For a 0.16.12 release

.github/codecov.yml

@@ -0,0 +1,5 @@
+coverage:
+  status:
+    project:
+      default:
+        threshold: 0.5%

Project.toml

@@ -1,7 +1,7 @@
 name = "MLJModels"
 uuid = "d491faf4-2d78-11e9-2867-c94bc002c0b7"
 authors = ["Anthony D. Blaom <[email protected]>"]
-version = "0.16.11"
+version = "0.16.12"
 
 [deps]
 CategoricalArrays = "324d7699-5711-5eae-9e2f-1d82baa6b597"

src/registry/Models.toml

@@ -10,6 +10,8 @@ ParallelKMeans = ["KMeans"]
 NaiveBayes = ["GaussianNBClassifier", "MultinomialNBClassifier"]
 MultivariateStats = ["LDA", "MultitargetLinearRegressor", "BayesianSubspaceLDA", "FactorAnalysis", "LinearRegressor", "ICA", "PPCA", "RidgeRegressor", "KernelPCA", "MultitargetRidgeRegressor", "SubspaceLDA", "BayesianLDA", "PCA"]
 DecisionTree = ["AdaBoostStumpClassifier", "DecisionTreeRegressor", "DecisionTreeClassifier", "RandomForestRegressor", "RandomForestClassifier"]
+MLJBalancing = ["BalancedBaggingClassifier"]
+Imbalance = ["RandomOversampler", "SMOTENC", "TomekUndersampler", "ClusterUndersampler", "SMOTE", "SMOTEN", "ROSE", "RandomUndersampler", "ENNUndersampler", "BorderlineSMOTE1", "RandomWalkOversampler"]
 Clustering = ["HierarchicalClustering", "DBSCAN", "KMeans", "KMedoids"]
 EvoLinear = ["EvoSplineRegressor", "EvoLinearRegressor"]
 XGBoost = ["XGBoostCount", "XGBoostRegressor", "XGBoostClassifier"]

src/registry/Project.toml

@@ -3,8 +3,10 @@ BetaML = "024491cd-cc6b-443e-8034-08ea7eb7db2b"
 CatBoost = "e2e10f9a-a85d-4fa9-b6b2-639a32100a12"
 EvoLinear = "ab853011-1780-437f-b4b5-5de6f4777246"
 EvoTrees = "f6006082-12f8-11e9-0c9c-0d5d367ab1e5"
+Imbalance = "c709b415-507b-45b7-9a3d-1767c89fde68"
 InteractiveUtils = "b77e0a4c-d291-57a0-90e8-8db25a27a240"
 LightGBM = "7acf609c-83a4-11e9-1ffb-b912bcd3b04a"
+MLJBalancing = "45f359ea-796d-4f51-95a5-deb1a414c586"
 MLJBase = "a7f614a8-145f-11e9-1d2a-a57a1082229d"
 MLJClusteringInterface = "d354fa79-ed1c-40d4-88ef-b8c7bd1568af"
 MLJDecisionTreeInterface = "c6f25543-311c-4c74-83dc-3ea6d1015661"

src/registry/src/Registry.jl

@@ -1,4 +1,4 @@
-module Registry
+module Registry 
 
 using Pkg
 import Pkg.TOML

src/registry/src/update.jl

@@ -52,7 +52,7 @@ To register all the models in GreatNewPackage with MLJ:
 - Quit your REPL session and make a trivial commit to your MLJModels
   branch to force pre-compilation in a new julia session when you run
   `using MLJModels`. (For technical reasons the registry is not loaded
-  in `__init__`()`, so without pre-compiliation the new ]registry is not
+  in `__init__()`, so without pre-compiliation the new ]registry is not
   available.)
 
 - Test that the interfaces load properly with

test/builtins/ThresholdPredictors.jl

@@ -6,7 +6,6 @@ using CategoricalDistributions
 import Distributions
 import MLJBase
 
-
 const MMI = MLJModels.MLJModelInterface
 
 X_ = NamedTuple{(:x1,:x2,:x3)}((rand(4), rand(4), rand(4)))
@@ -163,9 +162,10 @@ MMI.input_scitype(::Type{<:DummyDetector}) = MMI.Table
     @test MMI.predict(detector, fr, X) == fill("in", length(y_))
 
     # integration (y == ["in", "out", "out", "out"]):
+    accuracy(yhat, y) = sum(yhat .== y)/length(y)
     e = MLJBase.evaluate(detector, X_, y_,
                          resampling=MLJBase.Holdout(fraction_train=0.5),
-                         measure=MLJBase.accuracy)
+                         measure=accuracy)
     @test e.measurement[1] ≈ 0
 end