v0.15.15

MLJModels v0.15.15

Diff since v0.15.14

Closed issues:

• Interaction transformer (#476)
• Unsupported atomic model in BinaryThresholdPredictor should trigger informative error (#486)

Merged pull requests:

• Catch unsupported model types in BinaryThresholdPredictor. (#487) (@ablaom)
• For a 0.15.15 release (#488) (@ablaom)

v0.15.14

MLJModels v0.15.14

Diff since v0.15.13

Closed issues:

• Add EvoLinear.jl to model registry (#481)

Merged pull requests:

• Update registry: BetaML (renamings) and TextAnalysis (new docstrings) (#484) (@ablaom)
• Update model registry (#485) (@ablaom)

v0.15.13

MLJModels v0.15.13

Diff since v0.15.12

• (new model) Add EvoLinearRegressor, from EvoLinear.jl, to the model registry.

Merged pull requests:

• CompatHelper: bump compat for RelocatableFolders to 1, (keep existing compat) (#477) (@github-actions[bot])
• Registry update to add new EvoLinear package (#482) (@ablaom)
• For a 0.15.13 release (#483) (@ablaom)

v0.15.12

MLJModels v0.15.12

Diff since v0.15.11

• Add InteractionTransformer for generating polynomial features (#478) @olivierlabayle
• Add InteractionTransformer and HierarchicalClusterer (from Clustering.jl) to model registry

Merged pull requests:

• Add HeirarchicalClustering and InteractionTransformer to model registry (#479) (@ablaom)
• For a 0.15.12 release (#480) (@ablaom)

v0.15.11

MLJModels v0.15.11

Diff since v0.15.10

• Add Clustering.jl model, DBSCAN to the model registry

Merged pull requests:

• MLJ compliant docstrings for constant regressor/classifier (#472) (@josephsdavid)
• make relocatable (#473) (@jbrea)
• Update registry to add DBSCAN (#474) (@ablaom)
• For a 0.15.11 release (#475) (@ablaom)

v0.15.10

MLJModels v0.15.10

Diff since v0.15.9

• Update the model registry. This will make many new document strings available via doc("SomeModel", pkg="SomePkg"). It also add the following BetaML.jl models: BetaMLGMMImputer, BetaMLGMMRegressor, BetaMLGenericImputer, BetaMLMeanImputer, BetaMLRFImputer.
• Fix a bug with OneHotEncoder (#468) @olivierlabayle

Closed issues:

• DecisionTree struct cleanup (#177)
• fit OneHotEncoder without MLJBase (#465)
• A column with missing as first value trips OneHotEncoder (#467)

Merged pull requests:

• Rm redundant code (#462) (@ablaom)
• Reorganization of model registry code (#463) (@ablaom)
• fix one hot encoder when first value in column is missing (#468) (@olivierlabayle)
• For a 0.15.10 release (#469) (@ablaom)
• add hack to allow ci pass (#470) (@OkonSamuel)
• Update registry (#471) (@ablaom)

v0.15.9

MLJModels v0.15.9

Diff since v0.15.8

• Allow OneHotCoder to provide basic missing value support by propagating missing values (#457) @Chandu-4444
• (bug fix) Allow models internally implementing a "data front-end", such as EvoTrees models, to behave correctly when wrapped by BinaryThresholdPredictor(model=...) (#459) @OkonSamuel
• Have matching(X, y) issue a warning if y is a table with just one column (#460) @DilumAluthge
• Bump Julia requirement to 1.6

Closed issues:

• BinaryThresholdPredictor not working for models implementing reformat/selectrows interface (#440)
• OneHotEncoder with missing values or Type (#450)
• models("Outlier") not finding the OutlierDetectionPython models (#453)

Merged pull requests:

• Add missing value propagation to OneHotEncoder (#457) (@Chandu-4444)
• fix bug in data front-end issue#440 (#459) (@OkonSamuel)
• matching(X, y) should print a warning if y is a table and y only has one column (#460) (@DilumAluthge)
• For a 0.15.9 release (#461) (@ablaom)

v0.15.8

MLJModels v0.15.8

Diff since v0.15.7

• (new model) Add OneRuleClassifier from OneRule.jl to the model registry. Do using MLJModels; doc("OneRuleClassifier") for details (#455) @roland-KA

Merged pull requests:

• Add OneRuleClassifier to the registry (#455) (@ablaom)
• For a 0.15.8 release (#456) (@ablaom)

v0.15.7

MLJModels v0.15.7

Diff since v0.15.6

Closed issues:

• no method matching metadata_model (#446)

Merged pull requests:

• Remove the section of code labelled "code to be removed in 0.15.0" (#449) (@DilumAluthge)
• Update registry to reflect new docstrings at MLJLIBSVMInterface.jl (#451) (@ablaom)
• For a 0.15.6 release (#452) (@ablaom)

v0.15.6

MLJModels v0.15.6

Diff since v0.15.5

Merged pull requests:

• Every version of MLJModels that has human_name will require at least MLJModelInterface 1.4 (#447) (@DilumAluthge)
• For a 0.15.6 release (#448) (@ablaom)