Abstract
The finite sample performance of a number of linear regression estimators is investigated in a variety of parametric settings involving outliers. A Bayesian approach is shown to have good overall comparative performance. It is then shown how the same Bayesian methodology can be easily extended to robust nonparametric regression. The Bayesian analysis is carried out using the Gibbs sampler.
Original language | English |
---|---|
Pages (from-to) | 269-301 |
Number of pages | 33 |
Journal | Computational Statistics |
Volume | 11 |
Issue number | 3 |
State | Published - 1996 |
Keywords
- Gaussian mixture
- Gibbs sampling
- Nonparametric regression
- Outliers
- Regression splines
ASJC Scopus subject areas
- Statistics and Probability
- Statistics, Probability and Uncertainty
- Computational Mathematics