Skip to content

[RsR] Gamma shape and scale

3 messages · claudio, Matthias Kohl, Matias Salibian-Barrera

#
Dear All,
sorry for the silly question.
Does anyone have R code for the (plain) shape and scale Gamma 
model using M type estimators?

Thanks,
Claudio

--------------------------------------------------------------
Claudio Agostinelli
Dipartimento di Statistica
Universita' Ca' Foscari di Venezia
San Giobbe, Cannaregio 873
30121 Venezia
Tel: 041 2347446, Fax: 041 2347444
email: claudio at unive.it, www: www.dst.unive.it/~claudio
--------------------------------------------------------------
Per favore non mandatemi allegati in Word o PowerPoint.
Si veda http://www.gnu.org/philosophy/no-word-attachments.html

Please avoid sending me Word or PowerPoint attachments.
See http://www.gnu.org/philosophy/no-word-attachments.html
#
Hello Claudio,

my R package ROptEst can be used to compute optimally robust estimators 
for all L2-differentiable (differentiable in quadratic mean) parametric 
families which are based on a univariate distribution (e.g., the Gamma 
model). You can find it under (not yet on CRAN):
http://www.stamats.de/RobASt.htm
You will also need the packages distr, distrEx, RandVar and evd which 
are on CRAN.

Using Linux you might get an error by R CMD check, but this is due to 
some problems with S4 class protoypes which I haven't corrected yet and 
which do not affect the corresponding methods you need.

After the installation there is the R Skript "GammaModel.R" in the 
directory of .../ROptEst/Scripts which might help you. Moreover, you can 
find a short treatment of the Gamma model in Chapter 6 of my thesis
http://www.stamats.de/ThesisMKohl.pdf

Feel free to contact me directly, if you encounter problems!

Kind regards,
Matthias

  
    
#
Hi Claudio,

M-estimators in Hampel et al. (1986, pp 238-257) have been implemented by 
Alfio Marazzi. Look in:

http://www.iumsp.ch/Unites/us/Alfio/msp_programmes.htm

Full disclosure: these estimators were also implemented in the S-PLUS 
robust library.

Matias
On Wed, 11 Jan 2006, claudio wrote: