v0.2.0

CategoricalDistributions v0.2.0

Diff since v0.1.15

Significant changes for this release are discussed in #75. They are:

• Update the compatibility requirements to CategoricalArrays = "1"
• (breaking) Arrange that levels(::UnivariateFinite) returns a categorical vector, to mimic new behaviour of CategoricalArrays.jl. Previously the unwrapped version of this vector was returned.
• Deprecate the classes method. Instead, promote use of CategoricalArrarys.levels(::UnivariateFinite) or CategoricalArrays(levels(first(A)), if A is an array with UnivariteFinite element type.

Merged pull requests:

• Require CategoricalArrays 1.0 and address breaking changes (#75) (@ablaom)
• Create dependabot.yml and update action versions (#76) (@abhro)
• Update code blocks in docstrings (#77) (@abhro)
• add a docstring for Distribution.params(::UnivariateFinite) (#78) (@ablaom)
• Bump actions/checkout from 4 to 5 (#79) (@dependabot[bot])
• For a 0.2.0 release (#80) (@ablaom)

v0.1.15

CategoricalDistributions v0.1.15

Diff since v0.1.14

Merged pull requests:

• Fix bug in error for UnivariateFinite constructor (#73) (@ablaom)
• For a 0.1.15 release (#74) (@ablaom)

v0.1.14

CategoricalDistributions v0.1.14

Diff since v0.1.13

Merged pull requests:

• allow dispatch on mixed number types for + and - (#71) (@tiemvanderdeure)
• For a 0.1.14 release (#72) (@ablaom)

Closed issues:

• Dictionaries-based implementation appears 45x faster for rand (#64)

v0.1.13

CategoricalDistributions v0.1.13

Diff since v0.1.12

• (optimization) Speed up single sampling of an UnivariateDistribution with rand (#68).

Merged pull requests:

• Introduce optimizations to speed up rand(d) (#68) (@ablaom)
• For a 0.1.13 release (#69) (@ablaom)

v0.1.12

CategoricalDistributions v0.1.12

Diff since v0.1.11

Merged pull requests:

• Use @belapsed for performance measurements (#63) (@skleinbo)
• Fix implementation of rand (#66) (@ablaom)
• For a 0.1.12 release (#67) (@ablaom)

Closed issues:

• Remove ~1.9.00 from CI when julia 1.9 is released (#52)
• Relax a test to address poorer than expected performance on Windows? (#60)
• Fix incomplete implementation of rand(::UnivariateFinite, ...) (#65)

v0.1.11

CategoricalDistributions v0.1.11

Diff since v0.1.10

• (new feature) Implement modes for UnivariateFinite objects (#58) @skleinbo

Merged pull requests:

• Implement modes (#59) (@skleinbo)
• For a 0.1.11 release (#61) (@ablaom)

Closed issues:

• Implement modes for UnivariateFinite objects (#58)

v0.1.10

CategoricalDistributions v0.1.10

Diff since v0.1.9

Closed issues:

• fix ordering of classes in UnivariateFinite objects (#34)
• Bump julia to 1.6 (#38)
• make pdf perform common promotions (#39)
• Arithmetic on UnivariateFiniteArray (#42)
• Problem with indexing of UnivariateFiniteArray (#43)
• Unexpected broadcasting behaviour for pdf (#44)
• Unexpected behavior of pdf function (#47)
• Make UnicodePlots an optional dependency (#49)
• Investigate method ambiguities (#50)

Merged pull requests:

• Improve implementation of pdf (#40) (@bkamins)
• fix indexing bug #43 (#45) (@OkonSamuel)
• Optimize broadcasting on the pdf function (#48) (@OkonSamuel)
• Make UnicodePlots a weak dependency (for julia > 1.9) (#51) (@ablaom)
• For a 0.1.11 release (#53) (@ablaom)
• Specialize * and / to Number (#54) (@ablaom)
• For a 0.1.10 release (#55) (@ablaom)

v0.1.9

CategoricalDistributions v0.1.9

Diff since v0.1.8

• Fix a bug with order of levels in constructor (#34) @OkonSamuel

Merged pull requests:

• fix issue #34 (#35) (@OkonSamuel)
• For a 0.1.9 release (#36) (@ablaom)

v0.1.8

CategoricalDistributions v0.1.8

Diff since v0.1.7

Merged pull requests:

• Allow creation of UnivariateFinite object from any abstract array object. (#31) (@OkonSamuel)
• For a 0.1.8 release (#32) (@ablaom)

v0.1.7

CategoricalDistributions v0.1.7

Diff since v0.1.6

Closed issues:

• Easiest way to convert UnivariateFinite{MLJ.Multiclass{2}} to a single number (#26)

Merged pull requests:

• Update compat version of UnicodePlots (#29) (@zsz00)
• For 0.1.7 release (#30) (@ablaom)