You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I can only speak for myself, and unfortunately, I won't be able to implement this in the near future.
But I'm tagging @avehtari so he knows about the feature request.
I guess the most important step will be to find a (frequentist) R package that supports maximum-likelihood estimation for GLMs including monotonic effects. And by that, I'm referring only to non-multilevel, non-additive models and also only to the traditional and the latent projection. For the augmented-data projection (considering again only non-multilevel, non-additive models), special R packages offering weighted maximum-likelihood estimation for the respective outcome families will be needed.
Hello,
I know that projpred does not currently support monotonic effects but will we see this development in the near future?
Thank you in advance,
The text was updated successfully, but these errors were encountered: