Advanced

How many will be suffered by Cancer in 2020? Forecasting with Lee-Carter model with covariates

Ando, Takashi LU (2010) EKHR02 20101
Department of Economic History
Abstract
By using Lee-Carter model by sex and the origin of tumor we forecasted the cancer incidence. The usefulness of Lee-Carter methods to the incidence rates was shown. For forecasting the future the time trend parameter of Lee-Carter model behaves very close to Random-walk but it is not exactly Random-walk for most cases. It is better to leave the option open to AR(1) and not restrict ourselves to Random-walk. Another alternative would be including the exogenous information into the model if the information is available. This gives the model more precision and the ability to respond the changes in the information. Proportion of smokers did not have a huge impact on incidence rate while GDP seems to consistently affect the incidence rate. The... (More)
By using Lee-Carter model by sex and the origin of tumor we forecasted the cancer incidence. The usefulness of Lee-Carter methods to the incidence rates was shown. For forecasting the future the time trend parameter of Lee-Carter model behaves very close to Random-walk but it is not exactly Random-walk for most cases. It is better to leave the option open to AR(1) and not restrict ourselves to Random-walk. Another alternative would be including the exogenous information into the model if the information is available. This gives the model more precision and the ability to respond the changes in the information. Proportion of smokers did not have a huge impact on incidence rate while GDP seems to consistently affect the incidence rate. The author’s conclusion is that the model which included GDP would be the first choice followed by AR(1) or Random-walk model. Smoking (proportion of smokers in the society) should be considered on individual case basis and there might be better scale to measure smoking other than the proportion of smokers. (Less)
Please use this url to cite or link to this publication:
author
Ando, Takashi LU
supervisor
organization
course
EKHR02 20101
year
type
H2 - Master's Degree (Two Years)
subject
keywords
Cancer incidence, Lee-Carter, GDP, smoking
language
English
id
1612548
date added to LUP
2010-06-17 14:09:16
date last changed
2010-06-17 14:09:16
@misc{1612548,
  abstract     = {By using Lee-Carter model by sex and the origin of tumor we forecasted the cancer incidence. The usefulness of Lee-Carter methods to the incidence rates was shown. For forecasting the future the time trend parameter of Lee-Carter model behaves very close to Random-walk but it is not exactly Random-walk for most cases. It is better to leave the option open to AR(1) and not restrict ourselves to Random-walk. Another alternative would be including the exogenous information into the model if the information is available. This gives the model more precision and the ability to respond the changes in the information. Proportion of smokers did not have a huge impact on incidence rate while GDP seems to consistently affect the incidence rate. The author’s conclusion is that the model which included GDP would be the first choice followed by AR(1) or Random-walk model. Smoking (proportion of smokers in the society) should be considered on individual case basis and there might be better scale to measure smoking other than the proportion of smokers.},
  author       = {Ando, Takashi},
  keyword      = {Cancer incidence,Lee-Carter,GDP,smoking},
  language     = {eng},
  note         = {Student Paper},
  title        = {How many will be suffered by Cancer in 2020? Forecasting with Lee-Carter model with covariates},
  year         = {2010},
}