Merge pull request #1127 from JuliaAI/constructor

Add model wrappers to the Model Browser

Project.toml

@@ -8,6 +8,7 @@ CategoricalArrays = "324d7699-5711-5eae-9e2f-1d82baa6b597"
 ComputationalResources = "ed09eef8-17a6-5b46-8889-db040fac31e3"
 Distributed = "8ba89e20-285c-5b6f-9357-94700520ee1b"
 Distributions = "31c24e10-a181-5473-b8eb-7969acd0382f"
+FeatureSelection = "33837fe5-dbff-4c9e-8c2f-c5612fe2b8b6"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 MLJBalancing = "45f359ea-796d-4f51-95a5-deb1a414c586"
 MLJBase = "a7f614a8-145f-11e9-1d2a-a57a1082229d"
@@ -31,12 +32,13 @@ Tables = "bd369af6-aec1-5ad0-b16a-f7cc5008161c"
 CategoricalArrays = "0.8,0.9, 0.10"
 ComputationalResources = "0.3"
 Distributions = "0.21,0.22,0.23, 0.24, 0.25"
+FeatureSelection = "0.1.1"
 MLJBalancing = "0.1"
 MLJBase = "1"
 MLJEnsembles = "0.4"
 MLJFlow = "0.5"
 MLJIteration = "0.6"
-MLJModels = "0.16"
+MLJModels = "0.17"
 MLJTestIntegration = "0.5.0"
 MLJTuning = "0.8"
 OpenML = "0.2,0.3"
@@ -89,4 +91,41 @@ SymbolicRegression = "8254be44-1295-4e6a-a16d-46603ac705cb"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 
 [targets]
-test = ["BetaML", "CatBoost", "EvoLinear", "EvoTrees", "Imbalance", "InteractiveUtils", "LightGBM", "MLJClusteringInterface", "MLJDecisionTreeInterface", "MLJFlux", "MLJGLMInterface", "MLJLIBSVMInterface", "MLJLinearModels", "MLJMultivariateStatsInterface", "MLJNaiveBayesInterface", "MLJScikitLearnInterface", "MLJTSVDInterface", "MLJTestInterface", "MLJTestIntegration", "MLJText", "MLJXGBoostInterface", "Markdown", "NearestNeighborModels", "OneRule", "OutlierDetectionNeighbors", "OutlierDetectionPython", "ParallelKMeans", "PartialLeastSquaresRegressor", "PartitionedLS", "SelfOrganizingMaps", "SIRUS", "SymbolicRegression", "StableRNGs", "Suppressor", "Test"]
+test = [
+    "BetaML",
+    "CatBoost",
+    "EvoLinear",
+    "EvoTrees",
+    "Imbalance",
+    "InteractiveUtils",
+    "LightGBM",
+    "MLJClusteringInterface",
+    "MLJDecisionTreeInterface",
+    "MLJFlux",
+    "MLJGLMInterface",
+    "MLJLIBSVMInterface",
+    "MLJLinearModels",
+    "MLJMultivariateStatsInterface",
+    "MLJNaiveBayesInterface",
+    "MLJScikitLearnInterface",
+    "MLJTSVDInterface",
+    "MLJTestInterface",
+    "MLJTestIntegration",
+    "MLJText",
+    "MLJXGBoostInterface",
+    "Markdown",
+    "NearestNeighborModels",
+    "OneRule",
+    "OutlierDetectionNeighbors",
+    "OutlierDetectionPython",
+    "ParallelKMeans",
+    "PartialLeastSquaresRegressor",
+    "PartitionedLS",
+    "SelfOrganizingMaps",
+    "SIRUS",
+    "SymbolicRegression",
+    "StableRNGs",
+    "Suppressor",
+    "Test",
+]
+

