Nonparametric Bayesian grouping methods for spatial time-series data

We describe an approach for identifying groups of dynamically similar locations in spatial time-series data based on a simple Markov transition model. We give maximum-likelihood, empirical Bayes, and fully Bayesian formulations of the model, and describe exhaustive, greedy, and MCMC-based inference methods. The approach has been employed successfully in several studies to reveal meaningful relationships between environmental patterns and disease dynamics.