00000ctm a22000004i 4500 UP-99796217612829789 Buklod 20180823152214.0 m |o d | ta 180823s xx d r |||| u| (iLib)UPD-00367512443 DSTC rda eng DMLUC LG 995 2017 S8 C45 Celoso, Charlene Mae Y. author. Robust estimation of a dynamic spatiotemporal model with structural change for count data by Charlene Mae Y. Celoso. [iii], 131 leaves illustrations text rdacontent unmediated rdamedia volume rdacarrier June 2018. Thesis (M.S. Statistics)--University of the Philippines, Diliman. Abstract (A dynamic spatial-temporal model with count response is estimated with a hybrid of forward search algorithm and bootstrap embedded into the backfitting algorithm. Simulation studies indicated that compared to a similar method that uses Normal Regression, the proposed method performs better in terms of median absolute deviation (MAD) and the bias and standard error of parameter estimates for most of the scenarios that consider the number of locations and time points, the amount of contributions of covariates and the spatial externalities, the type of count data, and the type of structural change. However, caution must be exercised in using this method in prediction for some situations such as zero-inflated counts and the presence of structural change in both covariates and spatial externalities as the MAS under the proposed method is lower than that under the method with Normal Regression). Backfitting algorithm. FI UP UPD DSTC LG 995 2017 S8 C45 Thesis