Quantile Regression Analysis of Heterogeneous Data in Ultra-high Dimension
Abstract: Modern high-dimensional data are often heterogeneous in the sense that the covariates (predictors) often influence not only the location but also the dispersion or other aspects of the conditional distribution. Quantile regression enjoys some unique advantages for analyzing high dimensional heterogeneous data. By considering different conditional quantiles, we may obtain a more complete picture of the conditional distribution of a response variable given high dimensional covariates. The sparsity level is allowed to be different at different quantile levels. The talk will provide an overview of recent advances in the statistical theory and algorithms for high-dimensional penalized linear/nonlinear quantile regression with both convex and nonconvex penalty functions.
Short Bio: Lan Wang is a professor at School of Statistics, University of Minnesota. Her research interests include nonparametric and semiparametric statistics with focus on high-dimensional data analysis, quantile regression, estimating equations, censored data, model diagnostics and their applications. She is a Fellow of the Institute of Mathematical Statistics and is on the editorial board of Annals of Statistics, Journal of the American Statistical Associations.