Quantile Regression
Brian Neelon, Duke University 
*Stephen Portnoy, University of Illinois at Urbana-Champaign 

Keywords:

Quantile regression seeks to provide a more complete picture of regression problems by analyzing all the conditional quantiles of the response in terms of interpretable linear models. This is especially useful when population heterogeneity leads to variation in the regression parameters as the quantile probability changes. The basic ideas will be presented through examples and the underlying basis for quantile regression will be summarized. Recent work on censored regression quantiles (for survival models) will be presented.