glum 3.2.3

3.2.3 - 2026-03-18

Bug fix:

• Fixed incorrect call in InverseGaussianDistribution.log_likelihood. The previous implementation always returned NaN.

glum 3.2.2

3.2.2 - 2026-03-17

Bug fix:

• Fixed incorrect formula in glum.CloglogLink.inverse_derivative2. This affected observed information matrix computation and robust/clustered standard errors for models using the complementary log-log link.

glum 3.2.1

3.2.1 - 2026-03-16

Bug fix:

• Fixed an error when predicting at a specific alpha with categorical features.

Other changes:

• Downgraded log messages in align_df_categories and add_missing_categories from INFO to DEBUG, and deduplicated them so they are emitted only once per column per fitted model.

glum 3.2.0

3.2.0 - 2026-03-10

New features:

• Add solver="closed-form" for Gaussian identity-link models, using an analytical normal-equations solution for ridge/OLS, auto-selecting it under solver="auto" for unconstrained no-L1 cases, and falling back to least-squares for singular or ill-conditioned systems.
• :class:~glum.GeneralizedLinearRegressor and :class:~glum.GeneralizedLinearRegressorCV now accept Polars <https://pola.rs>_ DataFrames as input, in addition to pandas DataFrames and numpy arrays.

Bug fixes:

• Fixed predict(X, alpha_index=...) and predict(X, alpha=...) on :class:~glum.GeneralizedLinearRegressorCV, which previously raised an error. The CV estimator now refits on the full data over the entire alpha path for the best l1_ratio_.
• Fixed alpha path computation in :class:~glum.GeneralizedLinearRegressorCV: the alpha grid is now computed from the properly standardized feature matrix and per-feature P1 penalties (matching the base class), and accounts for offset when present.

Full Changelog: 3.1.3...v3.2.0

glum 3.1.3

3.1.3 - 2025-02-18

Bug fixes:

• Fixed deviance_path_ in :class:~glum.GeneralizedLinearRegressorCV being scaled down by a factor of n_folds because test fold weights were not normalized to sum to 1.
• Fixed :class:~glum.NegativeBinomialDistribution theta setter rejecting np.number types, causing dist.theta = dist.theta to raise a TypeError.

Other changes:

• We disabled fast math to avoid invalid results (e.g., when dividing by zero).
• The default alpha_max for the regularization path in the pure ridge case (l1_ratio=0) is now computed from the null model gradient, matching glmnet's behavior. Previously it was hardcoded to 10.
• When min_alpha_ratio is not set and n_samples < n_features, the default is now 1e-2 instead of 1e-6, matching glmnet's convention.

Full Changelog: 3.1.2...3.1.3

glum 3.1.2

3.1.2 - 2025-01-30

Other:

• Linux-aarch64 wheels are now built and uploaded to PyPI.

Full Changelog: 3.1.1...3.1.2

glum 3.1.1

3.1.1 - 2025-01-13

Bug fixes:

• Fixed a bug where glum.TweedieDistribution._rowwise_gradient_hessian and glum.TweedieDistribution._eta_mu_deviance would call functions with wrong arguments in the p = 3 case.
• Fixed glum.InverseGaussianDistribution not using the optimized gradient, Hessian and deviance implementations, as well as those derivatives having the wrong sign.

Other changes:

• We now build and test with Python 3.13 in CI.

Full Changelog: 3.1.0...3.1.1

glum 3.1.0

3.1.0 - 2024-11-11

New features:

• New argument max_inner_iter for classes GeneralizedLinearRegressor and GeneralizedLinearRegressorCV to control the maximum number of iterations of the inner solver in the IRLS-CD algorithm.
• New fitted attributes col_means_ and col_stds_ for classes GeneralizedLinearRegressor and GeneralizedLinearRegressorCV.
• GeneralizedLinearRegressor now prints more informative logs when fitting with alpha_search=True and verbose=True.

Bug fixes:

• Fixed a bug where GeneralizedLinearRegressor.fit would raise a dtype mismatch error if fit with alpha_search=True.
• Use data type (float64 or float32) dependent precision in solvers.

glum 3.0.2

What's Changed

• Fix bug in Wald test with term names without intercept by Matthias Schmidtblaicher (@MatthiasSchmidtblaicherQC) in #797
• Bump mamba-org/setup-micromamba from 1.8.1 to 1.9.0 by Dependabot (@dependabot) in #798
• remove libblas mkl dependency by Marc-Antoine Schmidt (@MarcAntoineSchmidtQC) in #799
• Use ruff as our linter and formatter by Marc-Antoine Schmidt (@MarcAntoineSchmidtQC) in #800
• Pre-commit autoupdate by @quant-ranger in #803
• Combine parsing of initial values by Luca Bittarello (@lbittarello) in #805
• Support for build with numpy 2.0 by Marc-Antoine Schmidt (@MarcAntoineSchmidtQC) in #809
• Bump pypa/cibuildwheel from 2.18.1 to 2.19.1 by Dependabot (@dependabot) in #808
• Bump pypa/gh-action-pypi-publish from 1.8.14 to 1.9.0 by Dependabot (@dependabot) in #807
• Switch to pixi by Martin Stancsics (@stanmart) in #802
• Prepare release 3.0.2 by Marc-Antoine Schmidt (@MarcAntoineSchmidtQC) in #812

Full Changelog: 3.0.1...3.0.2

glum 3.0.1

Bug fix

• We now support scikit-learn 1.5.