Συγκριτική μελέτη στοχαστικών μοντέλων θνησιμότητας και εφαρμογές με στοιχεία του πληθυσμού της Ελλάδας
Α comparative study of stochastic mortality models and applications using Greek population data
View/ Open
Keywords
Θνησιμότητα ; Στοχαστικά μοντέλαAbstract
The two core subjects of this thesis are the stochastic mortality models and the application of the Lee-Carter model to the Greek population. The research objectives were approached in a twofold way, both from a bibliographical and an empirical viewpoint. At the level of bibliographical discussion, the basic concepts of mortality and its prediction were initially defined, while the Lee-Carter model was presented in meticulous detail, and, then, was critically commented upon. Further, descriptions of the characteristics of other stochastic models of mortality prediction were also presented. This was followed by the application of the Lee-Carter model on empirical mortality data of the Greek population. More specifically, yearly mortality data by age (ages 0-108) for the period 1981-2019 were used to predict mortality for the period 2020-2039. The analysis, modeling and extraction of the results were implemented using the statistical environment R version 4.2.2. The results derived from the application of the stochastic model, showed a slight decline in mortality for the abovementioned period and a 66.6% explanatory power of the variability for the general population. The measures of modeling error were relatively low (MSE=0.001 and MAPE=0.09).