fixes #98

src/make_cue_matrix.jl

@@ -412,6 +412,15 @@ function make_combined_cue_matrix(
 )
 
     data_combined = copy(data_train)
+    data_val = copy(data_val)
+    for col in names(data_combined)
+        if any(map(x -> x isa InlineString, data_combined[!,col]))
+            data_combined[!, col] = String.(data_combined[!,col])
+        end
+        if any(map(x -> x isa InlineString, data_val[!,col]))
+            data_val[!, col] = String.(data_val[!, col])
+        end
+    end
     append!(data_combined, data_val)
 
     cue_obj_combined = make_cue_matrix(

src/make_semantic_matrix.jl

@@ -984,6 +984,15 @@ function make_combined_L_matrix(
 )
 
     data_combined = copy(data_train)
+    data_val = copy(data_val)
+    for col in names(data_combined)
+        if any(map(x -> x isa InlineString, data_combined[!,col]))
+            data_combined[!, col] = String.(data_combined[!,col])
+        end
+        if any(map(x -> x isa InlineString, data_val[!,col]))
+            data_val[!, col] = String.(data_val[!, col])
+        end
+    end
     append!(data_combined, data_val)
 
     make_L_matrix(

test/data/latin_train.csv

@@ -0,0 +1,4 @@
+Column1,Word,Lexeme,Person,Number,Tense,Voice,Mood
+1,vocoo,vocare,p1,sg,present,active,ind
+2,vocaas,vocare,p2,sg,present,active,ind
+3,vocat,vocare,p3,sg,present,active,ind

test/data/latin_val.csv

@@ -0,0 +1,7 @@
+Column1,Word,Lexeme,Person,Number,Tense,Voice,Mood
+10,clamaamus,clamare,p1,pl,present,active,ind
+11,clamaatis,clamare,p2,pl,present,active,ind
+12,clamant,clamare,p3,pl,present,active,ind
+13,vocaabam,vocare,p1,sg,past,active,ind
+14,vocaabaas,vocare,p2,sg,past,active,ind
+15,vocaabat,vocare,p3,sg,past,active,ind

test/data/semantic_train.csv

@@ -0,0 +1,3 @@
+X,Y,Z
+A,MAAAAAAA,P
+B,FAAAA,P

test/data/semantic_val.csv

@@ -0,0 +1,4 @@
+X,Y,Z
+C,F,S
+D,M,
+E,,P

test/make_cue_matrix_tests.jl

@@ -98,4 +98,28 @@ end
     @test cue_obj_val.f2i["oco"] == 3
     @test cue_obj_train.A[1, 3] == 0
     @test cue_obj_val.A[1, 3] == 0
+
+    latin_train = DataFrame(CSV.File(joinpath(
+        @__DIR__,
+        "data",
+        "latin_train.csv",
+    )))
+    latin_val = DataFrame(CSV.File(joinpath(
+        @__DIR__,
+        "data",
+        "latin_val.csv",
+    )))
+
+    # check that indeed there are columns with differing string types
+    println(typeof(latin_train.Word) != typeof(latin_val.Word))
+
+    # but make combined cue matrix still runs
+    cue_obj_train, cue_obj_val = JudiLing.make_combined_cue_matrix(
+        latin_train,
+        latin_val,
+        grams = 3,
+        target_col = :Word,
+        tokenized = false,
+        keep_sep = false,
+    )
 end

test/make_semantic_matrix_tests.jl

@@ -184,6 +184,31 @@ end
     @test S_val[1, :] == L.L[L.f2i["F"], :]
     @test S_val[2, :] == L.L[L.f2i["M"], :]
     @test S_val[3, :] == zeros(Float64, n_features)
+
+    train = DataFrame(CSV.File(joinpath(
+        @__DIR__,
+        "data",
+        "latin_train.csv",
+    )))
+    val = DataFrame(CSV.File(joinpath(
+        @__DIR__,
+        "data",
+        "latin_val.csv",
+    )))
+
+    # check that indeed there are columns with differing string types
+    println(typeof(train.Word) != typeof(val.Word))
+
+    # but make combined cue matrix still runs
+    S_train, S_val = JudiLing.make_combined_S_matrix(
+        train,
+        val,
+        ["Lexeme"],
+        ncol = n_features,
+        seed = seed,
+        isdeep = true,
+        add_noise = false,
+    )
 end
 
 @testset "make L matrix" begin