[R] exact AIC

[Prof Brian Ripley]

On Tue, 3 Jul 2007, Selene Zárate wrote:

> I am using the stepAIC function on the MASS package that does model
> selection by exact AIC. I'm not sure what the term 'exact AIC' means,
> and how it differs from  AIC.

It doesn't, but there are approximations to AIC in use.  This is explained 
in the book stepAIC supports (one is C_p, for example).

-- 
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