This project is my bachelor's thesis. It is about training a graph neural network to distinguish between neutral graph and segregated/integrated graph.
The repository contains code for producing graphs and calculating metrics for them.
To run datasets generation, run dataset.py. In this file's code you can change the desired number of graphs for train, valid and test datasets.
Parameters specified in globalenv.py can be modified.
GNN model Code for producing visualization of the graphs and metrics
Many thanks for Vladimir Maksimenko, who provided supervision for this project.