rm some obsolete testing

Author: ablaom

src/MLJModels.jl

@@ -85,4 +85,7 @@ include("model_search.jl")
 # include tools to load model code:
 include("loading.jl")
 
+# include tool for cloning the Model Registry project file:
+include("registry_project.jl")
+
 end # module

src/registry/README.md

@@ -1,3 +1,5 @@
+# The MLJ Model Registry
+
 The *MLJ Model Registry*, also known as the *model metadata database*,
 consists of the files in this directory:
 

test/builtins/Constant.jl

@@ -19,17 +19,10 @@ X = NamedTuple{(:x1,:x2,:x3)}((rand(10), rand(10), rand(10)))
     @test MLJBase.predict(model, fitresult, X)[7].μ ≈ d.μ
     @test MLJBase.predict_mean(model, fitresult, X) ≈ fill(1.5, 10)
 
-    d = MLJModels.info_dict(model)
-    @test d[:input_scitype] == MLJBase.Table
-    @test d[:target_scitype] == AbstractVector{MLJBase.Continuous}
-    @test d[:name] == "ConstantRegressor"
-    @test d[:load_path] == "MLJModels.ConstantRegressor"
-
-    d = MLJModels.info_dict(DeterministicConstantRegressor)
-    @test d[:input_scitype] == MLJBase.Table
-    @test d[:target_scitype] == AbstractVector{MLJBase.Continuous}
-    @test d[:name] == "DeterministicConstantRegressor"
-    @test d[:load_path] == "MLJModels.DeterministicConstantRegressor"
+    @test MLJBase.input_scitype(model) == MLJBase.Table
+    @test MLJBase.target_scitype(model) == AbstractVector{MLJBase.Continuous}
+    @test MLJBase.name(model) == "ConstantRegressor"
+    @test MLJBase.load_path(model) == "MLJModels.ConstantRegressor"
 end
 
 @testset "Classifier" begin
@@ -59,17 +52,11 @@ end
     fitresult, cache, report =  MLJBase.fit(model, 1, X, y, w)
     d = MLJBase.UnivariateFinite([y[1], y[2], y[4]], [1/3, 1/4, 5/12])
 
-    d = MLJModels.info_dict(model)
-    @test d[:input_scitype] == MLJBase.Table
-    @test d[:target_scitype] == AbstractVector{<:MLJBase.Finite}
-    @test d[:name] == "ConstantClassifier"
-    @test d[:load_path] == "MLJModels.ConstantClassifier"
-
-    d = MLJModels.info_dict(DeterministicConstantClassifier)
-    @test d[:input_scitype] == MLJBase.Table
-    @test d[:target_scitype] == AbstractVector{<:MLJBase.Finite}
-    @test d[:name] == "DeterministicConstantClassifier"
-    @test d[:load_path] == "MLJModels.DeterministicConstantClassifier"
+    @test MLJBase.input_scitype(model) == MLJBase.Table
+    @test MLJBase.target_scitype(model) == AbstractVector{<:MLJBase.Finite}
+    @test MLJBase.name(model) == "ConstantClassifier"
+    @test MLJBase.load_path(model) == "MLJModels.ConstantClassifier"
+
 end
 
 end # module

test/builtins/Transformers.jl

@@ -53,7 +53,6 @@ end
     @test round.(Int, MLJBase.transform(stand, f, [0,4,8])) == [-1.0,1.0,3.0]
     @test round.(Int, MLJBase.inverse_transform(stand, f, [-1, 1, 3])) ==
         [0, 4, 8]
-    infos = MLJModels.info_dict(stand)
 
     N = 5
     rand_char = rand("abcefgh", N)
@@ -157,14 +156,6 @@ end
     @test Xnew[4] == X[4]
     @test Xnew[5] == X[5]
 
-    infos = MLJModels.info_dict(stand)
-
-    @test infos[:name] == "Standardizer"
-    @test infos[:input_scitype] ==
-        Union{MLJBase.Table, AbstractVector{<:Continuous}}
-    @test infos[:output_scitype] ==
-        Union{MLJBase.Table, AbstractVector{<:Continuous}}
-
     # univariate case
     stand = Standardizer()
     f, _, _   = MLJBase.fit(stand, 1, [0, 2, 4])
@@ -296,11 +287,6 @@ end
     e = v - MLJBase.inverse_transform(t, f, MLJBase.transform(t, f, v))
     @test sum(abs, e) <= 5000*eps()
 
-    infos = MLJModels.info_dict(t)
-
-    @test infos[:name] == "UnivariateBoxCoxTransformer"
-    @test infos[:input_scitype] == AbstractVector{MLJBase.Continuous}
-    @test infos[:output_scitype] == AbstractVector{MLJBase.Continuous}
 end
 
 
@@ -383,12 +369,6 @@ end
     f, = MLJBase.fit(t, 0, X)
     @test_throws Exception MLJBase.transform(t, f, Xmiss)
 
-    infos = MLJModels.info_dict(t)
-
-    @test infos[:name] == "OneHotEncoder"
-    @test infos[:input_scitype] == MLJBase.Table
-    @test infos[:output_scitype] == MLJBase.Table
-
     # test the work on missing values
     X = (name   = categorical(["Ben", "John", "Mary", "John", missing], ordered=true),
          height = [1.85, 1.67, 1.5, 1.67, 1.56],

test/metadata.jl

@@ -4,7 +4,7 @@ using Test
 using MLJModels
 using MLJBase
 
-METADATA = MLJModels.METADATA
+METADATA = MLJModels.REGISTRY_METADATA
 pca = MLJModels.Handle("PCA", "MultivariateStats")
 cnst = MLJModels.Handle("ConstantRegressor", "MLJModels")
 i = MLJModels.info_given_handle(METADATA)[pca]

test/registry.jl

@@ -14,7 +14,7 @@ using Suppressor
 
     # make a clone of the MLJModel registry, to test `registry_project` method:
     filename, stream = mktemp()
-    for line in MLJModels.Registry.registry_project()
+    for line in MLJModels.registry_project()
         write(stream, line*"\n")
     end
     close(stream)