[RsR] Tutorial: Introduction to Robust Statistics with R

"ValenT" == Valentin Todorov <valentin.todorov at chello.at>
    on Fri, 6 Jun 2008 23:49:49 +0200 writes:

ValenT> Today I came across the announcement of the tutorial at the UseR! Conference 
    ValenT> in Dortmund "Introduction to Robust Statistics with R" and was surprised to 
    ValenT> see that it will "... use R functions mainly from packages 'MASS' and ..."

didn't you guess I was the culprit ?  :-)
I think I mainly mentioned MASS because it's been a standard R
package for a long time.

    ValenT> I wonder what does 'MASS' still has that 'robustbase' & cie do not have?

Probably not much, but
simple  M-estimate for regression with flexible choice of
psi/rho function is something not available as "high-level"
in robustbase::lmrob() nor robust::lmRob() I think.

E.g. if you have only categorical predictors, such "simple" M estimates 
may make much sense, no?

Regards,
Martin

Hey Martin, why do you read/write mails instead of watching the EM matches 
(sorry)?! I do not care much about soccer too but just opened a bottle of 
Portuguese wine to celebrate and would like to use the opportunity to 
congratulate our Portuguese robastniks.

    ValenT>Today I came across the announcement of the tutorial at the UseR! 
Conference
    ValenT>in Dortmund "Introduction to Robust Statistics with R" and was 
surprised to
    ValenT> see that it will "... use R functions mainly from packages 
'MASS' and ..."

didn't you guess I was the culprit?  :-)

Of course I did, therefore I allowed myself to pull your leg.

I think I mainly mentioned MASS because it's been a standard R
package for a long time.

Then you should mention also MINVOL, ROBETH, COVINTER, PROGRESS which were 
around for quite a long time so far and are in all caps too.

But seriously speaking, 'MASS' stroke my eyes and I am sure that this will 
be what a newcomer will take home (the advantage of all caps you know, we 
should rather have chosen ROBUSTBASE).


Best regards,

Valentin