soci209 - m3 - lowess aka loess

This page is hosted on AFS file server space, which is being shut down on November 13, 2018. If you are seeing this message, your service provider needs to take steps now. Visit afs.unc.edu for more information.

LOWESS

The LOWESS aka LOESS procedure is carried out in two phases: (1) initial fit, and (2) robustness iterations to reduce the influence of extreme observations (outliers).
The following three exhibit illustrate how the LOWESS algorithm works.

Exhibit: Initial fit for a point with x-value inside the range of x, with a = 0.5 (a is width of "window", i.e. proportion of observations used in calculating the fitted value corresponding to x) (Cleveland 1994, Figure 3.49 p. 177) [lowess01.gif]
Exhibit: Initial fit for a point on the boundary of the range of x (Cleveland 1994, Figure 3.50 p. 178) [lowess02.gif]
Exhibit: Robustness iterations to reduce influence of outliers (Cleveland 1994, Figure 3.51 p. 179) [lowess03.gif]

Created 20 Jan 2003