How do I a logistic regression without the intercept in GLM.jl?

Specific Domains Machine Learning
1

I can’t figure out how to fit a model without the intercept

The documentatioin just says the intercept is fitted by default but doesn’t show how too do it without the intercept? is that even possible?

image
image1634×675 71.1 KB

2 
using StatsModels

# add term(-1)

#e.g.

formula = Term(:target) ~ term(-1) + sum(Term.(Symbol.(names(last_2_years4[:, Not(:target)]))))
3

I answered in the other thread - here are two options with or without the @formula macro:

julia> using GLM

julia> df = (x = rand(10), y = rand(Bool, 10));

julia> glm(@formula(y ~ 0 + x), df, Bernoulli(), LogitLink())
StatsModels.TableRegressionModel{GeneralizedLinearModel{GLM.GlmResp{Vector{Float64}, Bernoulli{Float64}, LogitLink}, GLM.DensePredChol{Float64, LinearAlgebra.CholeskyPivoted{Float64, Matrix{Float64}, Vector{Int64}}}}, Matrix{Float64}}

y ~ 0 + x

Coefficients:
──────────────────────────────────────────────────────────────
      Coef.  Std. Error      z  Pr(>|z|)  Lower 95%  Upper 95%
──────────────────────────────────────────────────────────────
x  -1.32166     1.47523  -0.90    0.3703   -4.21306    1.56975
──────────────────────────────────────────────────────────────

julia> glm(reshape(df.x, 10, 1), df.y, Bernoulli(), LogitLink())
GeneralizedLinearModel{GLM.GlmResp{Vector{Float64}, Bernoulli{Float64}, LogitLink}, GLM.DensePredChol{Float64, LinearAlgebra.CholeskyPivoted{Float64, Matrix{Float64}, Vector{Int64}}}}:

Coefficients:
───────────────────────────────────────────────────────────────
       Coef.  Std. Error      z  Pr(>|z|)  Lower 95%  Upper 95%
───────────────────────────────────────────────────────────────
x1  -1.32166     1.47523  -0.90    0.3703   -4.21306    1.56975
───────────────────────────────────────────────────────────────
4

I want the terms to be constructed programmatically

5

Ah, that wasn’t clear from this question or the other thread, where the discussion was around using the lm function.

If you want to translate something you know how to write in @formula syntax into code, expanding the macro is helpful:

julia> @macroexpand(@formula(y ~ 0 + x))
:(StatsModels.Term(:y) ~ StatsModels.ConstantTerm(0) + StatsModels.Term(:x))

Also, term is the preferred function for construcing Term objects, as it will accept strings so you can do term.(names(df[!, Not(:target)]))

1 Like