Dynamic Gaussian Process Priors, with Applications to the Analysis of Space-time Data
Dani Gamerman, Esther Salazar, Edna Afonso Reis · 2007
Abstract This paper describes the class of dynamic Gaussian processes. These are obtained as straightforward extensions of Gaussian processes when the time dimension is also considered or of dynamic models when the space dimension is also considered. Properties of dynamic Gaussian processes are derived and illustrative examples are provided. They provide useful components to a number of different models designed to analyze space-time data. Their adequacy and limitations are also discussed. Illustrations to simulated data examples show the potential usefulness of these structures.