[R] mle.stepwise versus step/stepAIC

[Prof Brian Ripley]

You could ask the author of the contributed package 'wle', which is not 
part of the `R-software'.

The documentation is minimal, but the references are all to classical 
stepwise methods such as those in package leaps, that is they select 
columns in the model matrix and _not_ terms.

On Wed, 7 Dec 2005, Jim Brindle wrote:

> Hello,
>
> I have a question pertaining to the stepwise regression which I am trying to
> perform.  I have a data set in which I have 14 predictor variables
> accompanying my response variable.  I am not sure what the difference is
> between the function "mle.stepwise" found in the wle package and the
> functions "step" or "stepAIC"?  When would one use "mle.stepwise" versus
> "step/stepAIC"?  Other than the references in the R-software, is there
> anything which discusses the use of "mle.stepwise"?
>
> Thank you kindly in advance for any insight offered.
>
> Jim Brindle


-- 
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