Recent Releases of calibrated-explanations
calibrated-explanations - v0.5.1
v0.5.1 - 2024-11-27
Features
- String Targets Support:** Added support for string targets, enhancing flexibility in handling diverse datasets. Special thanks to our new contributor ww-jermaine for the efforts on this feature (issue #27).
- Out-of-Bag Calibration: Introduced support for out-of-bag calibration when using random forests from
sklearn, enabling improved calibration techniques directly within ensemble models. See the new notebook for examples. - Documentation Enhancements: Updated and refined documentation, including fixes to existing sections and the addition of doctests for helper functions to ensure accuracy and reliability.
- Minor updates: Added a
calibratedparameter to thepredictandpredict_probamethods to allow uncalibrated results. ### Fixes - Bug Fixes: Resolved multiple bugs to enhance stability and performance across the library.
What's Changed
- Implement support for string representation of classification targets by @ww-jermaine in https://github.com/Moffran/calibrated_explanations/pull/55
- Add support for string representation of classification targets in CalibratedExplainer class by @ww-jermaine in https://github.com/Moffran/calibrated_explanations/pull/57
- docs: remove duplicate Notes sections by @rudymatela in https://github.com/Moffran/calibrated_explanations/pull/60
- Add the first doctests on CI by @rudymatela in https://github.com/Moffran/calibrated_explanations/pull/61
- chore(deps): bump codecov/codecov-action from 4.6.0 to 5.0.2 by @dependabot in https://github.com/Moffran/calibrated_explanations/pull/63
- chore(deps): bump codecov/codecov-action from 5.0.2 to 5.0.7 by @dependabot in https://github.com/Moffran/calibrated_explanations/pull/64
New Contributors
- @ww-jermaine made their first contribution in https://github.com/Moffran/calibrated_explanations/pull/55
Full Changelog: https://github.com/Moffran/calibrated_explanations/compare/v0.5.0...v0.5.1
- Jupyter Notebook
Published by tuvelofstrom over 1 year ago
calibrated-explanations - v0.5.0
v0.5.0 - 2024-10-15
Features
- Improved the introduction in README.
- Added
calibrated_confusion_matrixinCalibratedExplainerandWrapCalibratedExplainer, providing a leave-one-out calibrated confusion matrix using the calibration set. The insights from the confusion matrix are useful when analyzing explanations, to determine general prediction and error distributions of the model. An example of using the confusion matrix in the analysis is given in paper Calibrated Explanations for Multi-class. - Embraced the update of
crepesversion 0.7.1, making it possible to add a seed when fitting. Addresses issue #43. - Updating terminology and functionality:
- Introducing the concept of ensured explanations.
- Changed the name of
CounterfactualExplanationtoAlternativeExplanation, as it better reflects the purpose and functionality of the class. - Added a collection subclass
AlternativeExplanationsinheriting fromCalibratedExplanations, which is used for collections ofAlternativeExplanation's. Collection methods referring to methods only available in theAlternativeExplanationare included in the new collection class. - Added an
explore_alternativesmethod inCalibratedExplainerandWrapCalibratedExplainerto be used instead ofexplain_counterfactual, as the name of the later is ambiguous. Theexplain_counterfactualis still kept for compatibility reasons but only forwards the call toexplore_alternatives. All files and notebooks have been updated to only callexplore_alternatives. All references to counterfactuals have been changed to alternatives, with obvious exceptions. - Added both filtering methods and a ranking metric that can help filter out ensured explanations.
- The parameters
rnk_metricandrnk_weighthas been added to the plotting functions and is applicable to all kinds of plots. - Both the
AlternativeExplanationclass (for an individual instance) and the collection subclassAlternativeExplanationscontains filter functions only applicable to alternative explanations, such ascounter_explanations,semi_explanations,super_explanations, andensured_explanations. counter_explanationsremoves all alternatives except those changing prediction.semi_explanationsremoves all alternatives except those reducing the probability while not changing prediction.super_explanationsremoves all alternatives except those increasing the probability for the prediction.- The concept of potential (uncertain) explanations is introduced. When the uncertainty interval spans across probability 0.5, an explanation is considered a potential. It will normally only be counter-potential and semi-potential, but can in some cases also be super-potential. Potential alternatives can be included or excluded from the above methods using the boolean parameter
include_potentials. ensured_explanationsremoves all alternatives except those with lower uncertainty (i.e. smaller uncertainty interval) than the original prediction.
