Reduced-Rank Factor Models for Network and Relational Array Data

• Activity Number: 379
• Type: Invited
• Date/Time: Tuesday, August 3, 2010 : 2:00 PM to 3:50 PM
• Sponsor: General Methodology
• Author(s): Peter Hoff*+
• Companies: University of Washington
• Keywords: multiplicative model ; PARAFAC ; regularization ; factor analysis ; Bayes ; MCMC

Abstract

Networks and other types of relational data are often represented as matrices or arrays. For example, social network data recorded at a set of discrete time points can be represented as a three dimensional array with sender, receiver and time dimensions. Often, the main patterns in such data can be represented in terms of a low-dimensional (reduced-rank) matrix or array. In this talk we present statistical estimation methods which allow for the incorporation of reduced rank arrays into more complex models for relational data, including binary network data or longitudinal relational data.