relax some checks on eltypes for contrasts and add tests (#242)

* relax some checks on eltypes for contrasts and add tests

* actually use CSV

* remove test for different eltype (wrong on 1.0)

* compat bounds for CSV

* drop 1.0, use WeakRefStrings in tests

Co-authored-by: Phillip Alday <me@phillipalday.com>
Co-authored-by: Alex Arslan <ararslan@comcast.net>
Co-authored-by: Phillip Alday <palday@users.noreply.github.com>

Author: 3 people

Project.toml

@@ -23,13 +23,15 @@ ShiftedArrays = "1"
 StatsBase = "0.33.5"
 StatsFuns = "0.9"
 Tables = "0.2, 1"
+WeakRefStrings = "1"
 julia = "1.6"
 
 [extras]
 CategoricalArrays = "324d7699-5711-5eae-9e2f-1d82baa6b597"
 DataFrames = "a93c6f00-e57d-5684-b7b6-d8193f3e46c0"
 Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
+WeakRefStrings = "ea10d353-3f73-51f8-a26c-33c1cb351aa5"
 
 [targets]
-test = ["CategoricalArrays", "DataFrames", "Statistics", "Test"]
+test = ["CategoricalArrays", "DataFrames", "Statistics", "Test", "WeakRefStrings"]

src/contrasts.jl

@@ -120,7 +120,7 @@ end
 # only check equality of matrix, termnames, and levels, and that the type is the
 # same for the contrasts (values are irrelevant).  This ensures that the two
 # will behave identically in creating modelmatrix columns
-Base.:(==)(a::ContrastsMatrix{C,T}, b::ContrastsMatrix{C,T}) where {C<:AbstractContrasts,T} =
+Base.:(==)(a::ContrastsMatrix{C}, b::ContrastsMatrix{C}) where {C<:AbstractContrasts} =
     a.matrix == b.matrix &&
     a.termnames == b.termnames &&
     a.levels == b.levels
@@ -166,18 +166,19 @@ function ContrastsMatrix(contrasts::C, levels::AbstractVector{T}) where {C<:Abst
     # 3. contrast levels missing from data: would have empty columns, generate a
     #    rank-deficient model matrix.
     c_levels = something(DataAPI.levels(contrasts), levels)
-    if eltype(c_levels) != eltype(levels)
-        throw(ArgumentError("mismatching levels types: got $(eltype(levels)), expected " *
-                            "$(eltype(c_levels)) based on contrasts levels."))
-    end
+    
     mismatched_levels = symdiff(c_levels, levels)
     if !isempty(mismatched_levels)
         throw(ArgumentError("contrasts levels not found in data or vice-versa: " *
                             "$mismatched_levels." *
-                            "\n  Data levels: $levels." *
-                            "\n  Contrast levels: $c_levels"))
+                            "\n  Data levels ($(eltype(levels))): $levels." *
+                            "\n  Contrast levels ($(eltype(c_levels))): $c_levels"))
     end
 
+    # do conversion AFTER checking for levels so users get a nice error message
+    # when they've made a mistake with the level types
+    c_levels = convert(Vector{T}, c_levels)
+
     n = length(c_levels)
     if n == 0
         throw(ArgumentError("empty set of levels found (need at least two to compute " *
@@ -187,7 +188,7 @@ function ContrastsMatrix(contrasts::C, levels::AbstractVector{T}) where {C<:Abst
                             "compute contrasts)."))
     end
 
-    # find index of base level. use contrasts.base, then default (1).
+    # find index of base level. use baselevel(contrasts), then default (1).
     base_level = baselevel(contrasts)
     baseind = base_level === nothing ?
               1 :

test/contrasts.jl

@@ -355,5 +355,25 @@
     @testset "Non-unique levels" begin
         @test_throws ArgumentError ContrastsMatrix(DummyCoding(), ["a", "a", "b"])
     end
+
+    @testset "other string types" begin
+        using WeakRefStrings
+
+        using StatsModels: ContrastsMatrix
+        using DataAPI: levels
+
+        x = ["a", "b", "c", "a", "a", "b"]
+        x1 = WeakRefStrings.String1.(x)
+        x1_levs = levels(x1)
+
+        @test issetequal(x, x1)
+
+        c1 = ContrastsMatrix(DummyCoding(), x1_levs)
+        c = ContrastsMatrix(DummyCoding(levels=["a", "b", "c"]), x1_levs)
+        @test c == c1
+        @test eltype(c.levels) == eltype(c1.levels)
+
+        @test_throws ArgumentError ContrastsMatrix(DummyCoding(levels=[1, 2, 3]), x1_levs)
+    end
 
 end