基于AEL-STM改进模型的高龄人口死亡率预测  

Prediction of Mortality of Elderly Population with an Improved AE-LSTM Model

作　　者：杨刚 易艳萍 孙超 Yang Gang;Yi Yanping;Sun Chao(School of science,Hunan Technology and Business University,Changsha 410205,China)

机构地区：[1]湖南工商大学理学院,长沙410205

出　　处：《数学理论与应用》2022年第3期100-113,共14页Mathematical Theory and Applications

基　　金：国家社科基金面上项目(No.15BJY122);湖南省教育厅科学研究重点项目(No.19A280)资助。

摘　　要：高龄人口死亡率预测是长寿风险度量和管理、养老金成本和债务评估的基础.基于高龄人口死亡率数据特征,本文建立一个AEL-STM改进模型对高龄人口死亡率进行预测.首先利用AE模型从高龄人口死亡率数据提取潜在时间因子,把它作为LSTM模型的输入变量,然后通过解码得到高龄人口死亡率预测值.同时,选取我国大陆1994–2018年60–89岁高龄人口死亡率作为样本数据进行实证分析.研究结果表明,AELSTM改进模型较传统的人口死亡率CBD模型预测精度有显著提高,且预测结果呈现较强鲁棒性.Prediction of mortality of the elderly population is the basis of longevity risk measurement and management,assessment of pension cost and debt.Based on the characteristics of mortality data of the elderly population,an improved AE-LSTM model is proposed to predict the mortality of the elderly population.Firstly,the AE model is used to extract the potential time factor from the mortality data of the elderly population.Then,the potential time factor is used as the input variable of the LSTM model.The mortality prediction value of the elderly population is obtained by decoding the AE model.At the same time,the mortality rate of the elderly aged 60–89 in China mainland from 1994 to 2018 is selected as the sample data for empirical analysis.The results show that the prediction accuracy of the improved AELSTM model is significantly higher than that of the traditional CBD model,and the prediction results show strong robustness.

关 键 词：高龄人口死亡率 AEL-STM改进模型 CBD模型 死亡率预测 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] C924.2[自动化与计算机技术—控制科学与工程]