Logistic regression with GLM : not working with Categorical Arrays?

Specific Domains Statistics
,
1

Hello there!
I’m trying to do logistic regression using GLM Package. My Y-variable is of type categorical array, the predictor is the usual Float64-array. When I’m trying to make a model with @formula(Y ~ X), I’ve got a mistake: MethodError: no method matching zero(::Type{CategoricalValue{Int64,UInt8}})
So is it possible to make logregr work without manual encoding of categorical array?

2

See GLM.jl/issues/240.
For a quick regression you could use a draft I have at Econometrics.jl (I still need to port a few things to it that I have in other packages, but the code there does it).

1 Like