Leto Peel

Research Homepage


For any questions or queries regarding the code, please contact me.

Multiscale mixing

We introduce an approach to localise Newman's assortativity measure so that we can describe the assortativity, across multiple scales, at the node level. Python code is available here. See our paper on multiscale mixing patterns in networks for further details.

The NeoSBM

A new type of stochastic blockmodel (SBM) that performs community detection and explores the space of partitions by interpolating between a metadata partition and the identified communities. Python code is available here. See our paper on the ground truth about metadata and community detection for further details.

2-Step Label Propagation

A method for semi-supervised learning in complex networks. Specifically we address the issue that class labels may not necessarily be assortative (i.e., connect to other nodes with the same class label). A Jupyter notebook is available here. For more details, see the paper

Change-point Detection in Time Evolving Networks

Python code for detecting change points in time evolving networks. See the included readme or run "python runNetworkChangePoint.py -h" for more info.

IMPORTANT: this code requires a specific version of the dendropy library available here

Active Discovery of Network Roles for Predicting the Classes of Network Nodes

Python code for active discovery of network roles for predicting the classes of network nodes. This code works with NetworkX graphs. Runs from the command line, run "python activelearn.py -h" or view the readme file for more info.
See the paper for more details.