Networks on the Right-Hand Side: Relating Networks to Outcomes

• Activity Number: 627
• Type: Invited
• Date/Time: Thursday, August 13, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): Tyler McCormick*
• Companies: University of Washington

Abstract

Scientists have long appreciated the relationship between the social environment and a multitude of individual behaviors. Understanding this relationship presents two statistical challenges: (i) interpretably representing network structure and (ii) linking local network structure to outcomes of interest. This talk explores these challenges in the context of multiview, continuous-time network data. We first present a latent space model for multiview networks, situations where multiple types of relations are observed for the same individuals. We then present preliminary work which develops a Bayesian sparse coding approach to capturing local structure in unstructured observational network data.