|
Abstract:
|
This course will provide a tutorial on the basics of quantile regression. Quantile regression concerns the estimation of the conditional quantile of a response variable given a set of covariates. By flexibly accommodating varying covariate effects at different quantile levels, quantile regression is able to provide a more complete picture of the relationship between the response variable and covariates than the traditional linear least squares regression. One significant feature of quantile regression is its ability to incorporate heterogeneous covariates effects, which allow the covariates to influence not only the location but also the shape of the conditional distribution. For the morning section, we will start with an introduction to the basics of the linear quantile regression, including how the conditional quantile is defined, how to fit a quantile regression model, and how to use existing statistical software to fit the model and conduct statistical inference. We will then discuss nonparametric quantile regression and semiparametric quantile regression (including partially linear, additive, varying coefficients models) to accommodate nonlinear covariate effects. For the afternoon section, we will focus on the applications of quantile regression in several important areas: missing data analysis, longitudinal data analysis, time-to-event data analysis (survival analysis) and high-dimensional data analysis. The instructors will demonstrates each application on real life data sets from economics, business, medicine and genetics and other fields and show how to use R and SAS to conduct the analysis. The participants are expected to know the basics of linear regression but no prior knowledge of quantile regression is required.
|