- The parameters
- Added a new form of plot for probabilistic predictions is added, clearly visualizing both the aleatoric and the epistemic uncertainty.
- A global plot is added, plotting all test instances with probability and uncertainty as the x- and y-axes. The area corresponding to potential (uncertain) predictions is marked. The plot can be invoked using the
plot(X_test)orplot(X_test, y_test)call. - A local plot for alternative explanations, with probability and uncertainty as the x- and y-axes, is added, which can be invoked from an
AlternativeExplanationor aAlternativeExplanationsusingplot(style='triangular'). The optimal use is when combined with thefilter_topparameter (see below), to include all alternatives, as follows:plot(style='triangular', filter_top=None).
- A global plot is added, plotting all test instances with probability and uncertainty as the x- and y-axes. The area corresponding to potential (uncertain) predictions is marked. The plot can be invoked using the
- Added prerpint and bibtex to the paper introducing ensured explanations:
- Löfström, T., Löfström, H., and Hallberg Szabadvary, J. (2024). Ensured: Explanations for Decreasing the Epistemic Uncertainty in Predictions. arXiv preprint arXiv:2410.05479.
- Bibtex:
bibtex @misc{lofstrom2024ce_ensured, title = {Ensured: Explanations for Decreasing the Epistemic Uncertainty in Predictions}, author = {L\"ofstr\"om, Helena and L\"ofstr\"om, Tuwe and Hallberg Szabadvary, Johan}, year = {2024}, eprint = {2410.05479}, archivePrefix = {arXiv}, primaryClass = {cs.LG} }
- Introduced fast explanations
- Introduced a new type of explanation called
FastExplanationwhich can be extracted using theexplain_fastmethod. It differs from aFactualExplanationin that it does not define a rule condition but only provides a feature weight. - The new type of explanation is using ideas from ConformaSight, a recently proposed global explanation algorithm based on conformal classification. Acknowledgements have been added.
- Introduced a new form av probabilistic regression explanation:
- Introduced the possibility to get explanations for the probability of being inside an interval. This is achieved by assigning a tuple with lower and upper bounds as threshold, e.g.,
threshold=(low,high)to get the probability of the prediction falling inside the interval (low, high]. - To the best of our knowledge, this is the only package that provide this functionality with epistemic uncertainty.
- Introduced the possibility to add new user defined rule conditions, using the
add_new_rule_conditionmethod. This is only applicable to numerical features. - Factual explanations will create new conditions covering the instance value. Categorical features already get a condition for the instance value during the invocation of
explain_factual. - Alternative explanations will create new conditions that exclude the instance value. Categorical features already get conditions for all alternative categories during the invocation of
explore_alternatives. - Parameter naming:
- The parameter indicating the number of rules to plot is renamed to
filter_top(previouslyn_features_to_show), making the call including all rules (filter_top=None) makes a lot more sense. ### Fixes
- Added checks to ensure that the learner is not called unless the
WrapCalibratedExplaineris fitted. - Added checks to ensure that the explainer is not called unless the
WrapCalibratedExplaineris calibrated. - Fixed incorrect use of
np.random.seed.
What's Changed
- chore(deps): update numpy requirement from <2.1,>=1.20 to >=1.20,<2.2 by @dependabot in https://github.com/Moffran/calibrated_explanations/pull/52
- chore(deps): bump codecov/codecov-action from 4.5.0 to 4.6.0 by @dependabot in https://github.com/Moffran/calibrated_explanations/pull/53
- Jupyter Notebook
Published by tuvelofstrom over 1 year ago
calibrated-explanations - v0.4.0
v0.4.0 - 2024-08-23
Features
- Paper updates:
- Calibrated Explanations for Regression has been accepted to Machine Learning. It is currently in press.
- Code improvements:
- Substantial speedup achieved through the newly implemented
explainmethod! This method implements the core algorithm while minimizing the number of calls to core.predict, substantially speeding up the code without altering the algorithmic logic of `calibratedexplanations. Theexplainmethod is used exclusively from this version on when callingexplainfactualorexplaincounterfactual`. - Re-ran the ablation study for classification, looking at the impact of calibration set size, number of percentile samplings for numeric features and the number of features.
- Uploaded a pdf version of the ablation study, making the results easier to overview.
- Re-ran the evaluation for regression, measuring stability, robustness and running times with and without normalization.
