Is there some reason not to shrink the coefficients of the indicator or
dummy variables for categories, nominal or ordinal, as well as the
coefficients of continuous variables?
David Smith
> Date: Sat, 28 Aug 2010 17:02:25 -0400
> From: Lars Chi <[log in to unmask]>
> Subject: Ridge/Lasso Regression - Query
>
> Hi,
>
> In most Textbook examples of Rigge/Lasso Regression input variables are
> numeric. In practice, how should I deal with nominal inputs? For each
> nominal input I would have n-1 coefficients (with n being the number of
> categories). But does it make sense to shrink nominal factors?
>
> Thanks in advance,
> Lars.
>
>
>
>
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|