Stay Informed:

COVID-19 (coronavirus) information
Zoom Links: Zoom Help | Teaching with Zoom | Zoom Quick Guide

UCSC-SOE-17-17: Joint Modeling of Longitudinal Relational Data and Exogenous Variables

Rajarshi Guhaniyogi and Abel Rodriguez
10/23/2017 09:07 AM
Applied Mathematics & Statistics
A fundamental aspect of relational data, such as data from social network along
with the attributes of its constituent actors, is the possibility of dependence between
network and the attributes over time. This article proposes a time varying stochastic
framework that jointly models co-evolution of the network and the attributes over
time. To be more specifi c, we propose time varying stochastic latent factor models
with shared latent parameters in modeling the network and the actor attributes. Our
model derives multiple advantages over the existing literature. Unlike the popular co-
evolution models, the proposed framework is flexible enough to allow dynamic actor
attributes to be measured in both ordinal and continuous scale. It speci fically provides
model based assessment of the set of predictors jointly in influencing relation between
nodes. Additionally, the model is easy to compute and readily yields inference and
prediction for missing link between nodes. We employ our model framework to study
co-evolution of international relations between 22 countries and the country speci c
indicators over a period of 11 years.

UCSC-SOE-17-17