Logistic and CSG Growth Models for Predicting Life Expectancy

Abstract

The tables that allow calculating the probability of death at a certain age by recording the number of births/deaths in a population are called life tables. The concept of life expectancy, which is a measure that determines how long a creature will live, is also determined by mortality rates obtained from life tables. It is also possible to model the expected lifetime with some nonlinear mathematical functions. One of the functions that is often used in modeling mortality rates is the logistic growth function. This study aims to propose a model that can be used as an alternative to the logistic growth model and to interpret the mortality rates of countries. In this study, the life expectancy of males and females in Türkiye, Singapore, Norway, and China was modeled using the logistic and the CSG growth model, which was newly introduced to the literature. When modeling the life expectancy of countries, the adjusted graph was drawn following the data of each growth model. Then, the performances of the logistic growth model and the CSG growth model were compared with R^2, RMSE, and MAPE statistical criteria. As a result of the comparison, it was revealed that the CSG growth model is more suitable than the logistic model for estimating life expectancy for overall data and for each gender. The originality of this study is the CSG model which is a new nonlinear model that predicts life expectancy effectively for related datasets.