07-10-2017 09:28 PM - edited 07-10-2017 10:28 PM
I want to do lasso variable selection to mixed model (fixed effect and random effect) for binary data.
Which SAS procedure enable me to do this?
If there is no adequate procedure, are there altanative ways?
(I come up with LOGISTIC procedure lasso selection to fixed effect.)
07-10-2017 10:24 PM
I don't think so, but you can via a macro. Take a look at: chrome-extension://oemmndcbldboiebfnladdacbdfmadadm/http://newprairiepress.org/cgi/viewcontent.cgi?article=1124&context=agstatconference
The link (in the paper) to the macro is no longer valid, but George is now an instructor at SAS, thus I'd be surprised if the macro isn't still available somewhere.
Art, CEO, AnalystFinder.com
07-11-2017 12:12 AM
Thank you Art!
The sas macro which you recommended looks helpful,
but I am looking for the way which can be done by SAS product
because it is difficult for me to judge the macro is generally accepted or not.
Thank you again,