Debón A, Montes F, Sala R
Dpt. Estadística e Investigación Operativa Aplicadas y Calidad, Universidad Politécnica de Valencia, E-46022, Valencia, Spain.
Lifetime Data Anal. 2006 Jun;12(2):223-44. doi: 10.1007/s10985-006-9005-1. Epub 2006 Jul 1.
The graduation of mortality data aims to estimate the probabilities of death at age x, q ( x ), by means of an age-dependent function, whose parameters are adjusted from the crude probabilities that are directly obtainable from the data. However, current life tables have a problem, the need for periodic updates due to changes in mortality over short periods of time. The table containing mortality rates for different ages in different years, q ( xt ), is called a dynamic life table, which captures mortality variation over time. This paper proposes a review of the most commonly used dynamic models and compares the results obtained by each of them when applied to mortality data from the Valencia Region (Spain). The result of the comparison leads us to the conclusion that the Lee-Carter method offers the best results for both sexes, while that based on Heligman and Pollard functions provides the best fit for men alone. Our working method is of additional interest as it may be applied to mortality data for a wide range of ages in any geographical location, allowing the most appropriate dynamic life table to be selected for the case at hand.
死亡率数据的分级旨在通过一个与年龄相关的函数来估计x岁时的死亡概率q(x)，该函数的参数根据可直接从数据中获得的粗略概率进行调整。然而，当前的生命表存在一个问题，即由于短时间内死亡率的变化需要定期更新。包含不同年份不同年龄死亡率q(xt)的表格称为动态生命表，它捕捉了死亡率随时间的变化。本文对最常用的动态模型进行了综述，并比较了将它们应用于西班牙巴伦西亚地区死亡率数据时各自得到的结果。比较结果使我们得出结论，Lee-Carter方法对男女都提供了最佳结果，而基于Heligman和Pollard函数的方法仅对男性提供了最佳拟合。我们的工作方法具有额外的意义，因为它可以应用于任何地理位置广泛年龄范围的死亡率数据，从而为手头的案例选择最合适的动态生命表。
