In a recent paper a new approach to forecasting based on the Bayesian principles of information theory was proposed and called the Poisson-gamma single-state model. In this paper a two-state version of the Poisson-gamma model is formulated by considering the uncertainty not only in the parameters but also in the model itself. This model is particularly useful for modelling epidemic data such as measles by considering two different situations (states) of the generating process at each time point: viz, state 1—no epidemic phase; state 2—epidemic phase.
Get full access to this article
View all access options for this article.
References
1.
CliffA DHaggettPOrdJ KBassetKDaviesD, 1975Elements of Spatial Structure (Cambridge University Press, Cambridge)
2.
CliffA DOrdJ K, 1978“Forecasting the progress of an epidemic” in Towards the Dynamic Analysis of Spatial Systems Eds MartinR LThriftN JBennettR J (Ron, London) pp 191–204
3.
JaynesE T, 1957Information Theory and Statistical Mechanics Physical Review106(4) 620–630
4.
ShannonC EWeaverW, 1949The Mathematical Theory of Communications (University of Illinois Press, Chicago, Ill.)
5.
SouzaR C, 1978A Bayesian Entropy Approach to ForecastingPhD thesis, Department of Statistics, University of Warwick, Coventry, England
6.
SouzaR CHarrisonP J, 1979“Steady state system forecasting: A Bayesian-entropy approach” internal report, Department of Statistics, University of Warwick, Coventry, England
