Merge pull request #141 from JuliaAI/dev

For a 1.4.0 release

Author: ablaom

Project.toml

@@ -1,7 +1,7 @@
 name = "MLJModelInterface"
 uuid = "e80e1ace-859a-464e-9ed9-23947d8ae3ea"
 authors = ["Thibaut Lienart and Anthony Blaom"]
-version = "1.3.6"
+version = "1.4.0"
 
 [deps]
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
@@ -18,9 +18,10 @@ CategoricalArrays = "324d7699-5711-5eae-9e2f-1d82baa6b597"
 DataFrames = "a93c6f00-e57d-5684-b7b6-d8193f3e46c0"
 Distances = "b4f34e82-e78d-54a5-968a-f98e89d6e8f7"
 InteractiveUtils = "b77e0a4c-d291-57a0-90e8-8db25a27a240"
+Markdown = "d6f4376e-aef5-505a-96c1-9c027394607a"
 ScientificTypes = "321657f4-b219-11e9-178b-2701a2544e81"
 Tables = "bd369af6-aec1-5ad0-b16a-f7cc5008161c"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 
 [targets]
-test = ["CategoricalArrays", "DataFrames", "Distances", "InteractiveUtils", "ScientificTypes", "Tables", "Test"]
+test = ["CategoricalArrays", "DataFrames", "Distances", "InteractiveUtils", "Markdown", "ScientificTypes", "Tables", "Test"]

src/MLJModelInterface.jl

@@ -20,6 +20,7 @@ const MODEL_TRAITS = [
     :supports_online,
     :docstring,
     :name,
+    :human_name,
     :is_supervised,
     :prediction_type,
     :abstract_type,

src/metadata_utils.jl

@@ -1,21 +1,3 @@
-"""
-    docstring_ext
-
-Internal function to help generate the docstring for a package. See
-[`metadata_model`](@ref).
-"""
-function docstring_ext(T; descr::String="")
-    package_name = MLJModelInterface.package_name(T)
-    package_url  = MLJModelInterface.package_url(T)
-    model_name   = MLJModelInterface.name(T)
-    # the message to return
-    message = "$descr"
-    message *= "\n→ based on [$package_name]($package_url)."
-    message *= "\n→ do `@load $model_name pkg=\"$package_name\"` to " *
-        "use the model."
-    message *= "\n→ do `?$model_name` for documentation."
-end
-
 """
     metadata_pkg(T; args...)
 
@@ -72,19 +54,30 @@ function metadata_pkg(
     parentmodule(T).eval(ex)
 end
 
+# Extend `program` (an expression) to include trait definition for
+# specified `trait` and type `T`.
+function _extend!(program::Expr, trait::Symbol, value, T)
+    if value !== nothing
+        push!(program.args, quote
+              MLJModelInterface.$trait(::Type{<:$T}) = $value
+              end)
+        return nothing
+    end
+end
+
 """
     metadata_model(`T`; args...)
 
 Helper function to write the metadata for a model `T`.
 
 ## Keywords
 
-* `input_scitype=Unknown` : allowed scientific type of the input data
-* `target_scitype=Unknown`: allowed sc. type of the target (supervised)
-* `output_scitype=Unknown`: allowed sc. type of the transformed data (unsupervised)
-* `supports_weights=false` : whether the model supports sample weights
-* `docstring=""` : short description of the model
-* `load_path=""` : where the model is (usually `PackageName.ModelName`)
+* `input_scitype=Unknown`: allowed scientific type of the input data
+* `target_scitype=Unknown`: allowed scitype of the target (supervised)
+* `output_scitype=Unkonwn`: allowed scitype of the transformed data (unsupervised)
+* `supports_weights=false`: whether the model supports sample weights
+* `supports_class_weights=false`: whether the model supports class weights
+* `load_path="unknown"`: where the model is (usually `PackageName.ModelName`)
 
