For a 0.2 release

.github/workflows/ci.yml

@@ -17,7 +17,7 @@ jobs:
       fail-fast: false
       matrix:
         version:
-          - '1.6'
+          - '1.10'
           - '1' # automatically expands to the latest stable 1.x release of Julia.
         os:
           - ubuntu-latest
@@ -65,4 +65,4 @@ jobs:
             using Documenter: DocMeta, doctest
             using StatisticalMeasures
             DocMeta.setdocmeta!(StatisticalMeasures, :DocTestSetup, :(using StatisticalMeasures); recursive=true)
-            doctest(StatisticalMeasures)'
+            doctest(StatisticalMeasures)'

Project.toml

@@ -1,7 +1,7 @@
 name = "StatisticalMeasures"
 uuid = "a19d573c-0a75-4610-95b3-7071388c7541"
 authors = ["Anthony D. Blaom <[email protected]>"]
-version = "0.1.7"
+version = "0.2.0"
 
 [deps]
 CategoricalArrays = "324d7699-5711-5eae-9e2f-1d82baa6b597"
@@ -31,7 +31,7 @@ ScientificTypesExt = "ScientificTypes"
 CategoricalArrays = "0.10"
 CategoricalDistributions = "0.1"
 Distributions = "0.25"
-LearnAPI = "0.1"
+LearnAPI = "0.2"
 LossFunctions = "0.10, 0.11"
 MacroTools = "0.5"
 OrderedCollections = "1"
@@ -41,7 +41,7 @@ ScientificTypesBase = "3"
 StatisticalMeasuresBase = "0.1"
 Statistics = "1"
 StatsBase = "0.33, 0.34"
-julia = "1.6"
+julia = "1.10"
 
 [extras]
 LossFunctions = "30fc2ffe-d236-52d8-8643-a9d8f7c094a7"

README.md

@@ -6,7 +6,7 @@ Measures (metrics) for statistics and machine learning.
 [![Coverage](https://codecov.io/gh/JuliaAI/StatisticalMeasures.jl/branch/master/graph/badge.svg)](https://codecov.io/github/JuliaAI/StatisticalMeasures.jl?branch=master)
 [![Docs](https://img.shields.io/badge/docs-dev-blue.svg)](https://juliaai.github.io/StatisticalMeasures.jl/dev/)
 
-[List](https://juliaai.github.io/StatisticalMeasures.jl/dev/auto_generated_list_of_measures#aliases) of provided measures and their aliases thereof.
+[List](https://juliaai.github.io/StatisticalMeasures.jl/dev/auto_generated_list_of_measures#aliases) of provided measures and their aliases.
 
 Powered by
 [StatisticalMeasuresBase.jl](https://juliaai.github.io/StatisticalMeasuresBase.jl/dev/).

docs/make_tools.jl

@@ -94,7 +94,7 @@ string.
         page *= "\n## Classification measures (non-probabilistic)\n\n"
         constructors = filter(all_constructors) do c
             traits = traits_given_constructor[c]
-            traits.kind_of_proxy == LearnAPI.LiteralTarget() &&
+            traits.kind_of_proxy == LearnAPI.Point() &&
                 traits.observation_scitype <: Union{
                     Union{Missing,Finite},
                     AbstractArray{<:Union{Missing,Finite}},
@@ -104,7 +104,7 @@ string.
         page *= "\n## Regression measures (non-probabilistic)\n\n"
         constructors = filter(all_constructors) do c
             traits = traits_given_constructor[c]
-            traits.kind_of_proxy == LearnAPI.LiteralTarget() &&
+            traits.kind_of_proxy == LearnAPI.Point() &&
                 traits.observation_scitype <: Union{
                     Union{Missing,Infinite},
                     AbstractArray{<:Union{Missing,Infinite}},

ext/LossFunctionsExt.jl

@@ -35,7 +35,7 @@ loss(measure) = measure
 
 @trait(
     S{<:DistanceLoss},
-    kind_of_proxy=LearnAPI.LiteralTarget(),
+    kind_of_proxy=LearnAPI.Point(),
     orientation=Loss(),
     external_aggregation_mode = Mean(),
 )

