Statistics Colloquium
March 30 (Fri) 11 a.m.
Alavi Commons Room, 6625 Everett Tower

Quantile Regression Variable Selection

Yonggang Yao
Research Statistician Developer
SAS Institute Inc.
Raleigh-Durham, North Carolina

Quantile regression aims to predict conditional quantiles. Compared to linear regression which is exclusively focusing on conditional means, quantile regression provides a comprehensive strategy to investigate entire conditional distribution. As a result, quantile regression can be naturally used to analyze heteroscedastic data without preceding data transforming, which is required by linear regression. In terms of variable selection, quantile regression models at different quantile levels may rely on different subsets of variables. Such a property is extremely valuable in both theoretical sense and practical sense. In this talk, I will introduce some popular quantile regression variable selection methods, and show several data analysis examples.

All statistics students are expected to attend.


Past colloquiums


Department of Statistics
3304 Everett Tower
Western Michigan University
Kalamazoo MI 49008-5152 USA
(269) 387-1420 | (269) 387-1419 Fax