- Improved the
safe_importto allowimport ... from ...constructs. - Restructured package
- Added a utils folder:
- Moved discretizers.py to utils
- Moved utils.py to utils and renamed to helper.py
- Made explanations public
- Made VennAbers and interval_regressor restricted
- Experimental functionality introduced:
- Several new experimental features have been introduced. These will be presented as Features once they are thoroughly tested and evaluated.
- Substantial speedup achieved through the newly implemented
- Code interface improvements:
- Added support for the
MondrianCategorizerfrom crepes in theWrapCalibratedExplainer. - Added wrapper functions in
WrapCalibratedExplainerredirecting toCalibratedExplainer: - Including
predict,predict_proba, andset_difficulty_estimator. - Moved any remaining implementations of functions in
WrapCalibratedExplainertoCalibratedExplainer. - Renamed the
plot_allandplot_explanationfunctions toplot. Updated all usages of theplotfunction. - Added
__len__and__getitem__toCalibratedExplanations. __getitem__allow indexing withint,slice, and lists (both boolean and integer lists). When more than one explanation is retrieved, a newCalibratedExplanationsis returned, otherwise, the indexedCalibratedExplanationis returned.
- Added support for the
- Documentation improvements:
- Restructured and extended the documentation.
- Updated the information at the entry page
- Added an API reference
- Improvements of the CI setup:
- Updated CI to run pytest before pylint.
- Updated CI to avoid running tests when commit message starts with 'info:' or 'docs:'.
- Testing improvements
- Improved tests to test
predictandpredict_probafunctions inCalibratedExplainerbetter. - Added several other tests to increase coverage. ### Fixes
- Improved tests to test
- Fixed minor errors in the
predictandpredict_probafunctions inCalibratedExplainer. - Several other minor bug fixes have also been made.
- Jupyter Notebook
Published by tuvelofstrom almost 2 years ago
calibrated-explanations - v0.3.5
v0.3.5 - 2024-07-24
Features
- Made several improvements of the
WrapCalibratedExplainer:WrapCalibratedExplaineris introduced as the default way to interact withcalibrated-explanationsin README.md. The benefit of having a wrapper class as interface is that it makes it easier to add different kinds of explanations.- Documentation of the functions has been updated.
- Initialization:
- The
WrapCalibratedExplainercan now be initialized with an unfitted model as well as with a fitted model. - The
WrapCalibratedExplainercan now be initialized with an already initializedCalibratedExplainerinstance, providing access to thepredictandpredict_probafunctions.
- The
- The
fitmethod will reinitialize the explainer if theWrapCalibratedExplainerhas already been calibrated, to ensure that theexplaineris adapted to the re-fitted model. - Added improved error handling.
- Made several other minor quality improving adjustments.
- Code coverage tests are added and monitored at Codecov.
- Tests are added in order to increase code coverage.
- Unused code is cleaned up.
- Updated the Further reading and citing section in the README:
- Added a reference and bibtex to:
- Löfström, T., Löfström, H., Johansson, U. (2024). Calibrated Explanations for Multi-class. Proceedings of the Thirteenth Workshop on Conformal and Probabilistic Prediction and Applications, in Proceedings of Machine Learning Research. In press.
bibtex @Booklet{lofstrom2024ce_multiclass, author = {Tuwe Löfström and Helena Löfström and Ulf Johansson}, title = {Calibrated Explanations for Multi-Class}, howpublished = {EasyChair Preprint no. 14106}, year = {EasyChair, 2024} }- Updated the docs/citing.md with the above changes. ### Fixes
- Discretizers are limited to the default alternatives for classification and regression. BinaryDiscretizer removed.
__repr__functions added. - Changed the
check_is_fittedfunction to remove ties to sklearn. - Made the
safe_importthrow anImportErrorwhen an import fail.
