bytjn1416124 / ai-recruitment-bias

The Bias Portfolio

This app is a web-based dashboard which analyses the input and output of an algorithm to identify where undesirable biases are present. Often, protected classes (such as gender) are encoded in a set of unprotected variables, therefore a naive algorithm is able to discriminate against (e.g.) gender without using access to the protected class in the first place.

Simple 'fairness' methods do not identify these relationships and are therefore unable to address key issues in the measurement methodology. This multivariate analysis highlights variables which are highly correlated with protected classes.

Resources

https://dash.plot.ly/

Installation

Clone the project and navigate to the folder. git clone https://github.com/rosscg/ai-recruitment-bias cd ai-recruitment-bias

Create and activate virtual environment (Mac). python3 -m venv venv source venv/bin/activate

Create and activate virtual environment (Windows Anaconda). conda update conda conda create -n venv python=3.6 anaconda source activate venv

Install dependencies with the command: pip install -r requirements.txt

Run the app with the command: python app.py

Visit http:127.0.0.1:8050/ in your web browser

Viewing Notebook

Run: jupyter notebook analysis/notebook.ipynb