Translate R Regression Model Formulae to 'Julia' Syntax

Metaprogramming utilities for converting R regression model formulae to equivalents in 'Julia' , via modifications to the abstract syntax tree. Supports translations in zero correlation random effects syntax, protection of expressions to be evaluated as-is, interaction terms, and more. Accepts strings or R formula objects and returns modified R formula objects where possible (or a modified string, if not a valid formula in R).

JuliaFormulae

A utility package for converting R regression model formula to Julia GLM.jl and MixedModels.jl syntax.

See {jlme} for an example application.

Installation

# install.packages("devtools")
devtools::install_github("yjunechoe/JuliaFormulae")

library(JuliaFormulae)

Supported conversions

• Zero correlation:

  julia_formula(y ~ x + (x || g))
  #> y ~ x + zerocorr(x | g)
  

• Protection:

  julia_formula(y ~ x + I(x * 2))
  #> y ~ x + protect(x * 2)
  

• Interaction terms:

  julia_formula(y ~ a:b)
  #> y ~ a & b
  

Example

julia_formula(
  y ~ a + I(a * 2) + b + a:b  + (a || g) + (b | g)
)
#> y ~ a + protect(a * 2) + b + (a & b) + zerocorr(a | g) + (b | 
#>     g)