For another approach which deals with discrete and continous parameters,
you might like to check out the following references.

C.S. Wallace and D.M. Boulton (1968), An information measure for
classification, Computer Journal, 11(2), pp185-194.

C.S. Wallace and D.L. Dowe (1999), Minimum Message Length and Kolmogorov
Complexity, Computer Journal, 42(4), pp270-283.

C.S. Wallace and P.R. Freeman (1987), Estimation and inference by
compact coding., J. R. Statist. Soc B, 49(3), pp240-265.


David Albrecht                    email: [EMAIL PROTECTED]   
School of Computer Science          tel: +61 3 9905 5526 
 and Software Engineering,          fax: +61 3 9905 5146 
Monash University,                  URL: http://www.csse.monash.edu.au/~dwa/
Clayton, VIC 3800, Australia        Monash Provider No: 00008C

On Mon, 19 Jan 2004, Martin C. Martin wrote:

> Hi,
> 
> Criteria such as AIC and BIC guide choices between models with 
> continuous parameters.  But what if a model has a discrete parameter? 
> There are tricky issues here.  For example, a parameter that has 10 
> possible values provides the same number of alternatives as two 
> parameters which can take on two and five values respectively.
> 
> Any pointers to discussions/theory/practice of model selection with 
> discrete parameters?
> 
> Thanks,
> Martin
> 

Reply via email to