Concatenation of NamedArrays while maintaining column names

General Usage
1

Hello everyone, I am trying to figure out how to concatenate NamedArrays while maintaining column names.

using NamedArrays

countriesq = ["USq", "CNq", "JPq", "FRq", "BDq", "ITq", "UKq"]
countriesr = ["USr", "CNr", "JPr", "FRr", "BDr", "ITr", "UKr"]
q = NamedArray(zeros(138, 7))
r = NamedArray(zeros(138, 7))
setnames!(q, countriesq, 2)
setnames!(r, countriesr, 2)

If I do

z = hcat(r, q)

I get numbers 1:14 as column names instead of countriesq and countriesr names.

Thank you very much for your help.

2

I’m not sure what problem’s causing this, or if NamedArrays.jl just doesn’t provide this feature (concatenating while preserving names), but DimensionalData.jl shouldn’t have this same problem.

1 Like
3

NamedArrays.jl don’t support keeping keys when concatenating: Keeping names in *cat · Issue #74 · davidavdav/NamedArrays.jl · GitHub. The issue is related to potentially nonunique keys after the operation.

There are multiple keyed/named array implementations that support this though.
For example, AxisKeys.jl do allow duplicate keys and arbitrary concatenation:

julia> using AxisKeys

julia> q = KeyedArray(rand(3, 2), (["USq", "CNq", "JPq"], 1:2))
2-dimensional KeyedArray(...) with keys:
↓   3-element Vector{String}
→   2-element UnitRange{Int64}
And data, 3×2 Matrix{Float64}:
           (1)         (2)
  ("USq")    0.330996    0.455291
  ("CNq")    0.914666    0.335489
  ("JPq")    0.178773    0.990683

julia> r = KeyedArray(rand(3, 2), (["USq", "CNq", "JPq"], 1:2))
2-dimensional KeyedArray(...) with keys:
↓   3-element Vector{String}
→   2-element UnitRange{Int64}
And data, 3×2 Matrix{Float64}:
           (1)         (2)
  ("USq")    0.600243    0.997009
  ("CNq")    0.887196    0.979967
  ("JPq")    0.171013    0.663274

julia> hcat(r, q)
2-dimensional KeyedArray(...) with keys:
↓   3-element Vector{String}
→   4-element Vector{Int64}
And data, 3×4 Matrix{Float64}:
           (1)         (2)         (1)         (2)
  ("USq")    0.600243    0.997009    0.330996    0.455291
  ("CNq")    0.887196    0.979967    0.914666    0.335489
  ("JPq")    0.171013    0.663274    0.178773    0.990683
1 Like