Changelog
All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
[1.0.2.1] - 2023-06-16
Fixed
Colours in performance gain analysis plot
[1.0.2] - 2023-06-14
Added
Performance gain analysis plot
[1.0.1.2] - 2023-06-02
Fixed
Imbalance Ratio calculated only on train subset (if it is available)
[1.0.1.1] - 2023-01-26
Fixed
Target encoding class Exception
[1.0.1] - 2023-01-26
The first release of the package
This is the first official release of the package.
It allows to perform all required computations for comparing and benchmarking balancing methods.
[1.0.0] - 2023-01-26
The first release of the package
This is the first official release of the package.
It allows to perform all required computations for comparing and benchmarking balancing methods.
[0.5.0] - 2023-01-26
Added
Centering the ModelProfile plots
Package description in README
Codes for conducting the experiments presented in the thesis
Changed
AutomaticTransformer
andExtensionAutomaticTransformer
contain smaller number of methodsprotein_homo
dataset removed from the benchmarking set
Fixed
Bug with plotting from
ModelProfileExplanationArray
only one variable in one column
[0.4.0] - 2023-01-18
Added
Parallel computing support in
explain
moduleVariable Importance option in
Explainer
andExplainerArray
classesVariable Importance explanation in
ModelPartsExplanation
andModelPartsExplanationArray
classes inexplain
moduleParameter name
curve_type
inModelPartsExplanation
andModelPartsExplanationArray
(previouslyExplainerResult
andExplainerResultArray
) changed toexplanation_type
Variable Importance comparison metric based on Wilcoxon statistical test
FDR correction on the results of multiple Wilcoxon statistical tests
Variable Importance plots and summary plots
Changed
Metric to compare PDP and ALE changed from VOD to SDD
Parameter name
curve_type
inExplainer
andExplainerArray
classes changed toexplanation_type
ExplainerResult
andExplainerResultArray
classes are changed toModelProfileExplanation
andModelProfileExplanationArray
Fixed
The swap of the majority and the minority class when data after balancing is perfectly balanced
[0.3.0] - 2023-01-02
Added
Compare method for
ExplainerResultArray
New plot method to summary benchmarking
Enable compare methods to return raw values
Benchmarking set is also available with only continuous variables
Removed
Two comparison metrics - only one, based on variance, is left
[0.2.0] - 2022-12-21
Added
Example dataset
Benchmarking set
Extracted
requirements.txt
file for environment setup without using PyPIUser manual in documentation
[0.1.0] - 2022-12-06
Added
All main components are implemented
Finished unit tests
Finished CI/CD pipeline