What's Changed
- Update crepes requirement from <0.7,>=0.6.2 to >=0.7.0,<0.8 by @dependabot in https://github.com/Moffran/calibrated_explanations/pull/44
- Update numpy requirement from <1.27,>=1.20 to >=1.20,<2.1 by @dependabot in https://github.com/Moffran/calibrated_explanations/pull/42
- Bump codecov/codecov-action from 4.0.1 to 4.5.0 by @dependabot in https://github.com/Moffran/calibrated_explanations/pull/46
- Bump actions/cache from 3 to 4 by @dependabot in https://github.com/Moffran/calibrated_explanations/pull/45
- Jupyter Notebook
Published by tuvelofstrom almost 2 years ago
calibrated-explanations - v0.3.4
v0.3.4 - 2024-07-10
Features
- Updated the Further reading and citing section in the README:
- Added a reference and bibtex to:
- Löfström, H., Löfström, T. (2024). Conditional Calibrated Explanations: Finding a Path Between Bias and Uncertainty. In: Longo, L., Lapuschkin, S., Seifert, C. (eds) Explainable Artificial Intelligence. xAI 2024. Communications in Computer and Information Science, vol 2153. Springer, Cham.
bibtex @InProceedings{lofstrom2024ce_conditional, author="L{\"o}fstr{\"o}m, Helena and L{\"o}fstr{\"o}m, Tuwe", editor="Longo, Luca and Lapuschkin, Sebastian and Seifert, Christin", title="Conditional Calibrated Explanations: Finding a Path Between Bias and Uncertainty", booktitle="Explainable Artificial Intelligence", year="2024", publisher="Springer Nature Switzerland", address="Cham", pages="332--355", abstract="While Artificial Intelligence and Machine Learning models are becoming increasingly prevalent, it is essential to remember that they are not infallible or inherently objective. These models depend on the data they are trained on and the inherent bias of the chosen machine learning algorithm. Therefore, selecting and sampling data for training is crucial for a fair outcome of the model. A model predicting, e.g., whether an applicant should be taken further in the job application process, could create heavily biased predictions against women if the data used to train the model mostly contained information about men. The well-known concept of conditional categories used in Conformal Prediction can be utilised to address this type of bias in the data. The Conformal Prediction framework includes uncertainty quantification methods for classification and regression. To help meet the challenges of data sets with potential bias, conditional categories were incorporated into an existing explanation method called Calibrated Explanations, relying on conformal methods. This approach allows users to try out different settings while simultaneously having the possibility to study how the uncertainty in the predictions is affected on an individual level. Furthermore, this paper evaluated how the uncertainty changed when using conditional categories based on attributes containing potential bias. It showed that the uncertainty significantly increased, revealing that fairness came with a cost of increased uncertainty.", isbn="978-3-031-63787-2" }- Updated the docs/citing.md with the above changes. ### Fixes
- Changed np.Inf to np.inf for compatibility reasons (numpy v2.0.0).
- Updated requirements for numpy and crepes to include versions v2.0.0 and v0.7.0, respecitvely.
- Jupyter Notebook
Published by tuvelofstrom almost 2 years ago
calibrated-explanations - v0.3.3
v0.3.3 - 2024-05-25
Features
- Changed how probabilistic regression is done to handle both validity and speed by dividing the calibration set into two sets to allow pre-computation of the CPS. Credits to anonymous reviewer for this suggestion.
- Added updated regression experiments and plotting for revised paper.
- Added a new
under the hooddemo notebook to show how to access the information used in the plots, like conditions and uncertainties etc. ### Fixes - Several minor updates to descriptions and notebooks in the repository.
- Jupyter Notebook
Published by tuvelofstrom about 2 years ago
calibrated-explanations - v0.3.2
v0.3.2 - 2024-04-14
Features
- Added Fairness experiments and plotting for the XAI 2024 paper. Added a
Fairnesstag for the weblinks. - Added multi-class experiments and plotting for upcoming submissions. Added a
Multi-classtag for weblinks. - Some improvements were made to the multi-class functionality. The updates included updating the VennAbers class to a more robust handling of multi-class (with or without Mondrian bins). ### Fixes
- Updated the requirement for crepes to v0.6.2, to address known issues with some versions of python.
- The pythonpath for pytest was added to pyprojects.toml to avoid module not found error when running pytest locally.
