PLACE:  Atanasoff 214

SPEAKER:

Dr. Mario Francisco-Fernandez, Universidad de A Coruna, SPAIN

TITLE:

A PLUG-IN BANDWIDTH SELECTOR FOR LOCAL
POLYNOMIAL REGRESSION UNDER DEPENDENCE

ABSTRACT:

Consider the fixed regression model where the error random variables are
coming from a strictly stationary stochastic process. In a situation like
this, automated bandwidth selection methods for nonparametric regression
break down. In this talk, we present a plug-in method for choosing the
smoothing parameter for local least squares estimators of the regression
function and its derivatives in the presence of correlated errors.
The theoretical performance of the bandwidth estimator for the local linear
estimator of the regression function is obtained. These results can be
extended to other settings, such as derivative estimation and multiple
nonparametric regression. Estimators of regression functionals and error
correlation based on local least squares ideas are developed in this
article. A simulation study illustrates the selection method proposed ..

Finally, we also consider some extensions to the problem of selecting the
smoothing parameter in the spatial case. In this setting, we propose some
techniques based on correcting some classical criteria like cross-validation
or generalized cross-validation, by including the covariances of the errors in
the method.

COFFEE:  2:45p.m., 104 Snedecor Hall