Summary of Robust Regression Algorithms

Humbolt, Allen (HumboltA@kochind.com)
Mon, 5 Jan 1998 12:49:00 -0600

Many thanks to David Ross, James Jetton, Weiguo Fan, Helmut Elsenbeer,
Brian Ripley, Bill Venables, and Doug Martin for their information about
locating robust regression code that can be run independent of Splus.
Those who had detailed comments about these sources of code also sent
their replies to the snews list, so I won't repeat their detailed
comments below.

Two sources of software mentioned more than once were:

1. ROBETH (A. Marazzi) Fortran Code
Documentation in Algorithms, Routines and S Functions for Robust
Statistics, book by Marazzi (1993, Wadsworth and Brooks/Cole)
AMARAZZI@IUMSP.UNIL.CH at a "modest price" (Fortran)
http://lib.stat.cmu.edu/apstat (sources and Splus interface)
http://www.hospvd.ch/iumsp/download/robeth-en.htm (later than
statlib)

2. PROGRESS, in Fortran
http://win-www.ula.ac.be/u/statis/programs.html
http://win-www.ula.ac.be/u/statis/index.html

Others sources mentioned were:

3. http://www.astro.psu.edu/statcodes/sc_correlregr.html in Fortran and
C/C++.

4. lmsreg Splus routine, as taken from Rousseeuw and Leroy's book (1987,
John Wiley, Robust Regression and Outlier Detection)

Doug Martin sent me an e-mail suggesting that he may post to s-news some
comments about the merit or lack of merit of robust procedures like
least trimmed squares. I think that could spark a fun and lively
debate. Go for it, Doug.

Allen Humbolt
Quantitative Analyst
HumboltA@kochind.com e-mail
(316) 828-7911 phone
(316) 828-8674 fax