Skip to content

0.8.7.dev0

Pre-release
Pre-release
Compare
Choose a tag to compare
@leliel12 leliel12 released this 08 Feb 19:11
· 7 commits to master since this release
6948771

Version 0.8.7.dev0

  • New Added functionality for user extension of scikit-criteria with
    decorators for creating aggregation and transformation models using
    functions.

    >>> from skcriteria.extend import mkagg, mktransformer
    >>>
    >>> @mkagg
    >>> def MyAgg(**kwargs):
    >>>     # Implementation of the aggregation function
    >>>
    >>> @mkagg(foo=1)
    >>> def MyAggWithHyperparam(**kwargs):
    >>>     # Implementation of the aggregation function with
    >>>     # hyperparameter 'foo'
    >>>
    >>> @mktransformer
    >>> def MyTransformer(**kwargs):
    >>>     # Implementation of the transformation function
    >>>
    >>> @mktransformer(bar=2)
    >>> def MyTransformerWithHyperparam(**kwargs):
    >>>     # Implementation of the transformation function with
    >>>     # hyperparameter 'bar'

    These decorators enable the creation of aggregation and transformation
    classes based on provided functions, allowing users to
    define decision-making models with less flexibility than traditional
    inheritance-based models.

    For more information check the tutorial Extending Aggregation and Transformation Functions

  • New Module: Introduced the skcriteria.testing module, exposing utility functions for for comparing objects created in Scikit-Criteria in a testing environment. These functions facilitate the comparison of instances of the DecisionMatrix, ResultABC, and RanksComparator classes.

    The assertion functions utilize pandas and numpy testing utilities for comparing matrices, series, and other attributes.

    Check the Reference for more information.

  • New The API of the agg, pipeline, preprocessing, and extend modules has
    been cleaned up to prevent autocompletion with imports from other modules.
    The imported modules are still present, but they are excluded when attempting
    to autocomplete. This functionality is achieved thanks to the context manager
    skcriteria.utils.cmanagers.hidden().

  • New All methods (agg and transformers) has a new get_method_name
    instance method.

  • Drop Drop support for Python 3.8