1. An innovation mortality prediction model with cohort effect.
- Author
-
Xiao, Hongmin, Zhao, Miaomiao, Li, Xiang, and Bai, Aiqin
- Subjects
- *
LEAST squares , *SINGULAR value decomposition , *DECOMPOSITION method , *TIME series analysis , *CHINESE people - Abstract
Considering that the cohort effect is added to the population mortality prediction model, the data information on the impact of birth year on mortality can be captured. In this article, the Linear Link model (LL) is extended to obtain the Extended Linear Link model (ELL) with cohort effect. Model fitting and forecasting employ a two-stage approach, first estimating the age term parameters of the extended model, and then forecasting the time parameters using an optimal combination ARIMA(p,d,q) of time series model. Using the mortality data of the Chinese population aged 0–89 from 1995 to 2018 to fit and predict the model, and to test the stability by rolling-window time frame. The results show that when the singular value decomposition method and the least square estimation method are adopted, the prediction effect of the Extended Linear Link model with cohort effect is significantly better than the previous Linear Link model, and the accuracy of fitting and predicting are improved. [ABSTRACT FROM AUTHOR]
- Published
- 2024
- Full Text
- View/download PDF