ext/ScientificTypesExt.jl

@@ -26,7 +26,7 @@ function StatisticalMeasures.measures(yhat, y; trait_filters...)
     filter(measures(; trait_filters...)) do (_, metadata)
         requirement1 = y_scitype <: metadata.observation_scitype
         proxy = metadata.kind_of_proxy
-        requirement2 = if proxy == LearnAPI.LiteralTarget()
+        requirement2 = if proxy == LearnAPI.Point()
             yhat_scitype <: metadata.observation_scitype
         elseif proxy == LearnAPI.Distribution()
             yhat_scitype <: Density{<:y_scitype}

src/continuous.jl

@@ -1,6 +1,6 @@
 # This section is for measures that take `Infinite` input (vectors, arrays or tables)
 # and where the prediction has the same form as the ground truth, i.e.,`kind_of_proxy =
-# LearnAPI.LiteralTarget` (in particular, is not a probabilistic prediction).
+# LearnAPI.Point` (in particular, is not a probabilistic prediction).
 
 # If a new measure can be derived from a single scalar function with zero or one
 # parameters, then follow the examples that use the `@combination` macro. Otherwise you
@@ -12,7 +12,7 @@
 # defines LPLoss and MultitargetLPLoss constructors:
 @combination(
     LPLoss(; p=2) = multimeasure(Functions.pth_power_of_absolute_difference, mode=Mean()),
-    kind_of_proxy = LearnAPI.LiteralTarget(),
+    kind_of_proxy = LearnAPI.Point(),
     observation_scitype = Infinite,
     orientation = Loss(),
     human_name = "``L^p`` loss",
@@ -99,7 +99,7 @@ const multitarget_mean_absolute_value = multitarget_l1
 @combination(
     LPSumLoss(; p=2) =
         multimeasure(Functions.pth_power_of_absolute_difference, mode=Sum()),
-    kind_of_proxy = LearnAPI.LiteralTarget(),
+    kind_of_proxy = LearnAPI.Point(),
     observation_scitype = Infinite,
     orientation = Loss(),
     human_name = "``L^p`` sum loss",
@@ -158,7 +158,7 @@ const multitarget_l2_sum = MultitargetLPSumLoss(2)
 # define constructors:
 @combination(
     RootMeanSquaredError() = multimeasure(Functions.absolute_difference, mode=RootMean()),
-    kind_of_proxy = LearnAPI.LiteralTarget(),
+    kind_of_proxy = LearnAPI.Point(),
     observation_scitype = Infinite,
     orientation = Loss(),
 )
@@ -220,7 +220,7 @@ const multitarget_root_mean_squared_error = multitarget_rms
 @combination(
 RootMeanSquaredLogError() =
     multimeasure(Functions.absolute_difference_of_logs, mode=RootMean()),
-    kind_of_proxy = LearnAPI.LiteralTarget(),
+    kind_of_proxy = LearnAPI.Point(),
     observation_scitype = Infinite,
     orientation = Loss(),
 )
@@ -285,7 +285,7 @@ const multitarget_root_mean_squared_log_error = multitarget_rmsl
 @combination(
     RootMeanSquaredLogProportionalError(; offset=1) =
         multimeasure(Functions.absolute_difference_of_logs, mode=RootMean()),
-    kind_of_proxy = LearnAPI.LiteralTarget(),
+    kind_of_proxy = LearnAPI.Point(),
     observation_scitype = Infinite,
     orientation = Loss(),
 )
@@ -337,7 +337,7 @@ const multitarget_rmslp1 = MultitargetRootMeanSquaredLogProportionalError()
 @combination(
     RootMeanSquaredProportionalError(; tol=eps()) =
         multimeasure(Functions.normalized_absolute_difference, mode=RootMean()),
-    kind_of_proxy = LearnAPI.LiteralTarget(),
+    kind_of_proxy = LearnAPI.Point(),
     observation_scitype = Infinite,
     orientation = Loss(),
 )
@@ -390,7 +390,7 @@ const multitarget_rmsp = MultitargetRootMeanSquaredProportionalError()
 @combination(
     MeanAbsoluteProportionalError(; tol=eps()) =
         multimeasure(Functions.normalized_absolute_difference, mode=Mean()),
-    kind_of_proxy = LearnAPI.LiteralTarget(),
+    kind_of_proxy = LearnAPI.Point(),
     observation_scitype = Infinite,
     orientation = Loss(),
 )
@@ -440,7 +440,7 @@ const multitarget_mape = MultitargetMeanAbsoluteProportionalError()
 # define constructors:
 @combination(
     LogCoshLoss() = multimeasure(Functions.log_cosh_difference, mode=Mean()),
-    kind_of_proxy = LearnAPI.LiteralTarget(),
+    kind_of_proxy = LearnAPI.Point(),
     observation_scitype = Infinite,
     orientation = Loss(),
 )
@@ -512,7 +512,7 @@ const RSquaredType = API.FussyMeasure{<:API.RobustMeasure{<:_RSquared}}
 @trait(
     _RSquared,
     consumes_multiple_observations = true,
-    kind_of_proxy = LearnAPI.LiteralTarget(),
+    kind_of_proxy = LearnAPI.Point(),
     observation_scitype = Union{Missing,Infinite},
     orientation = Score(),
     human_name = "R² coefficient",

