Open JamesYang007 opened 1 year ago
That’s what glmnet does for Gaussian errors and type.gaussian=“covariance” method. Perhaps try glmnet first with this option Trevor
Sent from Trevor Hastie"s iphone
On Dec 17, 2022, at 17:53, James Yang @.***> wrote:
It would be nice to have a glmnet-like function that takes in individual-level data X, y and performs BASIL. The covariance method as in ghostbasil would be ideal and the covariance matrix can be constructed on the fly as strong set variables come in.
— Reply to this email directly, view it on GitHubhttps://github.com/JamesYang007/ghostbasil/issues/4, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ADFNNRMNTEVVBH4KO3HEPYLWNZVCVANCNFSM6AAAAAATCH566A. You are receiving this because you are subscribed to this thread.Message ID: @.***>
Thanks! I thought Zihuai already tried the covariance method but then realized today the default setting was naive.
It would be nice to have a
glmnet
-like function that takes in individual-level dataX, y
and performs BASIL. The covariance method as inghostbasil
would be ideal and the covariance matrix can be constructed on the fly as strong set variables come in.