Releases: JuliaAI/MLJModels.jl
v0.18.7
MLJModels v0.18.7
This release has been identified as a backport.
Automated changelogs for backports tend to be wildly incorrect.
Therefore, the list of issues and pull requests is hidden.
v0.18.6
MLJModels v0.18.6
(new model) Add the Transformer wrapper, from the new package MLJWrappers.jl, to the MLJ Model Registry.
This release has been identified as a backport.
Automated changelogs for backports tend to be wildly incorrect.
Therefore, the list of issues and pull requests is hidden.
v0.18.5
MLJModels v0.18.5
Merged pull requests:
- Fix threshold predictor bug (#607) (@ablaom)
- Update MLJ model registry (#609) (@ablaom)
- Bump actions/checkout from 2 to 6 (#610) (@dependabot[bot])
- Bump codecov/codecov-action from 4 to 6 (#611) (@dependabot[bot])
- Bump julia-actions/cache from 1 to 3 (#612) (@dependabot[bot])
- Bump julia-actions/setup-julia from 1 to 2 (#613) (@dependabot[bot])
- For a 0.18.5 release (#614) (@ablaom)
Closed issues:
- BinaryThresholdPredictor cannot wrap pipelines and others (#606)
v0.18.4
MLJModels v0.18.4
- Update the MLJ Model Registry to include latest model metadata, including accounting for migration of several models from MLJModels.jl to MLJTransforms.jl.
Merged pull requests:
v0.18.3
MLJModels v0.18.3
- (enhancement) The dummy regressor,
DeterministicConstantRegressor, now supports multi-targets (matrices or tables).
Merged pull requests:
- Extend support in
DeterministicConstantRegressorto multitargets (#602) (@ablaom) - For a 0.18.3 release (#603) (@ablaom)
Closed issues:
- Problem with interface packages with releases pre-dating implementation of the MLJ interface. (#363)
v0.18.2
v0.18.1
MLJModels v0.18.1
Merged pull requests:
Closed issues:
- Add short docstrings (#49)
- Find a way to load metadata from init (#74)
- Adding MulticlassPerceptron model (#156)
- Reported issues with the registry procedure (#169)
- Cleanup - Consistent code style (#203)
- Add DecisionTrees from ScikitLearn (#254)
- Uniformize fields for GLM models ? (#277)
- Review of model registry (#321)
- Add more categorical feature encoders (#375)
v0.18.0
MLJModels v0.18.0
-
Remove the model registry tools rendered redundant by the external package MLJModelRegistryTools (#590)
-
(breaking) Remove the following models now provided by MLJTransforms.jl instead:
ContinuousEncoder,FillImputer,InteractionTransformer,OneHotEncoder,Standardizer,UnivariateBoxCoxTransformer,UnivariateDiscrcetizer,UnivariateFillImputer,UnivariateStandardizer,UnivariateTimeTypeToContinuous. (#592) -
(new models) Update the Model Registry to include the models provided by MLJTransforms, which includes the following completely new models:
FrequencyEncoder,TargetEncoder,CardinalityReducer,OrdinalEncoder,MissingnessEncoder,ContrastEncoder(#592)
Merged pull requests: