GLM in Julia 0.7

General Usage
1

I am trying to run some regressions in Julia 0.7 + Atom. I added the package GLM.
The command ols = lm(@formula(Y ~ X), data) gives the error

UndefVarError: At_mul_B! not defined

I have seen this issue mentioned somewhere, but it is not clear to me how to fix it.

Thanks

2

Almost all of the common linear algebra routines have moved to the standard library package LinearAlgebra so you need to add to your code using LinearAlgebra, likely an import will be added to GLM in the future to not make you do this yourself.

3

I am using “LinearAlgebra”, but I get the same exact error.

4

GLM is not yet ready for 0.7. See here for progress: Update for 0.7/1.0 by jacobadenbaum · Pull Request #242 · JuliaStats/GLM.jl · GitHub

5

I see. Any recommendations for empirical work (old, panel data, etc)?

6

It’s a bit tough at the moment if you are used to glm in the R world. From my testing, Julia’s GLM.jl isn’t very developed compared to the glm and broom combination in R

7

I just need one package that works, it doesn’t need to be GLM. It seems Julia was not thought for data work.

8

GLM.jl is ready to get a 1.0 compatibility release. You can checkout a branch that works in 1.0 (see Nosferican/GLM.jl or jacobadenbaum/GLM.jl).

1 Like
9

This package works with Julia 1.0 for linear models: https://github.com/matthieugomez/FixedEffectModels.jl