Fairness Service

To assess whether a model is predicting the desirable outcome equally well for all values of a sensitive attribute.

Prerequisites

• Python 3.9.12
• Docker installed if you wish to containerize the application

Installation

Clone the repository to your local machine:

git clone https://github.com/mobile-cloud-computing/spatial-fairness.git

Create and activate environment Variable:

python3 -m venv venv
source venv/bin/activate

Change directory to the cloned repository:

cd Fairness

Add files Add files from the link to the 'data' folder https://tartuulikool-my.sharepoint.com/:f:/g/personal/marasing_ut_ee/EkWFwdVoOX1PqYkxcAewgNoBGtxDn275IC9-Dt7uJ6qZ6g?e=Kc83V5

Install python requirements:

pip install -r requirements.txt

Run the application locally

python3 main.py

Build and run the server using Docker

sudo docker build -t fairness .
sudo docker run -p 8083:8083 fairness

Access Swagger UI documentation at:

• http://localhost:8083/docs
• Endpoint for the service: '/explain_fairness/file'

Debugging

To view the docker containers and images

sudo docker ps
sudo docker images

Clean rebuild docker image

sudo docker rmi -f <docker_image_id>
sudo docker system prune