v0.4.4

MLJEnsembles v0.4.4

Diff since v0.4.3

• Update compatibility requirements: CategoricalArrays="1", CategoricalDistributions="0.2"

Merged pull requests:

• Bump compats: CategoricalArrays="1.0", CategoricalDistributions="0.2" (#42) (@ablaom)
• For a 0.4.4 release (#45) (@ablaom)

v0.4.3

MLJEnsembles v0.4.3

Diff since v0.4.2

Merged pull requests:

• For a 0.4.3 release (#41) (@ablaom)

v0.4.2

MLJEnsembles v0.4.2

Diff since v0.4.1

Merged pull requests:

• only add method to Random._GLOBAL_RNG if it is defined (#38) (@KristofferC)
• Add tests for CPUThreads and CPUProceses (#39) (@ablaom)
• For a 0.4.2 release (#40) (@ablaom)

v0.4.1

MLJEnsembles v0.4.1

Diff since v0.4.0

Merged pull requests:

• Add extra serialization test (#32) (@ablaom)
• Notation tweak (#33) (@ablaom)
• fix data race caused by rngs (#34) (@OkonSamuel)
• Tweak CI - take II (#36) (@ablaom)
• For a 0.4.1 release (#37) (@ablaom)

Closed issues:

• Multithreading option not working (#27)

v0.4.0

MLJEnsembles v0.4.0

Diff since v0.3.3

• (breaking) Remove MLJBase as dependency, replacing it with StatisticalMeasuresBase.jl. Adapt code appropriately to the new measures API. To use MLJEnsembles.jl without MLJBase.jl, one may need to explicitly import StatisticalMeasures.jl. See also the MLJBase 1.0 migration guide for further details (#30).

Merged pull requests:

• Adapt to migration of measures MLJBase -> StatisticalMeasures (#30) (@ablaom)
• For a 0.4.0 release (#31) (@ablaom)

v0.3.3

MLJEnsembles v0.3.3

Diff since v0.3.2

Closed issues:

• Implement measures checks (#20)

Merged pull requests:

• CompatHelper: bump compat for StatsBase to 0.34, (keep existing compat) (#28) (@github-actions[bot])

v0.3.2

MLJEnsembles v0.3.2

Diff since v0.3.1

Merged pull requests:

• Relax an unecessarily strict test (#22) (@ablaom)
• For a 0.3.2 release (#23) (@ablaom)

v0.3.1

MLJEnsembles v0.3.1

Diff since v0.3.0

Merged pull requests:

• Allow for atomic models that have Missing in target_scitype (#19) (@ablaom)
• For a 0.3.1 release (#21) (@ablaom)

v0.3.0

MLJEnsembles v0.3.0

Diff since v0.2.0

• (enhancement, breaking) Adapt to MLJBase 0.20, and in particular the new serialization API, which allows serialisation of EnsembleModel(model=...) where model is not necessarily a pure-Julia model (#15) @olivierlabayle
• (bug fix) Enable safe wrapping of atomic models that implement a data front-end (#17) @OkonSamuel

Closed issues:

• Ensembling fails for atomic models implementing the reformat/selectrows interface for caching model specific data representations (#16)

Merged pull requests:

• add serialization of ensembles (#15) (@olivierlabayle)
• fix data front-end bug #16 (#17) (@OkonSamuel)
• For a 0.3.0 release (#18) (@ablaom)

v0.2.0

MLJEnsembles v0.2.0

Diff since v0.1.2

• (breaking) Change syntax EnsembleModel(atom=...) to EnsembleModel(model=...) for consistency with other MLJ model wrappers (eg, TunedModel) but additionally allow passing model as non-keyword argument, as in EnsembleModel(my_tree, ...).
• Add CategoricalDistributions.jl as a dependency
• Bump compat for MLJBase to 0.19 and address the fact that UnivariateFinite functionality is not defined in MLJBase 0.19, but in CategoricalDistributions.jl instead.

Merged pull requests:

• Use CategoricalDistributions for UnivariateFinite, instead of MLJBase (#13) (@ablaom)
• For a 0.2.0 release (#14) (@ablaom)