src/docstrings.jl

@@ -115,7 +115,7 @@ function docstring(signature; body="", footer="", scitype="")
         "use the syntax `measurements($m_str, ŷ, y)`. ")
     ret *= "Generally, an observation `obs` in `MLUtils.eachobs(y)` is expected to satisfy "*
         "`ScientificTypes.scitype(obs)<:`$scitype. "
-    # if kind_of_proxy == LearnAPI.LiteralTarget()
+    # if kind_of_proxy == LearnAPI.Point()
     #     ret *= "The same is true for `ŷ`. "
     # else
     #     ret *= "Each observation in `ŷ` should be this kind of proxy for actual "*

src/finite.jl

@@ -41,7 +41,7 @@ const ConfusionMatrixType = API.FussyMeasure{
 @trait(
     _ConfusionMatrix,
     consumes_multiple_observations=true,
-    kind_of_proxy=LearnAPI.LiteralTarget(),
+    kind_of_proxy=LearnAPI.Point(),
     observation_scitype=Union{Missing,Finite},
     orientation=Unoriented(),
     external_aggregation_mode=Sum(),
@@ -111,7 +111,7 @@ const confusion_matrix =  confmat
 # define both constructors:
 @combination(
     MisclassificationRate() = multimeasure(!=),
-    kind_of_proxy=LearnAPI.LiteralTarget(),
+    kind_of_proxy=LearnAPI.Point(),
     observation_scitype=Finite,
     orientation=Loss(),
 )
@@ -181,7 +181,7 @@ const multitarget_mcr = multitarget_misclassification_rate
 # define both constructors:
 @combination(
     Accuracy() = multimeasure(==),
-    kind_of_proxy=LearnAPI.LiteralTarget(),
+    kind_of_proxy=LearnAPI.Point(),
     observation_scitype=Finite,
     orientation=Score(),
 )
@@ -290,7 +290,7 @@ end
 @trait(
     _BalancedAccuracy,
     consumes_multiple_observations=true,
-    kind_of_proxy=LearnAPI.LiteralTarget(),
+    kind_of_proxy=LearnAPI.Point(),
     observation_scitype=Union{Missing,Finite},
     supports_weights=true,
     orientation=Score(),
@@ -357,7 +357,7 @@ const KappaType = API.FussyMeasure{<:API.RobustMeasure{<:_Kappa}}
 @trait(
     _Kappa,
     consumes_multiple_observations=true,
-    kind_of_proxy=LearnAPI.LiteralTarget(),
+    kind_of_proxy=LearnAPI.Point(),
     observation_scitype=Union{Missing,Finite},
     supports_weights=true,
     orientation=Score(),
@@ -415,7 +415,7 @@ const MatthewsCorrelationType = API.FussyMeasure{
 @trait(
     _MatthewsCorrelation,
     consumes_multiple_observations=true,
-    kind_of_proxy=LearnAPI.LiteralTarget(),
+    kind_of_proxy=LearnAPI.Point(),
     observation_scitype=Union{Missing,Finite},
     orientation=Score(),
     human_name = "Matthew's correlation",
@@ -518,7 +518,7 @@ const FScoreType = API.FussyMeasure{<:API.RobustMeasure{<:_FScore}}
 @trait(
     _FScore,
     consumes_multiple_observations=true,
-    kind_of_proxy=LearnAPI.LiteralTarget(),
+    kind_of_proxy=LearnAPI.Point(),
