15.06.2016: A gentle introduction to network data analysis (Prof. Dr. Göran Kauermann)

Statistical Models for Network Data Analysis – A Gentle Introduction

Abstract:

We give an introduction to statistical models for network data. Starting from ‘simple’ models we concentrate on Exponential Random Graph Models (ERGM) as a common tool for network data analyses. The ERGM describes the distribution of a network graph with an exponential family distribution, where the statistics are counts of edges, k-stars or triangles, for instance. This allows for meaningful interpretations of network data, as demonstrated in the talk.

Though the model class mirrors welcome properties of exponential families, the fitting of ERGMs is numerical a burden due to a non-feasible normalization constant in the exponential model. We discuss the state of the art of MCMC based estimation routines, which is used to extend the ERGM in two ways. First we include random nodal effects to compensate for heterogeneity in the nodes of a network. Second, we propose the use of smooth, non-linear network statistics to allow for more data driven modeling.

As applications we will discuss two network examples. The first network describes the international arm trade between states and its changes over time. The second example comes from facebook and describes the connection between facebook users.

This lecture takes place as part of the lecture series Vortragsreihe Psychologische Wissenschaft. 

For more information on Prof. Dr. Kauermann, please visit his homepage at the Ludwig-Maximilians-University in Munich.