[R] plot discrimnant analysis

[Prof Brian Ripley]

On Thu, 16 Oct 2003, Marta Rufino wrote:

> Hello,
> 
> Does anyone knows how to do the plots from discriminant analysis (lda and qda)?

There are plot methods supplied for lda with examples in the MASS scripts.
You could have tried
> methods(class="lda")
[1] coef.lda*        model.frame.lda* pairs.lda*       plot.lda*
[5] predict.lda*     print.lda*

    Non-visible functions are asterisked

and see also ldahist() (referenced from ?plot.lda).

I don't know what `the plots' from qda are, but there is _a_ plot in the
MASS scripts.

> Is there any computed function to do the stepwise procedure?

Which procedure?
There is no implementation of the stepwise LDA as supplied by SPSS, say.
The main problem would be computing signifcance levels for the test 
statistics (and also the very strong distributional assumptions needed).

-- 
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 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595