v1.1.3

MLJModelInterface v1.1.3

Diff since v1.1.2

Merged pull requests:

• Add list of model subtypes (#107) (@ablaom)
• For a 1.1.3 release (#108) (@ablaom)

v1.1.2

MLJModelInterface v1.1.2

Diff since v1.1.1

Closed issues:

• Implementing interfaces: towards a standardised approach (#20)
• Handling of accelerator / device in model interface (#93)

Merged pull requests:

• CompatHelper: bump compat for "ScientificTypesBase" to "2" (#105) (@github-actions[bot])
• For a 1.1.2 release (#106) (@ablaom)

v1.1.1

MLJModelInterface v1.1.1

Diff since v1.1.0

Merged pull requests:

• Address renaming of ScientificTypes (#101) (@ablaom)
• for a 1.1.1 release (#103) (@ablaom)

v1.1.0

MLJModelInterface v1.1.0

Diff since v1.0.0

• Make is_equal_except (and hence ==) for MLJType objects property-based instead of field-based (#97)
• Generalize definition of is_equal_except (and hence ==) to treat object properties declared as "deep" differently; see the doc-string for new trait deep_properties for details (#97)

Merged pull requests:

• Add CompatHelper (#95) (@ablaom)
• Fix invalid test that use deprecated categorical arrays methods (#96) (@ablaom)
• Make MLJType == property-based, instead of field-based (#97) (@ablaom)

v1.0.0

MLJModelInterface v1.0.0

Diff since v0.4.1

• Extend compatibility requirement for StatisticalTraits to include v1.0.

Closed issues:

• Roadmap to 1.0? (#69)

Merged pull requests:

• Extend compat for StatisticalTraits; bump to version to 1.0.0 (#94) (@ablaom)

v0.4.1

MLJModelInterface v0.4.1

Diff since v0.4.0

• Add support for iterative models (#91, JuliaAI/MLJ.jl#139)

Merged pull requests:

• Modify the test for distribution-fitters (#90) (@ablaom)
• Add support for iterative models (#91) (@ablaom)
• For a 0.4.1 release (#92) (@ablaom)

v0.4.0

MLJModelInterface v0.4.0

Diff since v0.3.9

This release is breaking for MLJModels (and hence MLJ) but should not be breaking for 3rd party packages implementing the MLJ model API for their models.

• Make selectrows(X, rows) return X instead of throwing an error when X is not a table, matrix or vector (#84)
• Import all model traits from new package StatisticalTraits.jl, which includes new trait supports_class_weights.

Closed issues:

• selectrows() doesn't work with AbstractDict (#77)

Merged pull requests:

• implement non-erroring fallback for selectrows to address #77 (#84) (@ablaom)
• Bump version = "0.4.0" (#87) (@ablaom)
• For a 0.4.0 release (#88) (@ablaom)

v0.3.9

MLJModelInterface v0.3.9

Diff since v0.3.8

Merged pull requests:

• Get model traits from new package StatisticalTraits.jl (#85) (@ablaom)
• For a 0.3.9 release (#86) (@ablaom)

v0.3.8

MLJModelInterface v0.3.8

Diff since v0.3.7

Closed issues:

• Add keyword aliases for key-word args in metadata_model and meta_data package (#73)

Merged pull requests:

• suggested fix for typename issue in 1.6+ (#80) (@tlienart)
• Fix name(::Type) method (#81) (@ablaom)
• For a 0.3.8 release (#82) (@ablaom)
• Add correct trait names to utilities for declaring trait values (#83) (@ablaom)

v0.3.7

MLJModelInterface v0.3.7

Diff since v0.3.6

Merged pull requests:

• Move from Travis CI to GitHub Actions CI (#74) (@DilumAluthge)
• Add facility for adding a data front-end to a model implementation (#76) (@ablaom)
• For a 0.3.7 release (#78) (@ablaom)