MoN9: Ninth Mathematics of Networks meeting, 18th June, St Andrews University

Richard Clegg (UCL) - Likelihood methods for comparing network evolution models

In the last twelve years a number of models have been proposed which ‘‘explain" the evolution of networks. Most notably, for example, the rewiring model of a Watts–Strogatz small world network and the preferential attachment model which Albert and Barabasi used to explain power laws in degree distributions.

This presentation describes a rigorous statistical method using likelihood estimates to compare two rival explanations for the evolution of a given target network. The method, known as FETA (Framework for Evolving Topology Analysis) is statistically rigorous and can assess which of several rival models is the most likely explanation for a given target data set. The method is tested on several target networks. Free code and data are provided at the FETA webpage.

Return to previous page

Contact: Richard G. Clegg ( or Keith Briggs (mailto:keith.briggs_at_bt_dot_com)