docs/ModelDescriptors.toml

@@ -11,7 +11,9 @@ AutoEncoder_BetaML = ["dimension_reduction"]
 BM25Transformer_MLJText = ["encoders", "text_analysis"]
 BaggingClassifier_MLJScikitLearnInterface = ["classification", "ensemble_models"]
 BaggingRegressor_MLJScikitLearnInterface = ["regression", "ensemble_models"]
-BalancedBaggingClassifier_MLJBalancing = ["class_imbalance", "classification"]
+BalancedBaggingClassifier_MLJBalancing = ["class_imbalance", "classification", "meta_algorithms"]
+BinaryThresholdPredictor_MLJModels = ["meta_algorithms", "classification"]
+BalancedModel_MLJBalancing = ["class_imbalance", "meta_algorithms"]
 BayesianLDA_MultivariateStats = ["dimension_reduction", "classification", "Bayesian_models"]
 BayesianLDA_MLJScikitLearnInterface = ["dimension_reduction", "classification", "Bayesian_models"]
 BayesianQDA_MLJScikitLearnInterface = ["dimension_reduction", "classification", "Bayesian_models"]
@@ -52,6 +54,7 @@ ElasticNetCVRegressor_MLJScikitLearnInterface = ["regression"]
 ElasticNetRegressor_MLJLinearModels = ["regression"]
 ElasticNetRegressor_MLJScikitLearnInterface = ["regression"]
 ENNUndersampler_Imbalance = ["class_imbalance"]
+EnsembleModel_MLJEnsembles = ["ensemble_models", "meta_algorithms"]
 EpsilonSVR_LIBSVM = ["regression"]
 EvoLinearRegressor_EvoLinear = ["regression"]
 EvoTreeClassifier_EvoTrees = ["classification", "ensemble_models", "iterative_models"]
@@ -89,6 +92,7 @@ IForestDetector_OutlierDetectionPython = ["outlier_detection"]
 ImageClassifier_MLJFlux = ["classification", "image_processing", "iterative_models"]
 INNEDetector_OutlierDetectionPython = ["outlier_detection"]
 InteractionTransformer_MLJModels = ["static_models"]
+IteratedModel_MLJIteration = ["iterative_models", "meta_algorithms"]
 KDEDetector_OutlierDetectionPython = ["outlier_detection"]
 KMeansClusterer_BetaML = ["clustering"]
 KMeans_Clustering = ["clustering", "dimension_reduction", ]
@@ -104,7 +108,7 @@ KNeighborsClassifier_MLJScikitLearnInterface = ["classification"]
 KNeighborsRegressor_MLJScikitLearnInterface = ["regression"]
 KPLSRegressor_PartialLeastSquaresRegressor = ["regression"]
 KernelPCA_MultivariateStats = ["dimension_reduction", ]
-KernelPerceptronClassifier_BetaML = ["classification"]
+KernelPerceptronClassifier_BetaML = ["classification", "neural networks"]
 LADRegressor_MLJLinearModels = ["regression"]
 LDA_MultivariateStats = ["classification", "dimension_reduction", ]
 LGBMClassifier_LightGBM = ["classification", "ensemble_models", "iterative_models"]
@@ -146,14 +150,14 @@ MultitargetGaussianMixtureRegressor_BetaML = ["regression", "distribution_fitter
 MultitargetKNNClassifier_NearestNeighborModels = ["classification"]
 MultitargetKNNRegressor_NearestNeighborModels = ["regression"]
 MultitargetLinearRegressor_MultivariateStats = ["regression"]
-MultitargetNeuralNetworkRegressor_BetaML = ["regression"]
-MultitargetNeuralNetworkRegressor_MLJFlux = ["regression", "iterative_models"]
+MultitargetNeuralNetworkRegressor_BetaML = ["regression", "neural networks"]
+MultitargetNeuralNetworkRegressor_MLJFlux = ["regression", "iterative_models", "neural networks"]
 MultitargetRidgeRegressor_MultivariateStats = ["regression"]
 MultitargetSRRegressor_SymbolicRegression = ["regression"]
-NeuralNetworkClassifier_BetaML = ["classification"]
-NeuralNetworkClassifier_MLJFlux = ["classification", "iterative_models"]
-NeuralNetworkRegressor_BetaML = ["regression"]
-NeuralNetworkRegressor_MLJFlux = ["regression", "iterative_models"]
+NeuralNetworkClassifier_BetaML = ["classification", "neural networks"]
+NeuralNetworkClassifier_MLJFlux = ["classification", "iterative_models", "neural networks"]
+NeuralNetworkRegressor_BetaML = ["regression", "neural networks"]
+NeuralNetworkRegressor_MLJFlux = ["regression", "iterative_models", "neural networks"]
 NuSVC_LIBSVM = ["classification"]
 NuSVR_LIBSVM = ["regression"]
 OCSVMDetector_OutlierDetectionPython = ["outlier_detection"]
@@ -171,8 +175,9 @@ PartLS_PartitionedLS = ["regression"]
 PassiveAggressiveClassifier_MLJScikitLearnInterface = ["classification"]
 PassiveAggressiveRegressor_MLJScikitLearnInterface = ["regression"]
 PegasosClassifier_BetaML = ["classification"]
-PerceptronClassifier_BetaML = ["classification", "iterative_models"]
-PerceptronClassifier_MLJScikitLearnInterface = ["classification", "iterative_models"]
+PerceptronClassifier_BetaML = ["classification", "iterative_models", "neural networks"]
+PerceptronClassifier_MLJScikitLearnInterface = ["classification", "iterative_models", "neural networks"]
+Pipeline_MLJBase = ["meta_algorithms"]
 ProbabilisticNuSVC_LIBSVM = ["classification"]
 ProbabilisticSGDClassifier_MLJScikitLearnInterface = ["classification"]
 ProbabilisticSVC_LIBSVM = ["classification"]
@@ -190,6 +195,7 @@ RandomForestImputer_BetaML = ["missing_value_imputation", "ensemble_models", "it
 RandomForestRegressor_BetaML = ["regression", "ensemble_models", "iterative_models"]
 RandomForestRegressor_DecisionTree = ["regression", "ensemble_models", "iterative_models"]
 RandomForestRegressor_MLJScikitLearnInterface = ["regression", "ensemble_models", "iterative_models"]
+Resampler_MLJBase = ["meta_algorithms"]
 RidgeCVClassifier_MLJScikitLearnInterface = ["classification"]
 RidgeCVRegressor_MLJScikitLearnInterface = ["classification"]
 RidgeClassifier_MLJScikitLearnInterface = ["classification"]
@@ -210,6 +216,7 @@ StableForestClassifier_SIRUS = ["classification"]
 StableForestRegressor_SIRUS = ["regression"]
 StableRulesClassifier_SIRUS = ["classification"]
 StableRulesRegressor_SIRUS = ["regression"]
+Stack_MLJBase = ["meta_algorithms", "ensemble_models"]
 SVC_LIBSVM = ["classification"]
 SVMClassifier_MLJScikitLearnInterface = ["classification"]
 SVMLinearClassifier_MLJScikitLearnInterface = ["classification"]
@@ -222,9 +229,11 @@ SpectralClustering_MLJScikitLearnInterface = ["clustering", "static_models"]
 Standardizer_MLJModels = ["encoders"]
 SubspaceLDA_MultivariateStats = ["classification", "dimension_reduction"]
 TomekUndersampler_Imbalance = ["class_imbalance"]
+TunedModel_MLJTuning = ["meta_algorithms"]
 TSVDTransformer_TSVD = ["dimension_reduction"]
 TfidfTransformer_MLJText = ["encoders", "text_analysis"]
 TheilSenRegressor_MLJScikitLearnInterface = ["regression"]
+TransformedTargetModel_MLJBase = ["meta_algorithms", "outlier_detection"]
 UnivariateBoxCoxTransformer_MLJModels = ["encoders"]
 UnivariateDiscretizer_MLJModels = ["encoders"]
 UnivariateFillImputer_MLJModels = ["missing_value_imputation"]

