parameters to be different for different data batches, was proposed by Luo and Song (2021) who required the homogeneous structure to be known, which is difficult to guarantee in actual application. In this paper, we develop an online renewable quantile regression method that relies on the use of the same parameters. quantile regression method that relies only on the current data and summary statistics of historical data, for both homogeneous and heterogeneous streaming data. The proposed methods are computationally efficient, can automatically detect the unknown potential homogeneous structure, and are The proposed methods are computationally efficient, can automatically detect the unknown potential homogeneous structure, and are robust to heavy-tailed noise and data with outliers. Asymptotic properties show that the proposed renewable estimators can achieve the same statistical efficiency as the oracle estimators based on individual level data. A numerical simulation and a real data analysis illustrate that the proposed methods perform well.