Abstract
In this paper we describe the main features of the Bergm package for the open-source R software which provides a comprehensive framework for Bayesian analysis of exponential random graph models: tools for parameter estimation, model selection and goodness-of- fit diagnostics. We illustrate the capabilities of this package describing the algorithms through a tutorial analysis of three network datasets.
Highlights
Interest in statistical network analysis has grown massively in recent decades and its perspective and methods are widely used in many scientific areas which involve the study of various types of networks for representing structure in many complex relational systems such as social relationships, information flows, protein interactions, etc. (see Salter-Townshend, White, Gollini, and Murphy (2012) for a recent review of statistical network models).Social network theory is based on the study of social relations between actors so as to understand the formation of social structures by the analysis of basic local relations
Exponential random graph models (ERGMs) (Frank and Strauss (1986); Wasserman and Pattison (1996); Robins, Pattison, Kalish, and Lusher (2007)) are one of the most important family of models conceived to capture the complex dependence structure of an observed network allowing a reasonable interpretation of the underlying process which is supposed to have produced these structural properties
These are classical graph-theoretic structures such as degrees, cycles, etc. which can be directly incorporated in ERGMs as sufficient statistics with corresponding parameters measuring their importance in the observed network
Summary
Interest in statistical network analysis has grown massively in recent decades and its perspective and methods are widely used in many scientific areas which involve the study of various types of networks for representing structure in many complex relational systems such as social relationships, information flows, protein interactions, etc. (see Salter-Townshend, White, Gollini, and Murphy (2012) for a recent review of statistical network models). The dependence hypothesis at the basis of these models is that the connections between actors (edges) self-organize into small structures called configurations or network statistics These are classical graph-theoretic structures such as degrees, cycles, etc. The Bergm package for R (R Development Core Team 2011) implements Bayesian analysis for Exponential Random Graph Models using the methods described by Caimo and Friel (2011) and Caimo and Friel (2013). The package provides a comprehensive framework for Bayesian inference and model selection using Markov chain Monte Carlo (MCMC) algorithms. The Bergm package has been continually improved in terms of speed performance over the last two years and one of the purposes of this paper is to highlight these improvements We feel that this package offers the end-user a feasible option for carrying out Bayesian inference for exponential random graphs. This paper does not provide an exhaustive description of all the functionality and options available, and more information about the commands and methods mentioned are available through the R help system within the package
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
