A1155
Title: Modeling longitudinal area data with zero-modified via GEE
Authors: Hong-Ding Yang - National Chiayi University (Taiwan) [presenting]
Abstract: Benefiting from the assumptions of the hurdle model, a parameter estimation method is proposed under the generation mechanism of repeated measurements of area data with zero-modified. However, the spatiotemporal correlation of data is unknown in practice; generalized estimating equations (GEE) are used to estimate the regression coefficients. GEE exhibits robustness even when the underlying distribution of the reaction is unknown. Therefore, it is assumed that the marginal distribution of responses follows a hurdle binomial distribution, accommodating the zero-inflation and zero-contraction cases. Furthermore, an iterative non-parametric technique is employed to update the working correlation matrix and utilize a jackknife approach to approximate the estimated variance of GEE, resulting in more valid and reliable interval estimates. Numerical results show that the proposed method is promising for analyzing complex spatiotemporal data with zero-modification characteristics. Comparative analyses demonstrate the superiority of the approach over alternative methods based on mean squared error measurements.
