Construction of Mortality Tables using Bidirectional LSTM Neural Networks to Predict Death Probabilities
DOI:
https://doi.org/10.5753/isys.2022.2230Keywords:
Mortality Table, Pension Funds, Life Expectancy, Forecast, Lee-Carter Model, Neural Network, LSTMAbstract
Mortality Tables are tables structured with mortality data, especially mortality rates observed at all ages, used in pension funds and life insurance markets. This article concerns the application of the neural network model to the construction of future mortality tables, using the Lee-Carter model for comparison. The proposed model was a LSTM (Long-Short Term Memory) Neural Network model, including a bidirectional variation. This network is characterized by the sequential processing of data over time. The data for the prediction came from historical mortality table data prepared by the IBGE (Brazilian Institute of Geography and Statistics) and the Human Mortality Database. The results point to a reasonable use as an auxiliary tool for predicting death probabilities.
Downloads
References
Bishop, C. M. (2006) "Pattern Recognition and Machine Learning". Berlin: Springer.
Bowers Jr., N.L., Gerber, H.U., Hickman, J.C., Jones, D., Nesbitt, C.J. (1986) "Actuarial Mathematics". The Society of Actuaries.
Brownlee, J. (2020) “Long Short-Term Memory Networks with Python”, Edition: v1.7, E-book.
Castro, L. G. (2014) “Nota Técnica sobre a Metodologia adotada pelo Ministério da Previdência Social na Extrapolação das Tábuas de Mortalidade IBGE para as idades acima de 80 anos”. Disponível em: [link]. Acessado em 20/09/2020.
Chollet, F. (2018) “Deep learning with Python”. Shelter Island, NY: Manning Publications Co. Print.
Cordeiro Filho, A. (2017) “Cálculo atuarial aplicado: teoria e aplicações: exercícios resolvidos e propostos”. 2. ed. -[3.Reimpr.]. São Paulo: Atlas.
Escovedo, T. e Koshiyama, A. (2020) “Introdução a Data Science - Algoritmos de Machine Learning e métodos de análise”, Casa do Código.
Goodfellow, I., Yoshua, B., Courville, A. (2016), “Deep Learning”. MIT Press.
Lee, R. D. e Carter, L. R. (1992) “Modelling and Forecasting U.S. Mortality”. Journal of the American Statistical Association. 87, 659-671.
Lindemann, B., Müller, T., Vietz, H., Jazdi, N. Weyrich, M. (2021) “A survey on long short-term memory networks for time series prediction”. Procedia CIRP, v. 99, pp. 650-655.
Lindholm, M., Palmborg, L. (2021) “Efficient Use of Data for LSTM Mortality Forecasting”. SSRN Electronic Journal, January 2021.
London, C. e Layton, E. (1899); New York, The Spectator company; [etc, etc.], Transactions of the second International Actuarial Congress, held in the hall of the Institute of Actuaries, Staple Inn, Holborn, London, May 16 to 20, 1898, by International Congress of Actuaries (2nd : 1898 : London). Disponível em: [link]. Acessado em: 20/09/2020.
Nascimento, J.D. e Escovedo, T. (2021) “Construction of Mortality Tables using LSTM Neural Networks”. In XVII Brazilian Symposium on Information Systems (SBSI).
Nigri, A, Levantesi, S., Marino, M., Scognamiglio, S., Perla, F. (2019) “A Deep Learning Integrated Lee–Carter Model”. Risks, v. 7, 33.
Pascariu, M.D., Canudas–Romo, V., Vaupel, J.W. (1992) “Modelling and forecasting mortality: The Double-Gap Life Expectancy Forecasting Model”. University of Southern Denmark. October.
Perla, F., Richman, R., Scognamiglio, S., Wuthrich, M. V. (2021) “Time-Series Forecasting of Mortality Rates using Deep Learning”. Scandinavian Actuarial Journal, v. 2021, 7.
Petneházi, G., Gáll, J. (2019) “Mortality rate forecasting: can recurrent neural networks beat the Lee-Carter model?”. Disponível em: [link]. Acesso em 20/02/2022.
Pirovano, S. (2020) “Projection of Mortality Tables: A Neural Network Approach”. Master of Science in Statistical and Actuarial Sciences Thesis. Università Cattolica del Sacro Cuore Milano. Disponível em: [link]. Acesso em 20/02/2020.
Rodrigues, J.A. (2008) “Gestão de Risco Atuarial”. 1. ed. São Paulo: Saraiva.
Souza, S. (2007) “Seguros: contabilidade, atuária e auditoria”. 2. ed.. São Paulo: Saraiva.
Zaccone, G., Karim, R., e Menshawy, A. (2018) “Deep Learning with TensorFlow”, Birmingham: Packt Publishing. Print.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2022 The authors
This work is licensed under a Creative Commons Attribution 4.0 International License.
