v1.6.1: Patch release: minor fixes

This patch release prevent errors from matplotlib version 3.5.1

v1.6.0: Explainability Quality Metrics

Features
Evaluate the relevance of your explainability using 3 metrics: Stability, Consistency and Compacity

• Stability #234
• Compacity #242
• Consistency #252

New backend added to Shapash : LIME #255

Bug fixes

• Multiple lime fixe (#263, #265, #267)
• replace parameter for acv backend #262
• fix remove files when generate report fail #250

Internal

• Update data loader with urldowload #269

v1.5.0: A better estimation of Shapley values with ACV backend

Features

• New backend added to Shapash : ACV (#245)
  • New backend parameter on the compile method on which we can select 'acv'
  • New tutorial to illustrate how to work with Shapash and ACV
  • More info about ACV here : https://github.com/salimamoukou/acv00
• lightgbm added to Shapash accepted list of models (#232)

Bug fixes

• Fix Shapash Report bug when using multiclass (#223)
• Fix an attribute in SmartExplainer object (#247)

v1.4.4 : patch release

This release fixes bugs in the WebApp (#223)

v1.4.2: patch release

• Fixes a bug with required python version for installation

v1.4.1: patch release

Bug fix :

• Fix Shapash installation with python 3.9

v1.4.0: Sets of features and correlation plot

• Groups of features compatible with:
  • Features importance plot (#188)
  • Contribution plot (#188)
  • Local plot (#207)
  • WebApp (#203)
  • SmartPredictor (#212)
• New tutorial to illustrate how to use groups of features (#211)
• New plot : correlation plot (#210) used in the Shapash report
• Added compatibility with python version 3.9

Shapash Report - Bug Fix

Minor Release - bugs fixed:

• Correlation exception
• generate_report() - kernel_name parameter (Papermill kernel)

Shapash Report

• Standalone report (#137)
  Standalone HTML file that contains the following information :
  • General information about the project
  • Dataset description
  • Model library, parameters, and other specifications
  • Dataset analysis
  • Global explainability of the model
  • Model performance
• Digit counter breaks on 0 (#156)
• Switch from SmartPredictor to SmartExplainer (#132)

Interaction plot and new features for WebApp

• WebApp Features: Title (#118), sort features in mask (#117), run app without prediction (#135)
• Interaction plot: Plot (#124), calculation of top interactions (#120, #122)
• BugFix: Xgboost (#127), Pickle (#131, #133)