Changelog

Version 0.8.7

  • 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

Version 0.8.6

  • New Rank reversal 1 implementhed in the RankInvariantChecker class

    >>> import skcriteria as skc
>>> from skcriteria.cmp import RankInvariantChecker
>>> from skcriteria.agg.similarity import TOPSIS

>>> dm = skc.datasets.load_van2021evaluation()
>>> rrt1 = RankInvariantChecker(TOPSIS())
>>> rrt1.evaluate(dm)
<RanksComparator [ranks=['Original', 'M.ETH', 'M.LTC', 'M.XLM', 'M.BNB', 'M.ADA', 'M.LINK', 'M.XRP', 'M.DOGE']]>

  • New The module skcriteria.madm was deprecated in favor of skcriteria.agg

  • Add support for Python 3.11.

  • Removed Python 3.7. Google collab now work with 3.8.

  • Updated Scikit-Learn to 1.3.x.

  • Now all cached methods and properties are stored inside the instance. Previously this was stored inside the class generating a memoryleak.

Version 0.8.3

  • Fixed a bug detected on the EntropyWeighted, Now works as the literature specifies

Version 0.8.2

  • We bring back Python 3.7 because is the version used in google.colab.

  • Bugfixes in plot.frontier and dominance.eq.

Version 0.8

  • New The skcriteria.cmp package utilities to compare rankings.

  • New The new package skcriteria.datasets include two datasets (one a toy and one real) to quickly start your experiments.

  • New DecisionMatrix now can be sliced with a syntax similar of the pandas.DataFrame.

    • dm["c0"] cut the c0 criteria.

    • dm[["c0", "c2"] cut the criteria c0 and c2.

    • dm.loc["a0"] cut the alternative a0.

    • dm.loc[["a0", "a1"]] cut the alternatives a0 and a1.

    • dm.iloc[0:3] cuts from the first to the third alternative.

  • New imputation methods for replacing missing data with substituted values. These methods are in the module skcriteria.preprocessing.impute.

  • New results object now has a to_series method.

  • Changed Behaviour: The ranks and kernels equals are now called values_equals. The new aequals support tolerances to compare numpy arrays internally stored in extra_, and the equals method is equivalent to aequals(rtol=0, atol=0).

  • We detected a bad behavior in ELECTRE2, so we decided to launch a FutureWarning when the class is instantiated. In the version after 0.8, a new implementation of ELECTRE2 will be provided.

  • Multiple __repr__ was improved to folow the Python recomendation

  • Critic weighter was renamed to CRITIC (all capitals) to be consistent with the literature. The old class is still there but is deprecated.

  • All the functions and classes of skcriteria.preprocessing.distance was moved to skcriteria.preprocessing.scalers.

  • The StdWeighter now uses the sample standar-deviation. From the numerical point of view, this does not generate any change, since the deviations are scaled by the sum. Computationally speaking there may be some difference from the ~5th decimal digit onwards.

  • Two method of the Objective enum was deprecated and replaced:

    • Objective.construct_from_alias() -> Objective.from_alias() (classmethod)

    • Objective.to_string() -> Objective.to_symbol()

    The deprecated methods will be removed in version 1.0.

  • Add a dominance plot DecisionMatrix.plot.dominance().

  • WeightedSumModel raises a ValueError when some value < 0.

  • Moved internal modules

    • skcriteria.core.methods.SKCTransformerABC -> skcriteria.preprocessing.SKCTransformerABC

    • skcriteria.core.methods.SKCMatrixAndWeightTransformerABC -> skcriteria.preprocessing.SKCMatrixAndWeightTransformerABC

Version 0.7

  • New method: ELECTRE2.

  • New preprocessing strategy: A new way to transform from minimization to maximization criteria: NegateMinimize() which reverses the sign of the values of the criteria to be minimized (useful for not breaking distance relations in methods like TOPSIS). Additionally the previous we rename the MinimizeToMaximize() transformer to InvertMinimize().

  • Now the RankingResult, support repeated/tied rankings and some methods were implemented to deal with these cases.

    • RankingResult.has_ties_ to see if there are tied values.

    • RankingResult.ties_ to see how often values are repeated.

    • RankingResult.untied_rank_ to get a ranking with no repeated values.

      repeated values.

  • KernelResult now implements several new properties:

    • kernel_alternatives_ to know which alternatives are in the kernel.

    • kernel_size_ to know the number of alternatives in the kernel.

    • kernel_where_ was replaced by kernelwhere_ to standardize the api.

Version 0.6

  • Support for Python 3.10.

  • All the objects of the project are now immutable by design, and can only be mutated troughs the object.copy() method.

  • Dominance analysis tools (DecisionMatrix.dominance).

  • The method DecisionMatrix.describe() was deprecated and will be removed in version 1.0.

  • New statistics functionalities DecisionMatrix.stats accessor.

  • The accessors are now cached in the DecisionMatrix.

  • Tutorial for dominance and satisfaction analysis.

  • TOPSIS now support hyper-parameters to select different metrics.

  • Generalize the idea of accessors in scikit-criteria througth a common framework (skcriteria.utils.accabc module).

  • New deprecation mechanism through the

  • skcriteria.utils.decorators.deprecated decorator.

Version 0.5

In this version scikit-criteria was rewritten from scratch. Among other things:

  • The model implementation API was simplified.

  • The Data object was removed in favor of DecisionMatrix which implements many more useful features for MCDA.

  • Plots were completely re-implemented using Seaborn.

  • Coverage was increased to 100%.

  • Pipelines concept was added (Thanks to Scikit-learn).

  • New documentation. The quick start is totally rewritten!

Full Changelog: https://github.com/quatrope/scikit-criteria/commits/0.5

Version 0.2

First OO stable version.

Version 0.1

Only functions.