     observation_scitype=Union{Missing,OrderedFactor{2}},
     orientation=Score(),
     human_name = "``F_β`` score",
@@ -707,7 +707,7 @@ for Measure in TRUE_POSITIVE_AND_COUSINS
             @trait(
                 $_Measure,
                 consumes_multiple_observations=true,
-                kind_of_proxy=LearnAPI.LiteralTarget(),
+                kind_of_proxy=LearnAPI.Point(),
                 observation_scitype=Union{Missing,OrderedFactor{2}},
                 orientation=$orientation,
                 human_name=$name,
@@ -865,7 +865,7 @@ for Measure in [
             @trait(
                 $_Measure,
                 consumes_multiple_observations=true,
-                kind_of_proxy=LearnAPI.LiteralTarget(),
+                kind_of_proxy=LearnAPI.Point(),
                 observation_scitype=Union{Missing,Finite},
                 orientation=$orientation,
                 human_name=$name,
@@ -1104,7 +1104,7 @@ for Measure in [
             @trait(
                 $_Measure,
                 consumes_multiple_observations=true,
-                kind_of_proxy=LearnAPI.LiteralTarget(),
+                kind_of_proxy=LearnAPI.Point(),
                 observation_scitype=Union{Missing,Finite},
                 supports_class_weights=true,
                 orientation=$orientation,
@@ -1322,7 +1322,7 @@ MulticlassFScoreType = API.FussyMeasure{
 @trait(
     _MulticlassFScore,
     consumes_multiple_observations=true,
-    kind_of_proxy=LearnAPI.LiteralTarget(),
+    kind_of_proxy=LearnAPI.Point(),
     observation_scitype=Union{Missing,Finite},
     supports_class_weights=true,
     orientation=Score(),

test/finite.jl

@@ -208,7 +208,7 @@ end
     for m in (accuracy, recall, StatisticalMeasures.precision, f1score, specificity)
         @test API.consumes_multiple_observations(m)
         m == accuracy || @test !API.can_report_unaggregated(m)
-        @test API.kind_of_proxy(m) == StatisticalMeasures.LearnAPI.LiteralTarget()
+        @test API.kind_of_proxy(m) == StatisticalMeasures.LearnAPI.Point()
         @test API.observation_scitype(m) <: Union{Missing,Finite}
         @test !API.can_consume_tables(m)
         m == accuracy && @test API.supports_weights(m)
@@ -409,7 +409,7 @@ end
         m == MulticlassTrueNegativeRate &&
             (@test API.human_name(m) == "multi-class true negative rate")
         @test API.observation_scitype(m) <: Union{Missing,Finite}
-        @test API.kind_of_proxy(m) == StatisticalMeasures.LearnAPI.LiteralTarget()
+        @test API.kind_of_proxy(m) == StatisticalMeasures.LearnAPI.Point()
         @test API.orientation(m) == API.Score()
         @test API.can_report_unaggregated(m) == false
         @test !API.supports_weights(m)