docs/make.jl

@@ -15,6 +15,7 @@ import MLJ.MLJModels
 import MLJ.MLJEnsembles
 import MLJ.ScientificTypes
 import MLJ.MLJBalancing
+import MLJ.FeatureSelection
 import ScientificTypesBase
 import Distributions
 using CategoricalArrays
@@ -37,7 +38,7 @@ isempty(problems) || error(
 # compose the individual model docstring pages:
 @info "Getting individual model docstrings from the registry and generating "*
     "pages for them, written at /docs/src/models/ ."
-for model in models()
+for model in models(wrappers=true)
     write_page(model)
 end
 
@@ -118,6 +119,7 @@ makedocs(
         IterationControl,
         CategoricalDistributions,
         StatisticalMeasures,
+        FeatureSelection,
     ],
     pages    = pages,
     warnonly = [:cross_references, :missing_docs],

docs/model_docstring_tools.jl

@@ -5,7 +5,7 @@ const PATH_TO_MODEL_DOCS = joinpath(@__DIR__, "src", "models")
 """
     remove_doc_refs(str::AbstractString)
 
-Removes `@ref` references from `str. For example, a substring of the form
+Removes `@ref` references from `str`. For example, a substring of the form
 "[`some.thing_like_this123!`](@ref)" is replaced with "`some.thing_like_this123!`".
 
