Abstract by Easton Huch
ideq: An R Package for Dynamic Spatio-temporal Models
We introduce a new R package titled ideq for dynamic spatio-temporal models, in particular the integrodifference equation (IDE) model. We show the model structure and describe the sampling algorithm used for the models supported by the package. We demonstrate the package's functionality using simulated data. Lastly, we perform a simulation study to compare three model types with varying sizes of state space. We find (1) that the IDE model performs better in terms of predictive accuracy than empirical orthogonal function (EOF) models on simulated data from a diffusion process; (2) that EOF models perform more similarly to each other than to the IDE model in terms of predictive accuracy; and (3) that increasing the state space size tends to improve the models' point estimates, but it can sometimes hamper the models' ability to characterize the predictive uncertainty.