NestedTerms.jl is a concept package for correct nested terms fitting .
Install
import Pkg
Pkg.add(url = "https://github.com/PharmCat/NestedTerms.jl.git")
Example
using NestedTerms
using StatsModels, CSV, DataFrames, CategoricalArrays
rds = CSV.File(joinpath(dirname(pathof(NestedTerms)), "..", "test", "csv", "rds12x2.csv")) |> DataFrame # DataFrame with nested factor Subject(Sequence)
transform!(rds, :Sequence => categorical, renamecols=false)
transform!(rds, :Subject => categorical, renamecols=false)
# Nested term exploit function term in @formula macros
formula = @formula(Var ~ (Subject in Sequence))
# Replace FunctionTerm to NestedTerm
formula = NestedTerms.process_nt(formula)
# Make schema and apply_schema
sch = schema(formula, rds)
as = apply_schema(formula, sch) $ upper level factor not added to the formula (but it shoud be there)
# Get model matrix for nested term
rmf, lmf = modelcols(as, rds)
Unsolvable problems
Can’t get real coefnames and levels until modelcols not called. apply_schema not able to modify NestedTerm because have no appropriate information.
Question!
Is any way to workaround to get information about data or one more modification should be added after apply_schema(formula, sch)?