 """
@@ -27,8 +27,8 @@ handle(model) = model.name*"_"*model.package_name
 **Private method.**
 
 Compose and write to file the documentation page for `model`. Here `model` is an entry in
-the MLJ Model Registry, i.e., an element of `MLJModels.models()`. The file name has the
-form `"ModelName_PackageName.md"`, for example,
+the MLJ Model Registry, i.e., an element of `MLJModels.models(; wrappers=true)`. The file
+name has the form `"ModelName_PackageName.md"`, for example,
 `"DecisionTreeClassifier_DecisionTree.md"`. Such a page can be referenced from any other
 markdown page in /docs/src/ like this: `[DecisionTreeClassifier](@ref
 DecisionTreeClassifier_DecisionTree)`.
@@ -56,6 +56,7 @@ const DESCRIPTORS_GIVEN_HANDLE =
 # determined the list of all descriptors, ranked by frequency:
 const descriptors = vcat(values(DESCRIPTORS_GIVEN_HANDLE)...)
 const ranking = MLJBase.countmap(descriptors)
+ranking["meta algorithms"] = 1e10
 const DESCRIPTORS = sort(unique(descriptors), by=d -> ranking[d], rev=true)
 const HANDLES = keys(DESCRIPTORS_GIVEN_HANDLE)
 
@@ -67,7 +68,7 @@ handle as key in /docs/src/ModelDescriptors.toml.
 
 """
 function models_missing_descriptors()
-    handles = handle.(models())
+    handles = handle.(models(wrappers=true))
     filter(handles) do h
         !(h in HANDLES)
     end
@@ -82,7 +83,7 @@ Return the list of  models with a given `descriptor`, such as "regressor", as
 these appear in /src/docs/ModelDescriptors.toml.
 
 """
-modelswith(descriptor) = filter(models()) do model
+modelswith(descriptor) = filter(models(wrappers=true)) do model
     descriptor in DESCRIPTORS_GIVEN_HANDLE[handle(model)]
 end
 

docs/src/list_of_supported_models.md

@@ -40,10 +40,11 @@ independent assessment.
 [Imbalance.jl](https://github.com/JuliaAI/Imbalance.jl) | - | RandomOversampler, RandomWalkOversampler, ROSE, SMOTE, BorderlineSMOTE1, SMOTEN, SMOTENC, RandomUndersampler, ClusterUndersampler,  ENNUndersampler, TomekUndersampler, | low | 
 [LIBSVM.jl](https://github.com/mpastell/LIBSVM.jl) | [MLJLIBSVMInterface.jl](https://github.com/JuliaAI/MLJLIBSVMInterface.jl) | LinearSVC, SVC, NuSVC, NuSVR, EpsilonSVR, OneClassSVM | high | also via ScikitLearn.jl
 [LightGBM.jl](https://github.com/IQVIA-ML/LightGBM.jl) | - | LGBMClassifier, LGBMRegressor | high | 
+[FeatureSelector.jl](https://github.com/JuliaAI/FeatureSelection.jl) | - | FeatureSelector, RecursiveFeatureElimination | low |
 [Flux.jl](https://github.com/FluxML/Flux.jl) | [MLJFlux.jl](https://github.com/FluxML/MLJFlux.jl) | NeuralNetworkRegressor, NeuralNetworkClassifier, MultitargetNeuralNetworkRegressor, ImageClassifier | low |
 [MLJBalancing.jl](https://github.com/JuliaAI/MLJBalancing.jl) | - | BalancedBaggingClassifier | low | 
 [MLJLinearModels.jl](https://github.com/JuliaAI/MLJLinearModels.jl) | - | LinearRegressor, RidgeRegressor, LassoRegressor, ElasticNetRegressor, QuantileRegressor, HuberRegressor, RobustRegressor, LADRegressor, LogisticClassifier, MultinomialClassifier | medium |
-[MLJModels.jl](https://github.com/JuliaAI/MLJModels.jl) (built-in) | - | ConstantClassifier, ConstantRegressor, ContinuousEncoder, DeterministicConstantClassifier, DeterministicConstantRegressor, FeatureSelector, FillImputer, InteractionTransformer, OneHotEncoder, Standardizer, UnivariateBoxCoxTransformer, UnivariateDiscretizer, UnivariateFillImputer,  UnivariateTimeTypeToContinuous, Standardizer, BinaryThreshholdPredictor | medium |
+[MLJModels.jl](https://github.com/JuliaAI/MLJModels.jl) (built-in) | - | ConstantClassifier, ConstantRegressor, ContinuousEncoder, DeterministicConstantClassifier, DeterministicConstantRegressor, FillImputer, InteractionTransformer, OneHotEncoder, Standardizer, UnivariateBoxCoxTransformer, UnivariateDiscretizer, UnivariateFillImputer,  UnivariateTimeTypeToContinuous, Standardizer, BinaryThreshholdPredictor | medium |
 [MLJText.jl](https://github.com/JuliaAI/MLJText.jl) | - | TfidfTransformer, BM25Transformer, CountTransformer | low |
 [MultivariateStats.jl](https://github.com/JuliaStats/MultivariateStats.jl) | [MLJMultivariateStatsInterface.jl](https://github.com/JuliaAI/MLJMultivariateStatsInterface.jl) | LinearRegressor, MultitargetLinearRegressor, RidgeRegressor, MultitargetRidgeRegressor, PCA, KernelPCA, ICA, LDA, BayesianLDA, SubspaceLDA, BayesianSubspaceLDA, FactorAnalysis, PPCA | high | 
 [NaiveBayes.jl](https://github.com/dfdx/NaiveBayes.jl) | [MLJNaiveBayesInterface.jl](https://github.com/JuliaAI/MLJNaiveBayesInterface.jl) | GaussianNBClassifier, MultinomialNBClassifier, HybridNBClassifier | low |

