seminario Gelfand a Varese

[sonia petrone <spetrone@unimailer.sm.uni-bocconi.it>]

                         AVVISO DI SEMINARIO
Martedi' 29 maggio, alle ore 16.00, nell'aula seminari del Dipartimento di Economia  dell'Universita' degli Studi dell'Insubria-Varese e Como
(via Ravasi 2, Varese)

                        ALAN E. GELFAND
                   University of Connecticut

terra' un seminario dal titolo:

            Bayesian Nonparametric Approaches for Median and  
                Quantile Regression Problems
Riassunto:
Median regression models become an attractive alternative to
mean regression models when employing flexible distributions
for the errors.  Classical approaches are typically algorithmic 
with desirable properties emerging asymptotically.  However, non-
parametric error models may be most attractive in the case of 
smaller sample sizes where parametric specifications are difficult
to justify.  Hence a Bayesian approach offering full and exact
inference may be appealing.

In this context we develop two fully Bayesian modeling approaches,
employing mixture models.  We then suggest two different approaches
to provide extension to the case of general quantile regression.
Also, in the case of survival data with censoring, we look at the 
problem not of median regression but of median residual life 
regression.  

For all of the above we provide illustrative examples as well as 
discussion of issues in model fitting. 

-----------
per ulteriori informazioni, potete scrivere a  
spetrone@eco.uninsubria.it

o consultare la pagina 
http://eco.uninsubria.it/webam/Seminari/index.htm