Bayesian nonparametric regression with a flexible error term distribution

Courtney Barnes, University of Texas at El Paso

Abstract

Datasets often exhibit heavy tailed behavior and standard analyses are often heavily influenced by outliers. We propose a nonparametric regression model whose error term distribution is a mixture of a normal and a Student t distribution. This results in a model that is more resistant to outliers compared to a model with a normal error term.^

Subject Area

Statistics

Recommended Citation

Barnes, Courtney, "Bayesian nonparametric regression with a flexible error term distribution" (2010). ETD Collection for University of Texas, El Paso. AAI1477770.
http://digitalcommons.utep.edu/dissertations/AAI1477770

Share

COinS