docs/src/transformers.md

@@ -30,7 +30,7 @@ MLJModels.OneHotEncoder
 MLJModels.ContinuousEncoder
 MLJModels.FillImputer
 MLJModels.UnivariateFillImputer
-MLJModels.FeatureSelector
+FeatureSelection.FeatureSelector
 MLJModels.UnivariateBoxCoxTransformer
 MLJModels.UnivariateDiscretizer
 MLJModels.UnivariateTimeTypeToContinuous

src/MLJ.jl

@@ -10,7 +10,7 @@ explicitly loaded from the model-providing package, using `@load`, for example.
 some common transformers, listed using `localmodels()` at startup, are immediately
 available, as are the following model wrappers: `Pipeline`, `TunedModel`, `EnsembleModel`,
 `IteratedModel`, `BalancedModel`, `TransformedTargetModel`, `BinaryThresholdPredictor`,
-and `Stack`.
+and `Stack`, `RecursiveFeatureSelection`.
 
 # Components
 
@@ -35,6 +35,9 @@ and `Stack`.
 - MLJBalancing.jl: Incorporation of oversampling/undersampling methods in pipelines, via
   the `BalancedModel` wrapper
 
+- FeatureSelection.jl: Transformers for feature selection, and the supervised model wrapper
+  `RecursiveFeatureSelection`.
+
 - MLJFlow.jl: Integration with MLflow workflow tracking
 
 - OpenML.jl: Tool for grabbing datasets from OpenML.org
@@ -60,6 +63,7 @@ import MLJBase.save
 using MLJEnsembles
 using MLJTuning
 using MLJModels
+@reexport using FeatureSelection
 using OpenML
 @reexport using MLJFlow
 @reexport using StatisticalMeasures
@@ -165,11 +169,11 @@ export Grid, RandomSearch, Explicit, TunedModel, LatinHypercube,
 # re-export from MLJModels:
 export models, localmodels, @load, @iload, load, info, doc,
     ConstantRegressor, ConstantClassifier,     # builtins/Constant.jl
-    FeatureSelector, UnivariateStandardizer,   # builtins/Transformers.jl
+    UnivariateStandardizer,
     Standardizer, UnivariateBoxCoxTransformer,
     OneHotEncoder, ContinuousEncoder, UnivariateDiscretizer,
     FillImputer, matching, BinaryThresholdPredictor,
-    UnivariateTimeTypeToContinuous, InteractionTransformer
+    UnivariateTimeTypeToContinuous, InteractionTransformer  # builtins/Transformers.jl
 
 # re-export from MLJIteration:
 export MLJIteration

test/integration.jl

@@ -53,6 +53,9 @@ FILTER_GIVEN_ISSUE = Dict(
             "MultiTaskLassoCVRegressor",
             "MultiTaskLassoRegressor",
         ]
+    "https://github.com/JuliaAI/FeatureSelection.jl/issues/15" =>
+        model -> model.package_name == "FeatureSelection" &&
+        model.name = "RecursiveFeatureElimination"
 )