Python 3.6
torch
torch_geometric

This is a small repository for my paper available at: https://arxiv.org/abs/2111.13519.

The 5_fold_cross_validation.py file provides the code to split graph data into sets where k-fold cross validation can be carried out.

The model.py and train_validation_test.py files provide the graph auto-encoder used in the paper and the code to optimise it.

This is just the basic code used in the paper, if you want anything additional feel free to contact me. I do not provide the PyTorch geometric graphs used as the news co-occurrence edges were kindly given for my personal use but if needed I can enquire about their distribution.

If this is found to be helpful in your work consider refrencing the paper:

 @misc{turner2021graph,
   title={Graph Auto-Encoders for Financial Clustering}, 
   author={Edward Turner},
   year={2021},
   eprint={2111.13519},
   archivePrefix={arXiv},
   primaryClass={q-fin.ST}
}.