[R] quadratic discriminant analysis ?

Daniel Amorèse Daniel.Amorese at geos.unicaen.fr
Fri May 17 11:19:27 CEST 2002


Le 2002.05.17 10:08, ripley at stats.ox.ac.uk a écrit :
> On Fri, 17 May 2002, [ISO-8859-1] Daniel Amorèse wrote:
> 
> > Hello,
> >
> > I am a beginner in R and data analysis.
> > I have the VR (MASS) library installed and "Modern Applied
> Statististics
> > with S-Plus" is my bedside book.
> > Could anybody provide me an R script (or pieces of an R script) to
> perform
> > a
> > stepwise quadratic discriminant analysis ?
> 
> That needs distribution theory not implemented in R (to my knowledge).
> 
> QDA is rather dependent on multivariate normality, and the stepwise
> variants are even more so.  It seems to be very rarely appropriate.
> 
> When I wanted an example, a student ran one for me in what I recalled was
> SPSS under VMS (and it looks like SPSS output), but last time I looked I
> could not see it on the PC version.
> 
> -- 
> Brian D. Ripley,                  ripley at stats.ox.ac.uk
> Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
> University of Oxford,             Tel:  +44 1865 272861 (self)
> 1 South Parks Road,                     +44 1865 272860 (secr)
> Oxford OX1 3TG, UK                Fax:  +44 1865 272595
> 
> -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
> r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
> Send "info", "help", or "[un]subscribe"
> (in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
> _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
> 
> 

Thanks for this answer.
Many people claim LDA and QDA are quite robust methods with regard to
multivariate normality.
Of course, this assertion is certainly not true when a stepwise analysis is
performed.
Especially if the stepwise analysis involves Wilks' lambda (this test
requires multinormality).
Thus, how is it possible to (reliably) select variables when performing qda
or lda through R ?
Thanks

-- 
Daniel AMORESE
Lab. M2C "Morphodynamique Continentale et Côtière"
UMR CNRS 6143 Caen/Rouen
Centre de Géomorphologie
UCBN (Université de Caen Basse-Normandie)
14032 Caen Cedex
FRANCE
Ph: +33 (0)2 31 56 57 19
Fax: +33 (0)2 31 56 57 57
Courriel: amorese at geos.unicaen.fr
http://www.geos.unicaen.fr/perso/da/daperso.html
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._



More information about the R-help mailing list