Michael Krabbe Borregaard
Michael Krabbe Borregaard
It was just that in GLM I could only find it defined for LinPredModel
My bad conscience comes back to bite me. I had a working PR ( https://github.com/JuliaData/DataFrames.jl/pull/1160 ) but then never added the extra tests necessary for merging. I'll see if it...
R seems to have one more degree of freedom, possibly because it drops the case where subj is 8. That doesn't explain the discrepancy though. Did you mention on discourse...
I can confirm that this fixes the issue - I guess this can be closed, or changed to a documentation issue on `allowrankdeficient`? @PharmCat the discussion of decomposition type is...
Sorry but I don't see exactly how that would solve the issue.
Oh, so it would need a separate implementation with a different keyword to get it to work?
I don't see why in practice we couldn't just Cholesky decompose the passed covariance matrix, transform the data and pass on to GLM, as suggested here: https://julialang.slack.com/archives/C6821M4KE/p1527713945000352 That's also how...
But I may not see all the complications here. I thought `wts` were variance weights for weighted regression, but I think you're saying they just represent multiple instances of the...
> or it can be a Vector{V} where V is itself a subtype of an AbstractVector and such that eltype(V) and length(V) is defined But that should be true for...
Well. In that case I don't understand what's meant by that part of the docs.