- Jupyter Notebook
Published by tuvelofstrom about 2 years ago
calibrated-explanations - v0.3.1
Features
- Added support for Mondrian explanations, using the
binsattribute. Thebinsattribute takes a categorical feature of the size of the calibration or test set (depending on context) indicating the category of each instance. For continuous attributes, thecrepes.extras.binningcan be used to define categories through binning. - Added
BinaryRegressorDiscretizerandRegressorDiscretizerwhich are similar toBinaryEntropyDiscretizerandEntropyDiscretizerin that it uses a decision tree to identify suitable discretizations for numerical features.explain_factualandexplain_counterfactualhave been updated to use these discretizers for regression by default. In a future version, the possibility to assign your own discretizer may be removed. Updated the Further reading and citing section in the README:
- Updated the reference and bibtex to the published version of the introductory paper:
- Löfström, H., Löfström, T., Johansson, U., and Sönströd, C. (2024). Calibrated Explanations: with Uncertainty Information and Counterfactuals. Expert Systems with Applications, 1-27.
bibtex @article{lofstrom2024calibrated, title = {Calibrated explanations: With uncertainty information and counterfactuals}, journal = {Expert Systems with Applications}, pages = {123154}, year = {2024}, issn = {0957-4174}, doi = {https://doi.org/10.1016/j.eswa.2024.123154}, url = {https://www.sciencedirect.com/science/article/pii/S0957417424000198}, author = {Helena Löfström and Tuwe Löfström and Ulf Johansson and Cecilia Sönströd}, keywords = {Explainable AI, Feature importance, Calibrated explanations, Venn-Abers, Uncertainty quantification, Counterfactual explanations}, abstract = {While local explanations for AI models can offer insights into individual predictions, such as feature importance, they are plagued by issues like instability. The unreliability of feature weights, often skewed due to poorly calibrated ML models, deepens these challenges. Moreover, the critical aspect of feature importance uncertainty remains mostly unaddressed in Explainable AI (XAI). The novel feature importance explanation method presented in this paper, called Calibrated Explanations (CE), is designed to tackle these issues head-on. Built on the foundation of Venn-Abers, CE not only calibrates the underlying model but also delivers reliable feature importance explanations with an exact definition of the feature weights. CE goes beyond conventional solutions by addressing output uncertainty. It accomplishes this by providing uncertainty quantification for both feature weights and the model’s probability estimates. Additionally, CE is model-agnostic, featuring easily comprehensible conditional rules and the ability to generate counterfactual explanations with embedded uncertainty quantification. Results from an evaluation with 25 benchmark datasets underscore the efficacy of CE, making it stand as a fast, reliable, stable, and robust solution.} }- Added Code and results for the Investigating the impact of calibration on the quality of explanations paper, inspiring the idea behind Calibrated Explanations.
- Added a bibtex to the software repository:
bibtex @software{Lofstrom_Calibrated_Explanations_2024, author = {Löfström, Helena and Löfström, Tuwe and Johansson, Ulf and Sönströd, Cecilia and Matela, Rudy}, license = {BSD-3-Clause}, title = {Calibrated Explanations}, url = {https://github.com/Moffran/calibrated_explanations}, version = {v0.3.1}, month = feb, year = {2024} }- Updated the docs/citing.md with the above changes.
Added a CITATION.cff with citation data for the software repository.
Fixes
Extended
__repr__to include additional fields whenverbose=True.Fixed a minor bug in the example provided in the README.md and the getting_started.md, as described in issue #26.
Added
utils.transform_to_numericand a clarification about known limitations in README.md as a response to issue #28.Fixed a minor bug in
FactualExplanation.__plot_probabilisticthat was triggered when no features where to be shown.Fixed a bug with the discretizers in
core.Fixed a bug with saving plots to file using the
filenameparameter.
What's Changed
- Bump actions/setup-python from 4 to 5 by @dependabot in https://github.com/Moffran/calibrated_explanations/pull/24
- Add notes encouraging PRs on the markdown docs by @rudymatela in https://github.com/Moffran/calibrated_explanations/pull/25
Full Changelog: https://github.com/Moffran/calibrated_explanations/compare/v0.3.0...v0.3.1
- Jupyter Notebook
Published by tuvelofstrom over 2 years ago
calibrated-explanations - v0.3.0
Features
- Updated to version 1.4.1 of vennabers. Added
precision=4to the fitting of the vennabers model to increase speed. - Preparation for weighted categorical rules implemented but not yet activated.
- Added a state-of-the-art comparison with scripts and notebooks for evaluating the performance of the method in comparison with
LIMEandSHAP: see ClassificationExperimentsota.py and ClassificationAnalysissota.ipynb for running and evaluating the experiment. Unzip results_sota.zip and run ClassificationAnalysissota.ipynb to get the results used in the paper Calibrated Explanations: with Uncertainty Information and Counterfactuals. - Updated the parameters used by
plot_allandplot_explanation. ### Fixes - Filtered out extreme target values in the quickstart notebook to make the regression examples more realistic.
