Forecasting causes of death by using compositional data analysis: the case of cancer deaths

Søren Kjærgaard, Yunus Emre Ergemen, Malene Kallestrup-Lamb, James Oeppen, Rune Lindahl-Jacobsen

Research output: Contribution to journalJournal articleResearchpeer-review

Abstract

Cause‐specific mortality forecasting is often based on predicting cause‐specific death rates independently. Only a few methods have been suggested that incorporate dependence between causes. An attractive alternative is to model and forecast cause‐specific death distributions, rather than mortality rates, as dependence between the causes can be incorporated directly. We follow this idea and propose two new models which extend the current research on mortality forecasting using death distributions. We find that adding age, time and cause‐specific weights and decomposing both joint and individual variation between different causes of death increased the forecast accuracy of cancer deaths by using data for French and Dutch populations.
Original languageEnglish
JournalJournal of the Royal Statistical Society, Series C (Applied Statistics)
Volume68
Issue number5
Pages (from-to)1351-1370
ISSN0035-9254
DOIs
Publication statusPublished - 13. Jun 2019

Fingerprint

Compositional Data
Mortality
Forecast
Forecasting
Data analysis
Cancer
Mortality Rate
Alternatives
Model
Cause of death
Mortality forecasting

Cite this

@article{e05bcba892644d2188fd6a1f70559d39,
title = "Forecasting causes of death by using compositional data analysis: the case of cancer deaths",
abstract = "Cause‐specific mortality forecasting is often based on predicting cause‐specific death rates independently. Only a few methods have been suggested that incorporate dependence between causes. An attractive alternative is to model and forecast cause‐specific death distributions, rather than mortality rates, as dependence between the causes can be incorporated directly. We follow this idea and propose two new models which extend the current research on mortality forecasting using death distributions. We find that adding age, time and cause‐specific weights and decomposing both joint and individual variation between different causes of death increased the forecast accuracy of cancer deaths by using data for French and Dutch populations.",
author = "S{\o}ren Kj{\ae}rgaard and Ergemen, {Yunus Emre} and Malene Kallestrup-Lamb and James Oeppen and Rune Lindahl-Jacobsen",
year = "2019",
month = "6",
day = "13",
doi = "10.1111/rssc.12357",
language = "English",
volume = "68",
pages = "1351--1370",
journal = "Journal of the Royal Statistical Society, Series C (Applied Statistics)",
issn = "0035-9254",
publisher = "Wiley",
number = "5",

}

Forecasting causes of death by using compositional data analysis: the case of cancer deaths. / Kjærgaard, Søren; Ergemen, Yunus Emre; Kallestrup-Lamb, Malene; Oeppen, James; Lindahl-Jacobsen, Rune.

In: Journal of the Royal Statistical Society, Series C (Applied Statistics), Vol. 68, No. 5, 13.06.2019, p. 1351-1370.

Research output: Contribution to journalJournal articleResearchpeer-review

TY - JOUR

T1 - Forecasting causes of death by using compositional data analysis: the case of cancer deaths

AU - Kjærgaard, Søren

AU - Ergemen, Yunus Emre

AU - Kallestrup-Lamb, Malene

AU - Oeppen, James

AU - Lindahl-Jacobsen, Rune

PY - 2019/6/13

Y1 - 2019/6/13

N2 - Cause‐specific mortality forecasting is often based on predicting cause‐specific death rates independently. Only a few methods have been suggested that incorporate dependence between causes. An attractive alternative is to model and forecast cause‐specific death distributions, rather than mortality rates, as dependence between the causes can be incorporated directly. We follow this idea and propose two new models which extend the current research on mortality forecasting using death distributions. We find that adding age, time and cause‐specific weights and decomposing both joint and individual variation between different causes of death increased the forecast accuracy of cancer deaths by using data for French and Dutch populations.

AB - Cause‐specific mortality forecasting is often based on predicting cause‐specific death rates independently. Only a few methods have been suggested that incorporate dependence between causes. An attractive alternative is to model and forecast cause‐specific death distributions, rather than mortality rates, as dependence between the causes can be incorporated directly. We follow this idea and propose two new models which extend the current research on mortality forecasting using death distributions. We find that adding age, time and cause‐specific weights and decomposing both joint and individual variation between different causes of death increased the forecast accuracy of cancer deaths by using data for French and Dutch populations.

U2 - 10.1111/rssc.12357

DO - 10.1111/rssc.12357

M3 - Journal article

VL - 68

SP - 1351

EP - 1370

JO - Journal of the Royal Statistical Society, Series C (Applied Statistics)

JF - Journal of the Royal Statistical Society, Series C (Applied Statistics)

SN - 0035-9254

IS - 5

ER -