Time-series forecasting of mortality rates using deep learning

被引：57

作者：

Perla, Francesca ^{[1
]}

Richman, Ronald ^{[2
]}

Scognamiglio, Salvatore ^{[1
]}

Wuthrich, Mario, V ^{[3
]}

机构：

[1] Univ Naples Parthenope, Dept Management & Quantitat Sci, Naples, Italy

[2] QED Actuaries & Consultants, Johannesburg, South Africa

[3] Swiss Fed Inst Technol, RiskLab, Dept Math, Zurich, Switzerland

来源：

SCANDINAVIAN ACTUARIAL JOURNAL | 2021年 / 2021卷 / 07期

关键词：

Mortality forecasting; recurrent neural networks; convolutional neural networks; representation learning; time-series forecasting; Lee– Carter model; human mortality database; STOCHASTIC MORTALITY; NEURAL-NETWORKS;

D O I：

10.1080/03461238.2020.1867232

中图分类号：

O1 [数学];

学科分类号：

0701 ; 070101 ;

摘要：

The time-series nature of mortality rates lends itself to processing through neural networks that are specialized to deal with sequential data, such as recurrent and convolutional networks. The aim of this work is to show how the structure of the Lee-Carter model can be generalized using a relatively simple shallow convolutional network model, allowing for its components to be evaluated in familiar terms. Although deep networks have been applied successfully in many areas, we find that deep networks do not lead to an enhanced predictive performance in our approach for mortality forecasting, compared to the proposed shallow one. Our model produces highly accurate forecasts on the Human Mortality Database, and, without further modification, generalizes well to the United States Mortality Database.

引用

页码：572 / 598

页数：27