A0573
Title: Bayesian clustering of spatial functional data based on random spanning trees
Authors: Bohai Zhang - Nankai University, China (China) [presenting]
Huiyan Sang - Texas A\&M University (United States)
Hui Huang - Sun Yat-Sen University (China)
Abstract: A Bayesian wavelets model is proposed for modeling and clustering the spatial functional data, where domain partitioning is achieved by operating on the spanning trees. By imposing a proper prior on the spanning trees, the resulting clusters can have arbitrary shapes and are spatially contiguous in the input domain. The within-cluster parameters are updated through Gibbs samplers, and the between-cluster parameters are updated using a reversible jump Markov chain Monte Carlo (RJ-MCMC) algorithm. The numerical results show that the proposed model can identify the true clusters and yield reasonable parameter estimates. We then apply our model to the mobility dataset of Harris County in Houston during the COVID-19 pandemic.
