0.6.4 - Bugfixes

• Replaced usage of deprecated scipy_mode. Fixed #39
• Fixed ValueError: invalid literal for int() with base 10 in predict_qd. Fixed #40
• Added input validators to raise human-readable error messages when the input is not correct. Fixes #37
• Fixed AttributeError: module 'numpy' has no attribute 'object'.. Fixes #38

See documentation page for details.

0.6.3 - Bugfixes

• Fixed ValueError with recent versions of SciPy, due to usage of sparse arrays with object dtype. Fixes #31
• Fixed IndexError when NaN values are present in the dataframe. Fixes #28

See documentation page for details.

0.6.2 - Warning filter

• Now filtering UserWarning in fit_selector_model even in the sklearn adapter. Fixes #20

See documentation page for details.

0.6.1 - Minor changes

• Now filtering UserWarning in fit_selector_model. See #20
• New build system: now using virtualenv instead of conda in nox sessions. Fixes #23
• New project layout to avoid bug with xunitparser. Fixes #18

See documentation page for details.

0.6.0 - sklearn api renames

• Migrated CI/CD from Travis to Github Actions + nox.

• selector_skl module renamed sklearn and QDSSelector renamed QDScreen. Fixes #16

See documentation page for details.

0.5.0 - First public working release

Initial release with:

• A main method qd_screen to get the (adjacency matrix of) the quasi-deterministic-forest, a QDForest object with string representation of arcs (Fixes #8).
• Possibility to keep_stats so as to analyse the (conditional) entropies in order to define a "good" threshold.
• A method <QDForest>.fit_selector_model(X) to fit a QDSelectorModel feature selection model able to select relevant features and to predict missing ones. Fixes #7
• Support for both pandas dataframes and numpy arrays as input. Fixes #2
• A Scikit-learn compliant feature selector QDSSelector, providing the exact same functionality as above but compliant with scikit-learn Pipelines. Fixes #1

Non-functional:

• Travis continuous integration, generating documentation and deploying releases on PyPi
• A package level __version__ attribute. Fixes #3
• Added py.typed for PEP561 compliance. Fixed #4
• Initial setup.py and setup.cfg

See documentation page for details.