Code repository for the Uppsala University Thesis in Data Science
For more details, refer to the full thesis at http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-538891
The installation of libraries listed in the requirements.txt file are required for running the application.
Run the following to install all required libraries:
pip install -r requirements.txtThe factory environment is defined in directory ./custom_environment. The environment is a Gymnasium environment.
The logic of the environment can be updated in ./custom_environment/environment.py.
Use ./custom_environment/environment_factory.py to define environment instantiation, the function create_factory_env
creates an environment instance.
The model files, i.e. ./custom_environment/job.py, ./custom_environment/machine.py, ./custom_environment/recipe.py contain
the blueprint of each model instance. You can update the behaviour of the specific models in these files.
All file names with _factory provide helper methods for easily creating instances of a model.
Run the ./manual_agent.py to start a manual agent and test out the environment.
The dispatch rule agents are edd_agent.py (Earliest Due Date), fifo_agent.py (First-In-First-Out)
and heuristic_agent.py (prioritizes jobs based on process time to deadline ratio).
The RL agents are dqn_agent.py (Deep Q-Network), a2c_agent.py (Asynchronous Advantage Actor Critic).
Code convention best practices is important in collaborative development for consistent readability and maintainability.
Contributions require being formatted and linted using the black and ruff libraries to be merged to the main branch.
Run the following to format all files:
black .Run the following to lint all files:
ruff .