[S] SUMMARY: quantile regression

Porth_Laurie_S/rmrs@fs.fed.us
Tue, 25 Aug 1998 09:55:37 -0600


Thank you very much to Vincent Carey, Thomas Yee, Pierre Duchesne, Brian Cade,
Kjetil Halvorsen, and Stephane Heritier for your quick response to my question.

It looks as though there is a lot of code available to run quantile regression
in S-Plus. Here is a summary of the responses:

-- On statlib, you will find Koenker's generalized L_1 approach (see Stat Sci
November 1997). The archive is called "quantreg", the function is "rq".
Additional material on this methodology was published in the last issue of the
newsletter of the ASA section on stat computing/graphics. To get that shar
archive go to http://lib.stat.cmu.edu/S/quantreg. More recent versions for
Windows or Unix platforms are available from Roger Koenker's web site
(http://www.econ.uiuc.edu).

-- At http://biosun1.harvard.edu/~carey/index.ssoft.html there is a link to
"lmsqreg", a penalized likelihood
approach following Cole and Green (Stat in Med 1992). It is not fully
bulletproofed and is currently limited
to one predictor. It provides graphics, diagnostics and model transfer tools
for 1-predictor problems.

-- Thomas Yee has written some code: see
http://www.stat.auckland.ac.nz/yee/vgam

A couple of people have offered their own code that is not posted at a
website. Thank you very much!

Thanks again to everyone!
Laurie Porth


-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news