This paper describes the LOESS procedure which is a new procedure in SAS/STAT software for performing local regression. Features of this procedure are outlined and a brief description of the fitting method is given. Examples are given illustrating the use of this procedure in obtaining fitted surfaces as well as prediction confidence limits for both univariate and multivariateregressor data. An automatic method for selecting the smoothing parameter based on a bias corrected Akaike information criterion is described and used in these examples.
Mendeley saves you time finding and organizing research
There are no full text links
Choose a citation style from the tabs below