adjust train_test_pairs fallbacks to allow (strategy, rows, y) args

src/resampling.jl

@@ -9,9 +9,12 @@ function ==(s1::S, s2::S) where S <: ResamplingStrategy
     return all(getfield(s1, fld) == getfield(s2, fld) for fld in fieldnames(S))
 end
 
+# fallbacks:
 train_test_pairs(s::ResamplingStrategy, rows, X, y, w) =
     train_test_pairs(s, rows, X, y)
 train_test_pairs(s::ResamplingStrategy, rows, X, y) =
+    train_test_pairs(s, rows, y)
+train_test_pairs(s::ResamplingStrategy, rows, y) =
     train_test_pairs(s, rows)
 
 """
@@ -137,13 +140,15 @@ function train_test_pairs(cv::CV, rows)
 end
 
 """
-    stratified_cv = StratifiedCV(; nfolds=6,  shuffle=false, rng=Random.GLOBAL_RNG)
+    stratified_cv = StratifiedCV(; nfolds=6,  
+                                   shuffle=false, 
+                                   rng=Random.GLOBAL_RNG)
 
 Stratified cross-validation resampling strategy, for use in
 `evaluate!`, `evaluate` and in tuning. Applies only to classification
 problems (`OrderedFactor` or `Multiclass` targets).
 
-    train_test_pairs(stratified_cv, rows, X, y)        # X is ignored
+    train_test_pairs(stratified_cv, rows, y)
  
 Returns an `nfolds`-length iterator of `(train, test)` pairs of
 vectors (row indices) where each `train` and `test` is a sub-vector of

test/resampling.jl

@@ -142,15 +142,15 @@ end
                                                 rows, nothing, y)
 
     # check class distribution is preserved in a larger randomized example:
-    N = 3
+    N = 30
     y = shuffle(vcat(fill(:a, N), fill(:b, 2N),
                         fill(:c, 3N), fill(:d, 4N))) |> categorical;
     d = fit(UnivariateFinite, y)
     pairs = MLJ.train_test_pairs(scv, 1:10N, nothing, y)
     folds = vcat(first.(pairs), last.(pairs))
     @test all([fit(UnivariateFinite, y[fold]) ≈ d for fold in folds])
 
-    
+
 end
 
 @testset "weights" begin