Abstract

In this talk, I will introduce two novel nonparametric Bayesian methods for learning spatial homogeneity pattern of functional data. Our methods have the advantage of effectively capturing both locally spatially contiguous clusters and globally discontiguous clusters. Posterior inferences are performed with an efficient Markov chain Monte Carlo (MCMC) algorithm. Simulation studies show that the inferences are accurate and the methods are superior compared to a wide range of competing methods. Two applications including state-level COVID-19 daily growth rates and income distributions across the US will be presented to reveal interesting findings based on proposed methods.