\magnification=1200
\baselineskip=20pt
\nopagenumbers
\font\big=cmr12 scaled \magstep2

\centerline{\bf STANFORD UNIVERSITY}
\centerline{\bf DEPARTMENT OF STATISTICS}
\centerline{\big DEPARTMENTAL SEMINAR}
\bigskip

\baselineskip=12pt
\centerline{4:15 p.m., Tuesday, July 31, 2001}
\centerline{Sequoia Hall Rm. 200}
\centerline{(Cookies at 3:45 in 1st Floor Lounge)}

\bigskip
\baselineskip=15pt
\centerline{\sl Marina Vannucci}
\centerline{\sl Texas A\&M University}
\bigskip
\centerline{\bf Bayesian variable selection methods in Chemometrics}

\bigskip

We consider the choice of explanatory variables in multivariate linear
regression with predictors arising as curves. Applications are to infrared
spectroscopy, where a large number (several hundred) of explanatory
variables is used, typically larger than the number of observations. We
approach the problem from Bayesian modeling, using mixing priors and MCMC
methods to explore the posterior distribution. We also investigate the use
of wavelet methods, where curves are represented through wavelet
coefficient sets. In our practical context we want to choose subsets which
are good for prediction of all responses simultaneously. We predict by
averaging over a set of likely a posteriori subsets but also look into
different prediction strategies that use single best models. 


\bye