Yes. Each model should have defined a likelihood function which computes the BIC/AIC. In the Linear regression, with the error Normality assumption, it would be RSS/ ( n − p − 1), as far as I know.
What do you think? 2016-01-02 13:39 GMT+01:00 Gael Varoquaux <gael.varoqu...@normalesup.org>: > On Fri, Jan 01, 2016 at 08:41:56PM +0100, Marco De Nadai wrote: > > I would expose it through a score function. In this way it can be called > to > > evaluate 2 models (let's say model A with 4 params and model B with 10). > > Moreover, this could also be called by feature_selection.RFECV. > > OK, but BIC is defined for a specific likelihood. I guess that what you > want is the likelihood associated to linear model with Gaussian > dstributions? > > Gaël > > > ------------------------------------------------------------------------------ > _______________________________________________ > Scikit-learn-general mailing list > Scikit-learn-general@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/scikit-learn-general > -- *Marco De Nadai* *Ph.D. student at Fondazione Bruno Kessler (FBK) - * *MobS Unit* *University of Trento* Via Sommarive, 18 - Povo 38123 Trento (TN) - Italy E-mail: dena...@fbk.eu LinkedIn: https://it.linkedin.com/in/marcodenadai
------------------------------------------------------------------------------
_______________________________________________ Scikit-learn-general mailing list Scikit-learn-general@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
