v2.8.1

Features

• Improved webapp responsiveness and layout consistency (headers, dropdowns, tables, inputs) by @guillaume-vignal — #676
• Refined font sizing with a mix of fixed and responsive units for better cross-screen readability by @guillaume-vignal — #676

Fixes

• Adjusted title height and alignment to avoid layout inconsistencies across browsers by @guillaume-vignal — #676
• Enforced pandas < 3 to prevent breaking changes and ensure data handling stability by @guillaume-vignal — #676

Full Changelog: v2.8.0...v2.8.1

v2.8.0

What's Changed

Major updates in this release:

• WebApp Explainability Expansion: Introduction of several new visual explainability tools in the Shapash WebApp, including Global–Local Feature Importance, Clusters by Explainability, significantly enhancing model interpretation capabilities.
• Modernized Packaging: Full support for numpy>=2.0.0 and dependency modernization, ensuring forward compatibility with the latest scientific Python ecosystem.

---

🚀 New Features & Enhancements

• Add Global–Local Feature Importance Plot to WebApp
  by @guillaume-vignal in #656

  • Introduces a new visualization combining global feature importance with local (instance-level) contributions, bridging the gap between model-level and individual explanations directly in the WebApp.

• Add Clusters by Explainability Plot Function and also to WebApp
  by @guillaume-vignal, @Yh-Cherif in #658, #671, #632

  • Provides a novel map-based visualization to explore and compare individuals based on their explanatory profiles.
  • Enables projection of Shapley contributions into a reduced space, facilitating the identification of explanatory patterns and clusters across observations.

• Add Column Ordering Support for Additional Data in WebApp
  by @guillaume-vignal in #643

  • Allows explicit control over the display order of additional data columns, improving readability and consistency in the WebApp.

• Add _error_ Column Support for Classification
  by @guillaume-vignal in [#663

  • Adds explicit support for classification error tracking in datasets and visualizations.

---

📊 Visualization & Projection Updates

• Add cat_num_threshold Parameter to distribution_plot
  by @guillaume-vignal in #646

  • Improves automatic handling of categorical vs numerical features in distribution plots.

---

⚙️ Technical Improvements & Performance

• Add Support for numpy>=2.0.0 and Modernize Dependencies
  by @guillaume-vignal in #650

  • Updates core dependencies to ensure compatibility with NumPy 2.x and future-proof the library.

• Update pyproject.toml
  by @guerinclement in #651

  • Refines build and packaging configuration in line with modern Python standards.

• Vectorize Classification Error Computation
  by @guillaume-vignal in #666

  • Improves performance and scalability when computing classification errors on large datasets.

---

🐛 Bug Fixes & Robustness

• Synchronize features_dict with Dataset Columns
  by @guillaume-vignalin #661

  • Ensures consistent alignment between declared features and actual dataset columns.

• Rename threading.py to custom_thread.py
  by @guillaume-vignal in #667

  • Prevents shadowing of Python’s built-in threading module, improving compatibility and reliability.

• Change WebApp Favicon
  by @guerinclement in #664

---

🧑‍💻 New Contributors

• @Yh-Cherif made his first contribution

---

Full Changelog

Compare v2.7.10...v2.8.0

v2.7.10

Features

• Added Python 3.13 support in pyproject.toml by @guerinclement — #635
• Refactored typing: replaced typing.List / Tuple with built-in list / tuple (PEP 585) by @guillaume-vignal — #636

Fixes

• Limited Dash to <3.0 due to breaking changes in Dash 3.x by @jasperges — #634
• Handled non-finite values in distance matrix for SmartPlotter.correlations to fix clustering errors by @guillaume-vignal — #638
• Webapp fixes: corrected date filtering and improved boolean type handling with pandas by @ZakariaRida96 — #640

New Contributors

• @jasperges (Dash version fix)
• @ZakariaRida96 (Webapp improvements)

Full Changelog: v2.7.9...v2.7.10

v2.7.9 - Bug Fix Release

What's Changed

• Fix display_model_analysis in Shapash Report to correctly retrieve sklearn version by @guillaume-vignal in #628

Full Changelog: v2.7.8...v2.7.9

v2.7.8 - Bug Fix Release

What's Changed

• Fixed an issue in the report where the contributions plot was not displayed. Thanks to @MLecardonnel in PR #622.

Full Changelog: v2.7.7...v2.7.8

v2.7.7

What's Changed

• Add Option to Display Interaction Plot by @guillaume-vignal in #621
• Restrict Plotly to versions below 6.0.0 due to unexpected bugs occurring randomly in certain Python environmentst by @guillaume-vignal in #621

Full Changelog: v2.7.6...v2.7.7

v2.7.6 - Bug Fix Release

What's Changed

• Limit scikit-learn version to < 1.6.0 by @MLecardonnel in #614

Full Changelog: v2.7.5...v2.7.6

v2.7.5

What's Changed

• Add Customizable Graph Size, Improve Title Alignment, and Fix Color Palette Selection in Shapash Explainability Quality Graphs by @guillaume-vignal in #609
• Enhance Plot Functionality and Consistency for Additional Visualizations by @guillaume-vignal in #612

Full Changelog: v2.7.4...v2.7.5

v2.7.4

What's Changed

• Fix for Feature Importance Local Plot Failure After Global Plot
  Resolved an issue where plotting the feature importance local plot after the global plot resulted in errors due to missing value computations. This fix ensures seamless transitions between global and local plots, improving plot stability and usability.
  Contributed by @guillaume-vignal in #605

• Dynamic Title Height Adjustment for Feature Importance Plot
  Introduced a new feature that dynamically adjusts the title position based on figure height to prevent overlap with plot content. This update provides improved readability and layout flexibility, especially for custom plot dimensions.
  Contributed by @guillaume-vignal in #607

Full Changelog: View the changes for v2.7.3...v2.7.4

v2.7.3 - Bug Fix Release

What's Changed

This release brings important fixes to enhance both the visualization capabilities and the stability of the Shapash package.

• Plot Title Overlap Fix: We have resolved an issue where the plot titles were incorrectly positioned, leading to overlaps with the graph when users specified different figure heights.

• Fix for Shortened Label Duplication: In cases where long feature names were programmatically shortened for better visualization, duplicate labels could lead to missing lines in the correlation matrix.

• Restoration of Missing Files in Shapash Package: Several important files such as *.ipynb, *.html, and *.j2 were missing in previous package versions, causing issues and preventing the generation of reports. These files are now properly included, restoring full functionality to the package.

Contributors

• @guillaume-vignal in PR #603

Full Changelog: v2.7.2...v2.7.3