This repository contains the code associated with our paper "A source separation approach to temporal graph modelling for computer networks". It enables reproduction of the experiments presented in the paper, including data preprocessing and analysis of the results.
There are four main directories:
code: implementation of the SNMF model, and Python script enabling its application to CSV-formatted datasets.data: preprocessing scripts for the datasets used in our experiments.results: compressed JSON file containing the detailed evaluation metrics computed in our quantitative experiments.notebooks: Jupyter notebooks designed to help explore and understand the experiment results.
The code is written in Python 3.9. To install the necessary dependencies:
pip install -r requirements.txt
This is sufficient for running the model using the NumPy backend. To use the PyTorch backend (including GPU support), you also need to install torch and torch-sparse.
The experiments.sh script reproduces the experiments presented in
the paper.
To reuse the model in other settings, all necessary classes can be
found in snmf.py.