Douglas Bates

The code for that method starts around line 130 of https://github.com/lrberge/fixest/blob/master/src/lm_related.cpp The linear algebra code we are using in Julia is LAPACK/OpenBLAS or MKL - code that has been developed...

`GlmResp` and `LmResp` were formulated a long time ago and a redesign may be warranted. There are several competing goals - generality, ease-of-use, efficiency for the most common cases -...

Having done some more benchmarking it looks like I am trying to solve a non-problem here. I keep convincing myself that the bottleneck in fitting a glm must be the...

See https://github.com/JuliaStats/StatsModels.jl/pull/28

The reason for having an intermediate step of Original Data Frame -> Model Frame -> Model Matrix in R was to isolate operations like skipping rows with missing data, only...

That was also the reason for the bizarre "Standard Non-Standard R Evaluation Model" that I mentioned in a slack discussion with @kleinschmidt . The idea was to have all the...

The example given by the OP is actually a case of `LinearModel`s being fit as `GeneralizedLinearModel`s so, in theory, there is an F-test. However, it is simplest to just fit...

I have lost track of the purpose here but there are a couple of observations that might help. - I'm not sure why fitting a null model needs a short-circuit...

That is sort-of what I do in [`MixedModels`](https://github.com/dmbates/MixedModels.jl) where a generalized linear mixed model (GLMM) is implemented in that package but using the `GLM.GlmResp` type. If the truth be told,...

On a somewhat related note, it would be helpful to allow `DataArrays.DataArray{Bool, 1}` in the second and subsequent arguments. At least I found that I needed to write an expression...