Releases: jrudar/TreeOrdination
Releases · jrudar/TreeOrdination
TreeOrdination v1.3.4
- Minor bug fix
TreeOrdination Version 1.3.3
- TreeOrdination can now take advantage of LANDMark's proximity measures (both using terminal nodes as features or all nodes in the decision path as features)
- Data is cast into the np.float32 dtype for the CLRClosureTransformer to reduce memory usage
- Updated LANDMark dependency to version 2.1.0
- Updated version to 1.3.3
TreeOrdination Version 1.3.2
- Bug fix: (1,n) dimensional samples were not being transformed correctly.
TreeOrdination Version 1.3.1
- Minor bug fixes and changes to README, API documentation
- Fixed tests
TreeOrdination Version 1.3.0
Updated hyper-parameters to expose re-sampling within LANDMark trees.
TreeOrdination Version 1.2.1
- Updated LANDMark dependency
TreeOrdination Version 1.2.0
- Removed unnecessary dependencies
- Switched to using 'shap' package
- Using hatchling for installation
TreeOrdination v1.1.1
Minor fix to pyproject.toml
TreeOrdination v1.1.0
What's Changed
- Minor pyproject.toml fixes by @peterk87 in #3
- Removed DEICODE dependency (cannot be used to transform new data)
- Removed pandas dependency - Not Used
- Improved readability and modularity of code
- Moved some functionality into other modules (included transformers, classes which calculate importance scores)
- Bug fixes
- Using alibi for feature importance
- Per-class and per-sample feature importance scores can now be plotted
- scikit-learn preprocessing steps can now be used to transform data through the 'transformer' parameter
- Added workflows (testing for Python 3.8 - 3.11), notebooks (for usage examples)
- Using pyproject.toml for install
- Updated project README to reflect new documentation
- Added CONTRIBUTION and API documentation
- Using imbalanced-learn for more flexible resampling through the 'resampler' parameter
- Added functionality to explore feature importance
- Fixed coding style using 'black'
- Bumped version to 1.1.0
- PyPI Release
New Contributors
TreeOrdination-v1.0.3
What's Changed
- Implemented down-sampling of an over-represented class. This may be extended in the future.
- Implemented a proper way to split the dataset into features which are to be scaled and those which are not (eg: binary features)
- Bug fixes