Abstract

A network time series is a multivariate time series where the individual series are known to be linked by some underlying network structure. Sometimes this network is known a priori, and sometimes the network has to be inferred, often from the multivariate series itself. Network time series are becoming increasingly common, long, and collected over a large number of variables. We are particularly interested in network time series whose network structure changes over time.

We describe some recent developments in the modeling and analysis of network time series via network autoregressive integrated moving average (NARIMA) process models. NARIMA models provide a network extension to a familiar environment that can be used to extract valuable information and aid prediction. As with classical ARIMA models, trend can impair the estimation of NARIMA parameters. The scope for trend removal is somewhat wider with NARIMA models and we exhibit some possibilities.

We will illustrate the prototypical operation of NARIMA modeling on data sets arising from human and veterinary epidemiology.

This is joint work with Kathryn Leeming (Bristol), Marina Knight (York) and Matt Nunes (Lancaster).