shap 0.50.0

A unified approach to explain the output of any machine learning model.

Stars: 25038, Watchers: 25038, Forks: 3474, Open Issues: 592

The shap/shap repo was created 9 years ago and the last code push was 9 hours ago.
The project is extremely popular with a mindblowing 25038 github stars!

How to Install shap

You can install shap using pip

pip install shap

or add it to a project with poetry

poetry add shap

Package Details

Author

None

License

MIT License

Homepage

None

PyPi:

https://pypi.org/project/shap/

Documentation:

https://shap.readthedocs.io/en/latest/index.html

GitHub Repo:

https://github.com/slundberg/shap

Classifiers

• Scientific/Engineering
• Scientific/Engineering/Artificial Intelligence

Related Packages

Errors

A list of common shap errors.

Code Examples

Here are some shap code examples and snippets.

GitHub Issues

The shap package has 592 open issues on GitHub

• build(deps): bump lodash from 4.17.21 to 4.17.23 in /javascript
• Support for channel-wise partitionshap
• build(deps): bump qs and express in /javascript
• rendering colors in SHAP text visualizations
• Failed Additivity Check
• Unpin numba on 3.14
• Make shap.plots.text compatible with numpy2.
• BUG: shap plot text not color in highlightings in Colab
• ENH add nsamples parameter to KernelExplainer call method
• ENH: CoalitionExplainer efficiency
• AttributeError: 'Sequential' object has no attribute 'predict'
• BUG: warnings
• ENH: Fit the model on entire dataset, vs doing a train/test split
• ENH: Keep row index in explanations of DataFrames
• BUG: PermutationExplainer does not expose output_names

See more issues on GitHub