About the Event:

Speaker: Matthew Nunes - University of Bath

Abstract: Environmental and energy time series often feature complex time-varying characteristics, such as temporal and cross-series dependence.  In this talk, I consider time series analysis problems in which the data are observed at nodes of a potentially large network structure. These data might evolve according to some  "network-spatial" dependence; some data will have an observed network, with others we can consider how to construct a network. I will introduce new models for such data, related to traditional, well-loved time series models and discuss a practical approach to analysis to take advantage of their flexibility. I will demonstrate the approach on wind energy series.