dash-feature-impact 0.0.1

A Dash component library for visualizing machine learning feature importance and impact. Create intuitive, interactive visualizations that show how individual features contribute to model predictions. Includes force plot visualizations that display feature contributions alongside prediction distributions, helping users understand how each variable drives model outputs. Ideal for model interpretation, debugging, and explaining predictions to stakeholders.

Dash Feature Impact

Dash Feature Impact is a Dash component library.

A Dash component library for visualizing machine learning feature importance and impact. Create intuitive, interactive visualizations that show how individual features contribute to model predictions. Includes force plot visualizations that display feature contributions alongside prediction distributions, helping users understand how each variable drives model outputs. Ideal for model interpretation, debugging, and explaining predictions to stakeholders.

Get started with:

1. Install Dash and its dependencies: https://dash.plotly.com/installation
2. Run python usage.py
3. Visit http://localhost:8050 in your web browser

Contributing

See CONTRIBUTING.md

Install dependencies

If you have selected install_dependencies during the prompt, you can skip this part.

1. Install npm packages

   $ npm install
   

2. Create a virtual env and activate.

   $ virtualenv venv
   $ . venv/bin/activate
   

   Note: venv\Scripts\activate for windows

3. Install python packages required to build components.

   $ pip install -r requirements.txt
   

4. Install the python packages for testing (optional)

   $ pip install -r tests/requirements.txt
   

Write your component code in src/lib/components/DashFeatureImpact.react.js.

• The demo app is in src/demo and you will import your example component code into your demo app.
• Test your code in a Python environment:
  1. Build your code

     $ npm run build
     

  2. Run and modify the usage.py sample dash app:

     $ python usage.py
     

• Write tests for your component.
  • A sample test is available in tests/test_usage.py, it will load usage.py and you can then automate interactions with selenium.
  • Run the tests with $ pytest tests.
  • The Dash team uses these types of integration tests extensively. Browse the Dash component code on GitHub for more examples of testing (e.g. https://github.com/plotly/dash-core-components)
• Add custom styles to your component by putting your custom CSS files into your distribution folder (dash_feature_impact).
  • Make sure that they are referenced in MANIFEST.in so that they get properly included when you're ready to publish your component.
  • Make sure the stylesheets are added to the _css_dist dict in dash_feature_impact/__init__.py so dash will serve them automatically when the component suite is requested.
• Review your code

Create a production build and publish:

1. Build your code:

   $ npm run build
   

2. Create a Python distribution

   $ python setup.py sdist bdist_wheel
   

   This will create source and wheel distribution in the generated the dist/ folder. See PyPA for more information.

3. Test your tarball by copying it into a new environment and installing it locally:

   $ pip install dash_feature_impact-0.0.1.tar.gz
   

4. If it works, then you can publish the component to NPM and PyPI:

   1. Publish on PyPI

      $ twine upload dist/*
      

   2. Cleanup the dist folder (optional)

      $ rm -rf dist
      

   3. Publish on NPM (Optional if chosen False in publish_on_npm)

      $ npm publish
      

      Publishing your component to NPM will make the JavaScript bundles available on the unpkg CDN. By default, Dash serves the component library's CSS and JS locally, but if you choose to publish the package to NPM you can set serve_locally to False and you may see faster load times.

5. Share your component with the community! https://community.plotly.com/c/dash

   1. Publish this repository to GitHub
   2. Tag your GitHub repository with the plotly-dash tag so that it appears here: https://github.com/topics/plotly-dash
   3. Create a post in the Dash community forum: https://community.plotly.com/c/dash