Title: A Bayesian approach to identify changepoints in spatio-temporal ordered categorical data: An application to COVID-19
Authors: Soudeep Deb - Indian Institute of Management Bangalore (India)
Candace Berrett - Brigham Young University (United States)
Siddharth Rawat - Indian Institute of Management Bangalore (India) [presenting]
Abstract: Although there is substantial literature on identifying structural changes for continuous spatio-temporal processes, that is not true for categorical spatio-temporal data. The purpose is to bridge that gap and propose a novel spatio-temporal model to identify changepoints in ordered categorical data. The model leverages an additive mean structure with separable gaussian space-time processes. Our proposed technique is defined in such a way that it can detect a shift in the mean structure as well as in the covariance structures in both the spatial and temporal associations. Our approach's capability to handle ordinal categorical data provides an added advantage from an application perspective. We implement the model through a Bayesian framework, which gives a computational edge over a classical method. For application, we use county-wise COVID-19 data from New York by categorizing the daily cases according to CDC guidelines. Our model is able to identify changepoints in the data and helps in providing interesting insights about the ``waves'' encountered during the pandemic.