[RsR] [R] M-estimator R function question
Martin Maechler
m@ech|er @end|ng |rom @t@t@m@th@ethz@ch
Mon Dec 5 12:25:56 CET 2005
>>>>> "Kjell" == Kjell Konis <konis using stats.ox.ac.uk>
>>>>> on Mon, 5 Dec 2005 10:29:11 +0000 writes:
Kjell> Here is an implementation of the OGK estimator
Kjell> completely in R. I haven't touched it for a while
Kjell> and I forget how thoroughly I tested it so use it
Kjell> with a bit of caution.
Kjell> http://www.stats.ox.ac.uk/~konis/pairwise.q
excellent! That's exactly what I meant (in my post two hours
ago):
The general OGK algorithm, but with the flexibility of the main
function having arguments for the "building block" functions, so
one can use different univariate sigma() or ``weight()'' functions!
When we'll start to discuss things along the covRobust()
proposal(s) from the "multivariate" working group, we'll also
have to think about function name(s) and the returned object
("output"), but I think our student will be able use your
function "as is".
Thanks a lot, Kjell!
Martin.
More information about the R-SIG-Robust
mailing list