两种死亡率预测方法的比较

Comparison of Two Mortality Prediction Methods

文章基于中国人口死亡率完整数据和不规则的有限数据,分别运用ARIMA方法和双随机过程对Lee-Carter模型的时间项进行预测,通过残差图和损失函数,对两种预测方法的稳健性进行检验,比较他们的预测效果。结果表明,经典ARIMA方法更适合目前中国的人口死亡率预测,更有利于进行长寿风险的研究。

Based on the complete data and irregular limited data of China’s population mortality rate,this paper uses ARIMA method and double stochastic process to predict the time term of Lee-Carter model respectively.And then,through residual plot and loss function,the paper tests the robustness of the two prediction methods,and also makes comparisons on their prediction effects.The results show that the classical ARIMA method is more suitable for China’s current population mortality rate prediction and is more conducive to longevity risk research.