SportsSchedulingRLGNN

Main Workflow

1. GraphXMLGeneration.py creates synthetic instances with a predefined # of teams and a guaranteed solution.
2. Graph.py generates Graph objects from instance xml files and saves them as pickles.
3. training.py trains RLAgent.py over the preprocessed instances from 1.
   • Hyperparams are stored in training.py
   • RLAgent.py is paramaterized by either s2v_scheduling.py or s2v_schedulingNew.py
4. RLAgent.py is evaluated against GreedyAgent.py and RandomAgent.py using the script in evaluation.py

Secondary Files

• GraphSummary.py generates a CSV summary of Graph instances
• NodeAndEdge.py a supporting class for Graph.py
• tablesandfigures.py generates training curves
• GraphVis.py creates a visual representation of a given Graph.