00000ctm a22000004a 4500 UP-99796217611000817 Buklod 20131127142505.0 m |o d | ta 131127s2013 xx d r |||| u| (iLib)UPD-00206905871 DSTC eng LG 995 2013 S8 L83 Lucagbo, Michael Daniel C. Estimation of isotonic spatio-temporal model with clustering. Michael Daniel C. Lucagbo. Quezon City University of the Philippines c2013. v, 60p ill. Thesis ( M.S.) --University of the Philippines. Abstract (The concept of contagion has gained a foothold in mathematical epidemiology and in the social sciences. Models which describe the transmission of a contagious entity have found applications in various disciplines. Spatio-temporal models are appropriate for describing transmission (of a disease, for instance) since they account for both the incidence of the disease from the time of its outbreak to the presents, as well as the pattern of spreading from the source to the susceptible units. This study introduces a spatio-temporal model of contagion, with possible clustering of spatial units. In addition, the study purposes the backfitting algorithm as an estimation procedure for the model. The model assumes that the degree of contamination in a spatial units is a function of the population, the incidence of the disease in the recent past, and the distance of the spatial unit from the source. FD-GMM bootstrap and monotone regression via integrated B-spline are embedded into the backfitting algorithm. The prediction performance of the estimation procedure is studied for varying cluster sizes, number of clusters, time series lengths, stationarity conditions, and degrees of misspecification. The results show that the procedures performs test when the model is correctly specified, the model is stationary, there are few clusters, the time series is long or the cluster sizes are much greater than the time series length). Fi-Thesis Fi Monograph UPD DARCHIVES LG 995 2013 S8 L83 UPD DSTC LG 995 2013 S8 L83 Thesis