glum 3.0.0a0

3.0.0a0 - 2023-08-17

Main change

• Ensure glum's compatibility with the new major release of tabmat (version 4.0.0) by @MartinStancsicsQC

Note that glum 3 is not compatible with pre-4.0.0 versions of tabmat.

Full Changelog: 2.5.2...v3.0.0a0

glum 2.5.2

2.5.2 - 2023-06-02

Bug fix

• Fixed a bug in glum.GeneralizedLinearRegressor when fit on a data set with a constant column and warm_start=True. See #645.

Other changes:

• We now pin llvm-openmp=11 when creating the wheel for macOS in line with scikit-learn. This should prevent segfaults on macOS with Intel architecture.
• Fixed a bug in the glm_benchmarks_analyze command line tool. See #642.
• Remove dev dependency on dask_ml.

glum 2.5.1

2.5.1 - 2023-05-19

Bug fix

• We fixed a bug in the computation of glum.distribution.NegativeBinomialDistribution.log_likelihood. Previously, this method just returned None.

glum 2.5.0

2.5.0 - 2023-04-28

New feature

• We added support for the Negative Binomial distribution. To use it, set the 'family' parameter of GeneralizedLinearRegressor or glum.GeneralizedLinearRegressorCV to 'negative.binomial'.

glum 2.4.1

What's Changed

• Fix typos by @jtilly in #606
• Fix issue with SplitMatrix in std_errors by @MarcAntoineSchmidtQC in #611
• Fix macOS build by @jtilly in #617
• Removes unused scikit-learn imports from _cd_fast.pyx by @adityagoel4512 in #615

New Contributors

• @adityagoel4512 made their first contribution in #615

Full Changelog: 2.4.0...2.4.1

glum 2.4.0

Other changes:

• glum._link.LogitLink has been made public.
• Apple Silicon wheels are now uploaded to PyPI.

glum 2.3.0

2.3.0 - 2023-01-06

Bug fix:

• A data frame with dense and sparse columns was transformed to a dense matrix instead of a split matrix by glum.GeneralizedLinearRegressor._set_up_and_check_fit_args. Fixed by calling tabmat.from_pandas on any data frame input.

New features:

• The following classes and functions have been made public:
  glum._distribution.BinomialDistribution, glum._distribution.ExponentialDispersionModel,
  glum._distribution.GammaDistribution, glum._distribution.GeneralizedHyperbolicSecant,
  glum._distribution.InverseGaussianDistribution, glum._distribution.NormalDistribution,
  glum._distribution.PoissonDistribution, glum._link.IdentityLink, glum._link.Link, glum._link.LogLink,
  glum._link.TweedieLink, glum._glm.get_family and glum._glm.get_link.
• The distribution and link classes now feature a more lenient equality check instead of the default identity check, so that, e.g., TweedieDistribution(1) == TweedieDistribution(1) returns True.

Glum 2.2.1

Fixing the 2.2.0 failed release.

Glum 2.2.0

2.2.0 - 2022-11-25

New features:

• Add an argument to GeneralizedLinearRegressorBase to drop the first category in a Categorical column using implementation in tabmat
• One may now request the Tweedie loss by setting the 'family' parameter of GeneralizedLinearRegressor and GeneralizedLinearRegressorCV to 'tweedie'.

Bug fixes:

• Setting bounds for constant columns was not working (bounds were internally modified to 0). A similar issue was preventing inequalities from working with constant columns. This is now fixed.

Other changes:

• No more builds for 32-bit systems with Python >= 3.8. This is due to scipy not supporting it anymore.

glum 2.1.2

2.1.2 - 2022-07-01

Other changes:

• Next attempt to build wheel for PyPI without --march=native.