Date:         Tue, 18 Jan 2011 09:52:43 -0500
Reply-To:     oloolo <dynamicpanel@YAHOO.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         oloolo <dynamicpanel@YAHOO.COM>
Subject:      Re: Logistic Regression on Autopilot - GLMSELECT adaptation
Comments: To: SUBSCRIBE SAS-L Olivier Van Parys <olivier.vanparys@GMAIL.COM>

would u please re-send their link
I can't see any link or reference here

On Tue, 18 Jan 2011 09:33:01 -0500, SUBSCRIBE SAS-L Olivier Van Parys
<olivier.vanparys@GMAIL.COM> wrote:

>Hi Everyone,
>
>The following presentation from Hastie & Park point towards a Lasso
adaptation
>for Logistic regression. While their code is available in R I was
wondering if
>simply using GLMSELECT in a linear probability setting was the best that
could
>be done in SAS to reproduce the objective function in Slide 5?
>Also do you see any value in scaling all the input variable within the 0-1
>range (perhaps after loging them) before using GLMSELECT?
>
>Olivier