https://github.com/drsoliddevil/mlr-gd - 0.3.0

mlr-gd 0.3.0 (2025-02-15)

0.3.0 is primarily a documentation update, this includes fully fledged documentation hosted on readthedocs No changes to the actual functionality of mlr-gd have been made since the last release (0.2.1).

Changes

• Created detailed documentation for mlr-gd with hosting on readthedocs.io
• Created github action for running all tests upon a push or pull-request to the main branch.
• Created security policy, containing instructions for reporting vulnerabilities and a list of supported versions.
• Created contribution guidelines for how you can contribute to mlr-gd.
• Created a code of conduct.
• Created a changelog, the changlog includes previous (functional) versions (0.2.1 and 0.1.2)
• Created templates for submitting issues, more specifically for bug reports and enhancement suggestions.
• Created template for writing release notes.
• Created checklist for releasing.
• Created template for submitting pull-requests.

Additional information

• Link to documentation: https://mlr-gd.readthedocs.io/en/latest/

Full Changelog: https://github.com/DrSolidDevil/mlr-gd/compare/v0.2.1...v0.3.0

- Python
Published by DrSolidDevil about 1 year ago

https://github.com/drsoliddevil/mlr-gd - 0.2.1

mlr-gd 0.2.1 (2025-02-05)

New Features

• Ability to use cost function other than mean squared error (including custom ones)

• New cost function. This version only comes with mean squared error and mean absolute error as pre-defined cost functions

• Compatability with pandas (DataFrame & Series). mlr-gd still does not require pandas to use, although it is required for some unit tests.

• Added testing with pytest. Added unit, integration and metadata tests.

Changes

• Cost function have been moved from melar/main.py to melar/cfuncs.py

• melar.LinearRegression now takes costfunction and costfunction_deriv as keyword arguments. This allows use of cost functions other than mean squared error, either other pre-defined cost functions (in cfuncs) or completly custom cost functions.

• Added docstrings to melar/__init__.py, melar/main.py and melar/cfuncs.py

• Added module level dunders to melar/__init__.py

• Added __slots__ and __repr__ to melar.LinearRegression For improved performance and ease of use/debugging.

Bug Fixes

• Fixed issue where the model could not handle 1 weight. The reason the model could not handle 1 weight was that when having weights_amount=1 in LinearRegression.__init__ resulted in melar.LinearRegression.weights being set to np.random.uniform(low=-1, high=1, size=1) which creates a np.ndarray with 1 element. You can not perform a dot product operation with an array with shape (1, ) and an array with the shape (n, ).

Additional information

• 0.2.0 was accidentally released to PyPi before merging with main branch, it had the dev note in README.md. Thus it was yanked and is considered a broken version. 0.2.1 is 0.2.0 that is merged and has dev note removed.

\ Full Changelog: https://github.com/DrSolidDevil/mlr-gd/compare/v0.1.2...v0.2.1

- Python
Published by DrSolidDevil about 1 year ago

https://github.com/drsoliddevil/mlr-gd - 0.1.2

mlr-gd 0.1.2 (2025-01-12)

Previous versions have been broken due to an issue with __init__.py this is the first functioning version of mlr-gd.

Changes

• Adjusted space above and below logo in README.md (#1)
• Added example code to README.md (#3)
• Fixed spelling

Bug Fixes

• Fixed issue where __init__.py could not import LinearRegression. (#2)

Additional information

• Previous versions (v0.1.0 & v0.1.1) have now been yanked from PyPi

\ Full Changelog: https://github.com/DrSolidDevil/mlr-gd/compare/v0.1.1...v0.1.2

- Python
Published by DrSolidDevil about 1 year ago

https://github.com/drsoliddevil/mlr-gd - 0.1.1 (Broken)

THIS VERSION IS BROKEN AND NOT FUNCTIONAL https://pypi.org/project/mlr-gd/

This is the first release (on github).

Changes:

• Same as 0.1.0 except a the README.md has installation instructions

- Python
Published by DrSolidDevil about 1 year ago