Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

don't fit a GLM when fixed effects are empty #657

Merged
merged 5 commits into from
Oct 26, 2022
Merged

don't fit a GLM when fixed effects are empty #657

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
4508a18
don't fit a GLM when fixed effects are empty
palday Oct 25, 2022
e8fd451
NEW entry
palday Oct 25, 2022
827964f
format
palday Oct 25, 2022
072a3e9
JuliaFormatter
palday Oct 25, 2022
264b194
Update src/generalizedlinearmixedmodel.jl
palday Oct 26, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 6 additions & 0 deletions NEWS.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,3 +1,8 @@


* Don't fit a GLM internally during construction of GLMM when the fixed effects are empty (better compatibility with
`dropcollinear` kwarg in newer GLM.jl) [#657]

MixedModels v4.8.0 Release Notes
==============================
* Allow predicting from a single observation, as long as `Grouping()` is used for the grouping variables. The simplified implementation of `Grouping()` also removes several now unnecessary `StatsModels` methods that should not have been called directly by the user. [#653]
Expand Down Expand Up @@ -380,3 +385,4 @@ Package dependencies
[#648]: https://github.com/JuliaStats/MixedModels.jl/issues/648
[#651]: https://github.com/JuliaStats/MixedModels.jl/issues/651
[#653]: https://github.com/JuliaStats/MixedModels.jl/issues/653
[#657]: https://github.com/JuliaStats/MixedModels.jl/issues/657
4 changes: 3 additions & 1 deletion docs/make.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,4 +19,6 @@ makedocs(;
],
)

deploydocs(;repo = "github.com/JuliaStats/MixedModels.jl.git", push_preview = true, devbranch = "main")
deploydocs(;
repo="github.com/JuliaStats/MixedModels.jl.git", push_preview=true, devbranch="main"
)
14 changes: 12 additions & 2 deletions src/generalizedlinearmixedmodel.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -403,8 +403,18 @@ function GeneralizedLinearMixedModel(
constresponse || updateL!(LMM)
# fit a glm to the fixed-effects only
T = eltype(LMM.Xymat)
gl = glm(LMM.X, y, d, l; wts=convert(Vector{T}, wts), offset=convert(Vector{T}, offset))
β = coef(gl)
# newer versions of GLM (>1.8.0) have a kwarg dropcollinear=true
# which creates problems for the empty fixed-effects case during fitting
# so just don't allow fitting
# XXX unfortunately, this means we have double-rank defiency detection
# TODO: construct GLM by hand so that we skip collinearity checks
# TODO: extend this so that we never fit a GLM when initializing from LMM
dofit = size(LMM.X, 2) != 0 # GLM.jl kwarg
gl = glm(LMM.X, y, d, l;
wts=convert(Vector{T}, wts),
dofit,
offset=convert(Vector{T}, offset))
palday marked this conversation as resolved.
Show resolved Hide resolved
β = dofit ? coef(gl) : T[]
u = [fill(zero(eltype(y)), vsize(t), nlevs(t)) for t in LMM.reterms]
# vv is a template vector used to initialize fields for AGQ
# it is empty unless there is a single random-effects term
Expand Down