Deployment

Docker

This project can be deployed using docker compose. Please run the following commands in the project root directory for:

• Tests

docker compose run tests

• Jupyter lab on the localhost:8888

docker compose up juplab

• Streamlit application, with interactive training session and results dashboard:

docker compose up app

In case changes are introduced in the code or the build pipeline, you can run the above commands by adding the --build option, e.g. docker run tests --build.

Web Application

The current web application allows the user to fit models based on uniform or custom demand. To do so, the user needs to define a demand distribution or upload relevant demand data, then choose the dual-sourcing model parameters and finally setup the neural network architecture and training parameters. Once the user completes these steps, the resulting plots are presented in the Results tab. This process is visualized in the following video: