add some tests

Author: ablaom

src/LearnDataFrontEnds.jl

@@ -4,7 +4,8 @@
 Module providing the following commonly applicable data front ends for implementations of
 the [LearnAPI.jl](https://juliaai.github.io/LearnAPI.jl/dev/) interface:
 
-- [`Saffron`](@ref): good for most supervised regressors operating on structured data
+- [`Saffron`](@ref): good for most supervised leaners, typically regressors, operating on
+  structured data
 
 - [`Sage`](@ref): good for most supervised classifiers operating on structured data
 

src/backends.jl

@@ -22,12 +22,13 @@ Additionally, when `observations = fit(learner, data)` and the
 been implemented, one has:
 
 - `observations.target`: length `n` target vector (`multitarget=false`) or size `(q, n)`
-  target matrix (`multivariate=true`)
+  target matrix (`multivariate=true`); this array has the same element type as the
+  user-provided one in the `Saffron` case
 
 # Specific to `Sage`
 
-If [`Sage`](@ref)`(multitarget=..., code_type=...)` has been implemented, the above
-target representation has an integer element type controlled by `code_type`, and we
+If [`Sage`](@ref)`(multitarget=..., code_type=...)` has been implemented, then
+`observations.target` has an integer element type controlled by `code_type`, and we
 additionally have:
 
 - `observations.classes`: A categorical vector of the ordered target classes, as actually

src/constants.jl

@@ -11,3 +11,8 @@ const DOC_FORMULAS = """
 const ERR_BAD_LEVELS = ArgumentError(
     "`levels` must be one of these: `:raw`, `:int`, `:small`. "
 )
+
+const ERR_EXPECTED_CATEGORICAL = ArgumentError(
+    "Targets (or target columns of a table) must be `CategoricalArray`, "*
+        "or subarrays thereof. "
+)

src/saffron.jl

@@ -1,9 +1,10 @@
 """
     Saffron(; multitarget=false, view=false)
 
-A LearnAPI.jl data front end implemented for some supervised regressors consuming
-structured data. If `learner` implements this front end, then `data` in the call
-[`LearnAPI.fit`](@extref)`(learner, data)` can take any of the following forms:
+A LearnAPI.jl data front end implemented for some supervised learners, typically
+regressors, consuming structured data. If `learner` implements this front end, then `data`
+in the call [`LearnAPI.fit`](@extref)`(learner, data)` can take any of the following
+forms:
 
 - `(X, y)`, where `X` is a feature matrix or table and `y` is a target vector, matrix or
   table
@@ -145,6 +146,10 @@ finalize(x, names, y, ::Saffron{<:Any,<:Any,IntCode}) =
 finalize(x, names, y, ::Saffron{<:Any,<:Any,SmallIntCode}) =
     finalize(x, names, y, CategoricalArrays.refcode)
 function finalize(x, names, y, int)  # here `int` is `levelcode` or `refcode` function
+    y isa Union{
+        CategoricalArrays.CategoricalArray,
+        SubArray{<:Any, <:Any, <:CategoricalArrays.CategoricalArray},
+    } || throw(ERR_EXPECTED_CATEGORICAL)
     l = LearnDataFrontEnds.classes(y)
     u = unique(y)
     mask = map(in(u), l)

src/sage.jl

@@ -73,6 +73,9 @@ training.
 
 # Implementation
 
+If a core algorithm is happy to work with a `CategoricalArray` target, without
+integer-encoding it, consider using the [`Saffron`](@ref) frontend instead.
+
 For learners of type `MyLearner`, with `LearnAPI.fit(::MyLearner, data)` returning
 objects of type `MyModel`, implement the `Sage` data front
 by making these declarations:

test/saffron.jl

@@ -10,6 +10,10 @@ import StatsModels.@formula
 
 # include("_some_learners.jl")
 
+@testset "bad `Saffron` constructor argument" begin
+    @test_throws LearnDataFrontEnds.ERR_BAD_LEVELS Saffron(code_type=:junk)
+end
+
 n = 20
 rng = StableRNG(345)
 c, t, a, t2 = randn(rng, n), rand(rng, n), rand(rng, n), rand(rng, n)

test/sage.jl

@@ -171,6 +171,13 @@ end
     @test CA.levels(yy) == CA.levels(y)
 end
 
+@testset "informative error if fitobs supplied non-categorical target" begin
+    @test_throws(
+        LearnDataFrontEnds.ERR_EXPECTED_CATEGORICAL,
+        fitobs("learner", ((; x=fill(1, 3)), [1 ,2, 3]), Sage())
+    )
+end
+
 # from test/_some_learners.jl:
 learner = LearnerReportingNames()
 model = fit(learner, X)