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
Hi! I tried to run the score testing on 4 different categories (each on 1 core with 16gb allocated to it) on a model with 1 covariate (2 levels) and 4885 categories. 3 of the tests ran fine, they took less than two hours to finish, but one of them just doesn't finish running (> 15 hours, same parameters as the others), is that something you've experienced? Do some categories just take a really long time to converge?
The text was updated successfully, but these errors were encountered:
This is something that I've experienced - in an analysis that I ran with ~8,000 categories, I found that the median test took 91 minutes, while some tests took over 5 days. Hopefully this helps!
Great question, @almita -- my initial reaction is either that the likelihood is very flat, or that it has a really unusual shape making the locally polynomial approximations poor. I know it's not a great answer but I hope it helps!
FYI We have a few projects in the works to speed up cases exactly like this -- currently documented as #89 and #90 -- but two of our core team members are currently on well-deserved vacations. Please stay tuned for updates and sorry to hear you have to wait a while for this pesky score tests... and thanks for your patience waiting for high quality inference ⭐
Hi! I tried to run the score testing on 4 different categories (each on 1 core with 16gb allocated to it) on a model with 1 covariate (2 levels) and 4885 categories. 3 of the tests ran fine, they took less than two hours to finish, but one of them just doesn't finish running (> 15 hours, same parameters as the others), is that something you've experienced? Do some categories just take a really long time to converge?
The text was updated successfully, but these errors were encountered: