Presented by: 
Alan Huang (UQ)
Mon 11 Sep, 2:00 pm - 3:00 pm

Poisson models for counts make the (very) strong assumption that the conditional variance is equal to the conditional mean. In this talk, we look at a recently-proposed extension of the Conway-Maxwell-Poisson distribution that can handle both overdispersed and underdispersed counts. The exponential family form of these distributions leads to particularly simple estimation and inference procedures from both a frequentist and Bayesian point of view.


Huang, A (2017) “Mean-parametrized Conway-Maxwell-Poisson models for dispersed counts”, Statistical Modelling, 17, 1—22

Huang, A. and Kim, S.I. (2017) “Bayesian inferences for dispersed counts”, in prep.