A0551
Title: A generalized estimating equation approach to zero-modified responses with spatiotemporal dependence
Authors: Hong-Ding Yang - National Chiayi University (Taiwan) [presenting]
Abstract: A novel hurdle binomial model is proposed within the generalized estimating equations (GEE) framework to address both zero inflation and zero deflation in count data exhibiting spatiotemporal correlation. By assuming the marginal distribution of the response variable follows a hurdle binomial distribution, the approach accommodates deviations from conventional zero-modified models, particularly when the proportion of zeros is low. GEE is employed for robust estimation of regression coefficients under unknown true distributions and iteratively updates the working correlation matrix using a nonparametric technique. Variance estimates are approximated via a jackknife procedure, yielding efficient and reliable inference. Simulation studies demonstrate the proposed method's ability to handle complex spatiotemporal datasets with zero-adjustment features. Comparative analyses against hurdle models estimated using alternative methods, evaluated by mean squared error, reveal the superior performance of the approach.
