M-estimator R function question

"Rand" == Rand Wilcox <rwilcox at usc.edu>
    on Fri, 02 Dec 2005 10:42:48 -0800 (PST) writes:

Rand> In the robust library of S+, it is possible to compute
    Rand> Rocke's constrained M-estimator of scatter and
    Rand> location. Can't find anything within R that does the
    Rand> same thing. Can someone tell me whether any package
    Rand> exists for computing it?

I can't offhand.
There's the cov.rob functions in 'MASS' and the  'rrcov' package
which do similar (better ? :-) things.

Note that for a few weeks now, there's the R-SIG-robust mailing
list [--> CC to there],  on which we plan to talk about and coordinate the
development of more robust methods for R --- as an offspring
from the Treviso workshop "Robustness and R".

Unfortunately, some things (e.g. workshop "minutes") are still a
bit in a waiting state.

One thing we'd be very interested is the OGK estimator of
Maronna and Zamar (JASA, 2002).  Unfortunately, there, too, some
of the authors sold their code exclusively to Insightful (the
S-plus company).

    Rand> Thanks.

You're welcome.

    Rand> 	Rand Wilcox Professor Dept of Psychology
    Rand> University of Southern California Los Angeles, CA
    Rand> 90089-1061 FAX: 213-746-9082

Martin Maechler, ETH Zurich

Dear Rand,
unfortunately, right now none of the R packages can compute Rocke's
constrained M-estimator of scatter and location. I have implemented this
estimator as function covMTB in rrcov - see
http://www.dst.unive.it/rsr/Todorov-treviso.pdf , slide 35 - but I have
still some open problems to solve before releasing it to CRAN. I hope to be
able to do this even before Christmas. And if you look at the next slide in
this presentation, you will see that there is an working implementation of
OGK, as mentioned by Martin. Again, I hope to release it very soon.

Best regards,
Valentin



----- Original Message ----- 
From: "Martin Maechler" <maechler at stat.math.ethz.ch>
To: "Rand Wilcox" <rwilcox at usc.edu>
Cc: <r-help at stat.math.ethz.ch>; <R-SIG-Robust at stat.math.ethz.ch>
Sent: Friday, December 02, 2005 10:22 PM
Subject: Re: [RsR] [R] M-estimator R function question

"Rand" == Rand Wilcox <rwilcox at usc.edu>
    on Fri, 02 Dec 2005 10:42:48 -0800 (PST) writes:

    Rand> In the robust library of S+, it is possible to compute
    Rand> Rocke's constrained M-estimator of scatter and
    Rand> location. Can't find anything within R that does the
    Rand> same thing. Can someone tell me whether any package
    Rand> exists for computing it?

I can't offhand.
There's the cov.rob functions in 'MASS' and the  'rrcov' package
which do similar (better ? :-) things.

Note that for a few weeks now, there's the R-SIG-robust mailing
list [--> CC to there],  on which we plan to talk about and coordinate the
development of more robust methods for R --- as an offspring
from the Treviso workshop "Robustness and R".

Unfortunately, some things (e.g. workshop "minutes") are still a
bit in a waiting state.

One thing we'd be very interested is the OGK estimator of
Maronna and Zamar (JASA, 2002).  Unfortunately, there, too, some
of the authors sold their code exclusively to Insightful (the
S-plus company).

    Rand> Thanks.

You're welcome.

    Rand> Rand Wilcox Professor Dept of Psychology
    Rand> University of Southern California Los Angeles, CA
    Rand> 90089-1061 FAX: 213-746-9082

Martin Maechler, ETH Zurich

_______________________________________________
R-SIG-Robust at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-robust

One observation about the following.

One thing we'd be very interested is the OGK estimator of
Maronna and Zamar (JASA, 2002).  Unfortunately, there, too, some
of the authors sold their code exclusively to Insightful (the
S-plus company).

Insightful has implemented the procedure in some way I ignore (actually, 
I don't like the way it works), but unfortunately I got not a single dollar 
from that!. Anybody is free to implement the method (which is extremely 
simple) in whatever language.
            Ricardo