An interactive visualization system for identifying and understanding biases in machine learning models.
A live demo is available at the following link: https://visual-auditor.surge.sh/
It runs on most modern web browsers. We suggest you use Google Chrome.
This section describes how you to up the visual auditor web application for development.
Run the following commands:
$ git clone https://github.com/poloclub/visual-auditor.git
$ cd visual-auditor
$ cd visual-auditor-app
$ npm installIt's unlikely, but you may need to install some basic JavaScript-related dependencies (e.g., npm).
In the project directory, you can run:
Runs the app in the development mode.
Open http://localhost:3000 to view it in the browser.
The page will reload if you make edits.
You will also see any lint errors in the console.
Builds the app for production to the build folder.
It correctly bundles React in production mode and optimizes the build for the best performance.
The build is minified and the filenames include the hashes.
Your app is ready to be deployed!
See the section about deployment for more information.
This section describes how to set up the visual auditor notebook widget for development.
Run the following commands:
$ git clone https://github.com/poloclub/visual-auditor.git
$ cd visual-auditor
$ cd visual-auditor-package
$ npm installIt's unlikely, but you may need to install some basic JavaScript-related dependencies (e.g., npm).
Launch Jupyter Notebook (or a computational notebook of your choice) and navigate to the visual-auditor-package/notebook-widget/visual-auditor directory. Choose between the adult.ipynb, german_credit.ipynb, or customer_churn.ipynb demo notebook files to test the visual auditor within an example data science workflow.
In the project directory, you can run:
Builds the app for production to the build folder.
It bundles the application into a single index.html file.
To update the notebook widget, copy the contents of this file over to the bundle.html file within the notebook-widget/visual-auditor directory.
The Visual Auditor was developed and maintained by David Munechika, Jay Wang, and Polo Chau from the Polo Club of Data Science at Georgia Tech.
The Visual Auditor is available under the MIT License. The Visual Auditor uses the D3.js which is licensed under the ISC License and React.js which is licensed under the MIT License.