- Fixed bugs related to how plots can be saved to file.
- Fixed an issue where
add_conjunctionswithmax_rule_size=3did not work.
- Jupyter Notebook
Published by tuvelofstrom over 2 years ago
calibrated-explanations - v0.2.3
Features
- Added an evaluation folder with scripts and notebooks for evaluating the performance of the method.
- One evaluation focuses on stability and robustness of the method: see ClassificationExperimentstab_rob.py and ClassificationAnalysisstab_rob.ipynb for running and evaluating the experiment.
- One evaluation focuses on how different parameters affect the method regarding time and robustness: see ClassificationExperimentAblation.py and ClassificationAnalysisAblation.ipynb for running and evaluating the experiment.
Fixes
- Fix in
CalibratedExplainerto ensure that greater-than works identical as less-than. - Bugfix in
FactualExplanation._get_rules()which caused an error when categorical labels where missing.
- Jupyter Notebook
Published by tuvelofstrom over 2 years ago
calibrated-explanations - v0.2.2
v0.2.2 - 2023-10-03
Fixes
Smaller adjustments and fixes.
- Jupyter Notebook
Published by tuvelofstrom over 2 years ago
calibrated-explanations - v0.2.1
v0.2.1 - 2023-09-20
Fixes
The wrapper file with helper classes CalibratedAsShapExplainer and CalibratedAsLimeTabularExplanainer has been removed. The as_shap and as_lime functions are still working.
- Jupyter Notebook
Published by tuvelofstrom over 2 years ago
calibrated-explanations - v0.2.0
v0.2.0 - 2023-09-19
Features
- Added a
WrapCalibratedExplainerclass which can be used for both classificaiton and regression. - Added quickstart_wrap to the notebooks folder.
- Added LIME_comparison to the notebooks folder. ### Fixes
- Removed the dependency on
shapandscikit-learnand closed issue #8. - Updated the weights to match LIME's weights (to ensure that a positive weight has the same meaning in both).
- Changed name of parameter
y(representing the threshold in probabilistic regression) tothreshold.
- Jupyter Notebook
Published by tuvelofstrom almost 3 years ago
calibrated-explanations - v0.1.1
v0.1.1 - 2023-09-14
Features
- Exchanged the slow
VennABERS_by_deffunction for theVennAbersclass in thevenn-aberspackage. ### Fixes - Low and high weights are correctly assigned, so that low < high is always the case.
- Adjusted the number of decimals in counterfactual rules to 2.
- Jupyter Notebook
Published by tuvelofstrom almost 3 years ago
calibrated-explanations - v0.1.0
Version v0.1.0 Release Notes (Initial Release)
🎉 Welcome to the first release of calibrated-explanations!
Features
- Performance: Fast, reliable, stable and robust feature importance explanations.
- Calibrated Explanations: Calibration of the underlying model to ensure that predictions reflect reality.
- Uncertainty Quantification: Uncertainty quantification of the prediction from the underlying model and the feature importance weights.
- Interpretation: Rules with straightforward interpretation in relation to the feature weights.
- Factual and Counterfactual Explanations: Possibility to generate counterfactual rules with uncertainty quantification of the expected predictions achieved.
- Conjunctive Rules: Conjunctive rules conveying joint contribution between features.
Enhancements
- Multiclass Support: Multiclass support has been added since the original version developed for the paper Calibrated Explanations: with Uncertainty Information and Counterfactuals.
- Regression Support: Support for explanations from standard regression was developed and is described in the paper Calibrated Explanations for Regression.
- Probabilistic Regression Support: Support for probabilistic explanations from standard regression was added together with regression and is described in the paper mentioned above.
- Conjunctive Rules: Since the original version, conjunctive rules has also been added.
- Code Structure: The code structure has been improved a lot. The
CalibratedExplainer, when applied to a model and a collection of test instances, creates a collection class,CalibratedExplanations, holdingCalibratedExplanationobjects, which are eitherFactualExplanationorCounterfactualExplanationobjects. Operations can be applied to all explanations in the collection directly throughCalibratedExplanationsor through each individualCalibratedExplanation(see the documentation).
Bug Fixes
- Bug Fixes: Numerous.
Installation
To install this version, you can use pip:
```bash pip install calibrated-explanations==0.1.0
- Jupyter Notebook
Published by tuvelofstrom almost 3 years ago