v1.5.2

What's Changed

In particular fixes an oversight where light GBM regressors in a sklearn pipeline didn't work.

• switch to Gurobi Sphinx theme by @mattmilten in #326
• Take newest setup from gurobipy-pandas by @pobonomo in #363
• Add a path from light GBM to sklearn pipelines by @pobonomo in #365

Full Changelog: v1.5.1...v1.5.2

v1.5.1

What's Changed

• Add tests for python 3.12 by @pobonomo in #299
• Fix some inconsistencies with column_transformer and get_convertor by @pobonomo in #294
• [GH316] Avoid logistic function for classification with Logistic Regression by @venaturum in #318
• [#315] Respect missing bias by @bzfwunde in #325

New Contributors

• @venaturum made their first contribution in #317
• @bzfwunde made their first contribution in #325

Full Changelog: v1.5.0...v1.5.1

v1.5.0

What's Changed

• Add a model for LightGBM by @pobonomo in #270
• Deal with empty list of column in column transformer by @pobonomo in #291
• Fixes for scikit-learn 1.4

Full Changelog: v1.4.0...v1.5.0

v1.4.1

What's Changed

• Deal with empty list of column in column transformer by @pobonomo in #291

Full Changelog: v1.4.0...v1.4.1

v1.4.0

What's Changed

The main update of this release is to use the new nonlinear capabilities in Gurobi 11 for modeling
the logistic regression. Namely, if the user has Gurobi 11, the attribute
FuncNonLinear is set for
the nonlinear constraints with the logistic function created by the package.

In our tests this results in better results (smaller errors in the solution and sometimes faster).

The corresponding parts of the documentation have been updated. In particular the
Student admission example
uses this and there is no more discussion on tuning the piecewise-linear approximation.

Other smaller updates regard the documentation that has been reorganized. In particular with the goal of documenting
better the internal objects that the package constructs and uses.

• Test with both gurobi 10.0.3 and 11.0.0 by @twbraam in #241
• Updates for Gurobi 11 by @pobonomo in #240
• Accept list of lists of variables as input to add predictor function by @pobonomo in #247
• Restructure and enhance documentation by @pobonomo in #256

Full Changelog: v1.3.3...v1.4.0

v1.3.3

• Fixed a bug with xgboost when verbose is asked
• More meaningful error if using a version of gurobipy that doesn't work

Full Changelog: v1.3.2...v1.3.3

v1.3.2

What's Changed

• Add support for XGBRegressor in SKlearn pipeline by @pobonomo in #200

Full Changelog: v1.3.1...v1.3.2

v1.3.1

What's Changed

Small changes for compatibility with newer version of tensorflow (2.13.0).

Update dependencies for testing and with the new tensorflow python 3.11 can now be tested.

• update to Tensorflow2.13.0 by @pobonomo in #185
• Bump xgboost from 1.7.5 to 1.7.6 by @dependabot in #188
• Bump ipykernel from 6.23.1 to 6.23.2 by @dependabot in #187
• Bump numpy from 1.24.3 to 1.25.0 by @dependabot in #189
• Tensorflow 2.13 by @pobonomo in #190
• Bump ipykernel from 6.23.2 to 6.23.3 by @dependabot in #191
• Bump pypa/gh-action-pypi-publish from 1.8.6 to 1.8.7 by @dependabot in #194
• Bump jupytext from 1.14.6 to 1.14.7 by @dependabot in #195
• Bump pandas from 2.0.2 to 2.0.3 by @dependabot in #196
• Bump scikit-learn from 1.2.2 to 1.3.0 by @dependabot in #198

Full Changelog: v1.3.0...v1.3.1

v1.3.0

What's Changed

• Add predictor constraint for XGBoost by @pobonomo in #178

Also updated dependencies

Full Changelog: v1.2.3...v1.3.0

v1.3.0rc1

What's Changed

Add support for XGBoost's gradient boosting tree based regressors

• Add predictor constraint for XGBoost by @pobonomo in #178

Full Changelog: v1.2.2...v1.3.0rc1