Contemporary inference in state space models

  • Gorst-Rasmussen, Anders (Project Participant)
  • Dethlefsen, Claus (Project Participant)
  • Lundbye-Christensen, Søren (Project Participant)

Project Details


Longitudinal data, i.e. data collected over time from one ore more subjects/units is often described by a state space model. A flexible class of models for serially correlated non Gaussian data is generalized linear state space models or dynamic generalized linear models. The sources of variation in such models are 1. variation between subjects, 2. serial correlation within subjects, and 3. observational noise. Internationally this is a field with much activity, especially with application of modern simulation based inference techniques. Cooperation with Niels Trolle Andersen and Jørn Atterman, Biostatistics, Aarhus University; Bent Jørgensen, University of Southern Denmark, Odense.
Effective start/end date19/05/201001/09/2013