 ## Example
 
@@ -93,43 +86,192 @@ metadata_model(KNNRegressor,
     input_scitype=MLJModelInterface.Table(MLJModelInterface.Continuous),
     target_scitype=AbstractVector{MLJModelInterface.Continuous},
     supports_weights=true,
-    docstring="K-Nearest Neighbors classifier: ...",
     load_path="NearestNeighbors.KNNRegressor")
 ```
 """
 function metadata_model(
     T;
     # aliases:
-    input=Unknown,
-    target=Unknown,
-    output=Unknown,
-    weights::Bool=false,
-    descr::String="",
-    path::String="",
+    input=nothing,
+    target=nothing,
+    output=nothing,
+    weights::Union{Nothing,Bool}=nothing,
+    class_weights::Union{Nothing,Bool}=nothing,
+    descr::Union{Nothing,String}=nothing,
+    path::Union{Nothing,String}=nothing,
 
     # preferred names, corresponding to trait names:
     input_scitype=input,
     target_scitype=target,
     output_scitype=output,
-    supports_weights=weights,
-    docstring=descr,
-    load_path=path,
+    supports_weights::Union{Nothing,Bool}=weights,
+    supports_class_weights::Union{Nothing,Bool}=class_weights,
+    docstring::Union{Nothing,String}=descr,
+    load_path::Union{Nothing,String}=path,
+    human_name::Union{Nothing,String}=nothing
 )
-    if isempty(load_path)
-        pname = MLJModelInterface.package_name(T)
-        mname = MLJModelInterface.name(T)
-        load_path = "MLJModels.$(pname)_.$(mname)"
-    end
-    ex = quote
-        MLJModelInterface.input_scitype(::Type{<:$T}) = $input_scitype
-        MLJModelInterface.output_scitype(::Type{<:$T}) = $output_scitype
-        MLJModelInterface.target_scitype(::Type{<:$T}) = $target_scitype
-        MLJModelInterface.supports_weights(::Type{<:$T}) = $supports_weights
-        MLJModelInterface.load_path(::Type{<:$T}) = $load_path
-
-        function MLJModelInterface.docstring(::Type{<:$T})
-            return MLJModelInterface.docstring_ext($T; descr=$docstring)
+
+    program = quote end
+
+    # Note: Naively using metaprogramming to roll up the following
+    # code does not work. Only change this if you really know what
+    # you're doing.
+    _extend!(program, :input_scitype, input_scitype, T)
+    _extend!(program, :target_scitype, target_scitype, T)
+    _extend!(program, :output_scitype, output_scitype, T)
+    _extend!(program, :supports_weights, supports_weights, T)
+    _extend!(program, :supports_class_weights,supports_class_weights, T)
+    _extend!(program, :docstring, docstring, T)
+    _extend!(program, :load_path, load_path, T)
+    _extend!(program, :human_name, human_name, T)
+
+    parentmodule(T).eval(program)
+end
+
+# TODO: After `human_name` trait is added as model trait, include in
+# example given in the docstring for `doc_header`.
+
+"""
+    MLJModelInterface.doc_header(SomeModelType)
+
+Return a string suitable for interpolation in the document string of
+an MLJ model type. In the example given below, the header expands to
+something like this:
+
+>    `FooRegressor`
+>
+>Model type for foo regressor, based on [FooRegressorPkg.jl](http://existentialcomics.com/).
+>
+>From MLJ, the type can be imported using
+>
+>
+>    `FooRegressor = @load FooRegressor pkg=FooRegressorPkg`
+>
+>Construct an instance with default hyper-parameters using the syntax
+>`model = FooRegressor()`. Provide keyword arguments to override
+>hyper-parameter defaults, as in `FooRegressor(a=...)`.
+
+Ordinarily, `doc_header` is used in document strings defined *after*
+the model type definition, as `doc_header` assumes model traits (in
