Not understood [🇷 for BE/BA]

posted by ElMaestro  – Denmark, 2020-08-04 00:55 (1296 d 01:46 ago) – Posting: # 21821
Views: 13,751

Hi Pharmcat,


❝ Do you calculate REML for all data at once or calculate REML subject by subject?


REML is not generally done on a per-subject basis. Each model is assumed to have an optimal fit, given the data and the variance/covariance structure. The optimal fit, when putatively found, corresponds to the fixed effects and the variance components, and the log likelihood serves as objective function.
I am sure I am missing something clever in your question. Can you describe what you meant?

❝ Is it possible to use function caching in R?


You can always write some public members to that effect where necessary and appropriate. I don't think R has a built-in optimizer which takes care of it as the language is interpreted, but other users of this forum should confirm; I am not expert on the inner workings of R.
I do not see how this would affect the REML process - can you describe with more words what your perspective on caching was, then we'll possibly figure out something?

Pass or fail!
ElMaestro

Complete thread:

UA Flag
Activity
 Admin contact
22,899 posts in 4,806 threads, 1,652 registered users;
28 visitors (0 registered, 28 guests [including 9 identified bots]).
Forum time: 01:42 CET (Europe/Vienna)

Statistics is, or should be, about scientific investigation
and how to do it better, but many statisticians believe
it is a branch of mathematics.    George E.P. Box

The Bioequivalence and Bioavailability Forum is hosted by
BEBAC Ing. Helmut Schütz
HTML5