[R] lars / lasso with glm

[Liaw, Andy]

> From: Thomas Lumley
> 
> On Tue, 31 May 2005, Bliese, Paul D LTC USAMH wrote:
> 
> > We have been using Least Angle Regression (lars) to help identify
> > predictors in models where the outcome is continuous.  To 
> do so we have
> > been relying on the lars package.  Theoretically, it should 
> be possible
> > to use the lars procedure within a general linear model 
> (glm) framework
> > - we are particular interested in a logistic regression model.  Does
> > anyone have examples of using lars with logistic regression?
> >
> 
> I think the problem is that the nice efficient algorithm for 
> lars doesn't 
> work with generalized linear models because the weights depend on the 
> fitted means.
> 
>  	-thomas


Jerry Friedman's PathSeeker is similar to LAR, and can do (2-class)
classification, using a different loss function.  See his web site for tech
report and software.

Andy