+particular, `package_name` and `package_url`) to be defined; see also
+[`MLJModelInterface.metadata_pkg`](@ref).
+
+
+### Example
+
+Suppose a model type and traits have been defined by:
+
+```
+mutable struct FooRegressor
+    a::Int
+    b::Float64
+end
+
+metadata_pkg(FooRegressor,
+    name="FooRegressorPkg",
+    uuid="10745b16-79ce-11e8-11f9-7d13ad32a3b2",
+    url="http://existentialcomics.com/",
+    )
+metadata_model(FooRegressor,
+    input=Table(Continuous),
+    target=AbstractVector{Continuous},
+    descr="La di da")
+```
+
+Then the docstring is defined post-facto with the following code:
+
+```
+const HEADER = MLJModelInterface.doc_header(FooRegressor)
+
+@doc \"\"\"
+\$HEADER
+
+### Training data
+
+In MLJ or MLJBase, bind an instance `model` ...
+
+<rest of doc string goes here>
+
+\"\"\" FooRegressor
+```
+
+"""
+function doc_header(SomeModelType)
+    name = MLJModelInterface.name(SomeModelType)
+    human_name = MLJModelInterface.human_name(SomeModelType)
+    package_name = MLJModelInterface.package_name(SomeModelType)
+    package_url = MLJModelInterface.package_url(SomeModelType)
+    params = MLJModelInterface.hyperparameters(SomeModelType)
+
+    ret =
+        """
+        ```
+        $name
+        ```
+
+        Model type for $human_name, based on
+        [$(package_name).jl]($package_url), and implementing the MLJ
+        model interface.
+
+        From MLJ, the type can be imported using
+
+        ```
+        $name = @load $name pkg=$package_name
+        ```
+
+        Do `model = $name()` to construct an instance with default hyper-parameters.
+        """ |> chomp
+
+    ret *= " "
+
+    isempty(params) && return ret
+
+    p = first(params)
+    ret *=
+        """
+        Provide keyword arguments to override hyper-parameter defaults, as in
+        `$name($p=...)`.
+        """ |> chomp
+
+    return ret
+end
+
+"""
+    synthesize_docstring
+
+Private method.
+
+Generates a value for the `docstring` trait for use with a model which
+does not have a standard document string, to use as the fallback. See
+[`metadata_model`](@ref).
+
+"""
+function synthesize_docstring(M)
+    package_name = MLJModelInterface.package_name(M)
+    package_url  = MLJModelInterface.package_url(M)
+    model_name   = MLJModelInterface.name(M)
+    human_name   = MLJModelInterface.human_name(M)
+    hyperparameters = MLJModelInterface.hyperparameters(M)
+
+    # generate text for the section on hyperparameters
+    text_for_params = ""
+    if !is_wrapper(M)
+        model = M()
+        isempty(hyperparameters) || (text_for_params *= "# Hyper-parameters")
+        for p in hyperparameters
+            value = getproperty(model, p)
+            text_for_params *= "\n\n- `$p = $value`"
         end
     end
-    parentmodule(T).eval(ex)
+
+    ret = doc_header(M)
+    if !isempty(text_for_params)
+        ret *=
+            """
+
+            $text_for_params
+
+            """
+    end
+    return ret
 end

src/model_traits.jl

@@ -13,11 +13,12 @@ const DeterministicDetector = Union{
 
 const StatTraits = StatisticalTraits
 
-StatTraits.docstring(M::Type{<:MLJType}) = name(M)
-
 function StatTraits.docstring(M::Type{<:Model})
-    return "$(name(M)) from $(package_name(M)).jl.\n" *
-        "[Documentation]($(package_url(M)))."
+    docstring = Base.Docs.doc(M) |> string
+    if occursin("No documentation found", docstring)
+        docstring = synthesize_docstring(M)
+    end
+    return docstring
 end
 
 StatTraits.is_supervised(::Type{<:Supervised}) = true