s-news
[Top] [All Lists]

Lowess polynomial degree

To: s-news@lists.biostat.wustl.edu
Subject: Lowess polynomial degree
From: Xin Tian <xint@stat.rutgers.edu>
Date: Wed, 27 Feb 2002 01:17:37 -0500 (EST)
In-reply-to: <200202261122.FAA24019@wubios.wustl.edu>
Dear Splus users,

  I am using Splus (2000). I have a question about the Splus smoothing
procedure " lowess".  I would like to know the degree of polynomials used
for the local fit for lowess.

The help menu says " Lowess uses robust locally linear fits " and it
based on Cleveland (1979). Does this mean for each point, it use
weighted simple linear regression (polynomial with degree=1 ? ). Lowess
does not allow to specify degree. I tried lowess (with iteration=0)
and loess (set degree =1), the results are close, diff < 0.05, but not the
same.  

Any help will be greatly appreciated.

Sincerely,

Shirley






<Prev in Thread] Current Thread [Next in Thread>
  • Lowess polynomial degree, Xin Tian <=