A novel multivariate Poisson distribution: forecasting the number of cancer deaths in Iran

Forecasting the number of cancer deaths in the future years will be the primary basis for policy-making in the field of cancer prevention. The forecasting can be made using statistical models. Accordingly, this article introduces a new multivariate Poisson distribution. This distribution has Poisson marginals and a joint probability mass function that is easier to understand than previously known distributions. The article also discusses some statistical properties of the distribution. A bivariate integer-valued autoregressive model (BINAR(۱)) is constructed based on the bivariate version of the proposed distribution and the model is applied to forecast the number of deaths caused by cancer in Iran based on the number of deaths from ۲۰۱۱ to ۲۰۱۸.

Keywords:

Bivariate integer-valued autoregressive time series , Cancer , Forecasting , Multivariate distribution , Multivariate Poisson distribution.