[R] stepwise discriminant analysis

[Prof Brian Ripley]

On Fri, 25 Apr 2003 J.Illian at abertay.ac.uk wrote:

> Hi all,
> some days ago I sent off a query on stepwise discriminat analysis and hardly
> got any reply. That's why I am trying this again now. I probably wasn;t
> specific enough the last time I did it.
> 
> I now about the step function for glm etc. and I also know how to do
> discriminat analysis in R but I was wondering whether there is an equivalent
> of the step funcion for dicrimant analysis models?

Not that I know of.  If there were, it would not be what is usually known 
as `stepwise discriminant analysis', since that usually means a series of 
T tests, not selection by AIC.

-- 
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