Douglas Bates
Douglas Bates
Sounds good. Thanks for looking into that.
@palday My plan is to make the show method for a `MixedModelProfile` just show the `tbl` field. I like the way the Table is set up now. I still need...
@palday Do you have any insight into the test failures? As far as I can see I haven't modified anything that should cause that test to fail? It may be...
> @dmbates I bet this is from one of the recent changes in GLM.jl .... I have an idea how to fix and will ping you soon slightly_smiling_face @palday That...
@palday I am leaning towards profiling the standard deviations of the random effects separately from the correlation parameters. I think your introduction of the capability to set a value of...
@palday I have been thinking about the fields of the MixedModelProfile type. I feel it would be best to keep a shallow copy of the LinearMixedModel object itself in the...
cc: @JeffBezanson, @StefanKarpinski Do you think this would be straightforward to accomplish, assuming that the chunks of `src` and `dest` are aligned?
I'm not sure you would want that particular formula for the GLM case. As the name implies, the Iteratively Reweighted Least Squares (IRLS) algorithm iterates on the `W` matrix. Would...
It might be a good idea to benchmark the singular value decomposition. As an iterative method it is definitely more expensive than direct methods like Cholesky and QR but some...
@pdeffebach It appears that your R code is not using the Cholesky factorization in R but a specific method, which from the name I expect is coded in C++, called...