ModelMatrix(@formula(0~1), (x = 1:4, y = 5:8)) fails

[behinger]

I expected this to return a "empty" ModelMatrix (it is not really empty, as I wanted to get only the intercept via modelmatrix(mf)

MWE

df = (x = 1:4,)
mf = ModelFrame(@formula(x ~ 1 ), df) # works
mf = ModelFrame(@formula(0 ~ 1 +x ), df) # works
mf = ModelFrame(@formula(0 ~ 1 ), df) # throws error

The following error is thrown.

ERROR: ArgumentError: collection must be non-empty
Stacktrace:
 [1] first(itr::NamedTuple{(), Tuple{}})
   @ Base ./abstractarray.jl:425
 [2] missing_omit(d::NamedTuple{(), Tuple{}})
   @ StatsModels ~/.julia/packages/StatsModels/fK0P3/src/modelframe.jl:57
 [3] missing_omit(data::NamedTuple{(:x,), Tuple{UnitRange{Int64}}}, formula::FormulaTerm{ConstantTerm{Int64}, ConstantTerm{Int64}})
   @ StatsModels ~/.julia/packages/StatsModels/fK0P3/src/modelframe.jl:70
 [4] ModelFrame(f::FormulaTerm{ConstantTerm{Int64}, ConstantTerm{Int64}}, data::NamedTuple{(:x,), Tuple{UnitRange{Int64}}}; model::Type{StatisticalModel}, contrasts::Dict{Symbol, Any})
   @ StatsModels ~/.julia/packages/StatsModels/fK0P3/src/modelframe.jl:81
 [5] ModelFrame(f::FormulaTerm{ConstantTerm{Int64}, ConstantTerm{Int64}}, data::NamedTuple{(:x,), Tuple{UnitRange{Int64}}})
   @ StatsModels ~/.julia/packages/StatsModels/fK0P3/src/modelframe.jl:73
 [6] top-level scope
   @ REPL[175]:1

[kleinschmidt]

Thanks for the report, definitely not an edge case I'd anticipated 😁 I think it's a pretty straightforward fix to add a check that d is non-empty in missing_omit(d) (and return it as-is if it is) if you'd like to make a PR!