GL(M)M's as a future target?

[jrhadams]

I was curious if generalised linear models and/or other models that use link functions were a potential future feature.

Given the sommer's rather large scope, I was surprised to not see any mention of GLM's.

[covaruber]

It is a potential future feature but I am not close to get there. I have very little time to invest in the development of sommer and my next big step is the addition of the asreml average information algorithm for the n > p (more observations than coefficients) situation. Then maybe 2022 can be a year to work on GLMs but to be honest I am not an expert in the topic of GLMs for mixed models, I've been waiting for someone to contribute in this area but not found yet :) I'll be happy to know if you are or know someone :) cheers.

[jrhadams]

I'd like to be, but I'm unfortunately not yet equipped wrt the inner workings of GLMM's. :laughing:

Anyway, yes I can understand why implementation of AI is taking up the majority of your energy (btw, really looking forward to that when it rolls out!).

If I bump into anyone with a good understanding of GLM's, I'll be sure give them a nudge this way. :smirk:

Thanks!

[covaruber]

Thanks for the suggestion though, for sure I would love to go in that direction. Let me know if you meet someone ;) I will come back to you in the future.

Cheers, Eduardo

[covaruber]

MME- version has been added to sommer through the mmec function. Still doesn't have GLMs but we have moved one step forward.

Cheers,