In the past years there has been considerable interest in count data models, particularly
in the actuarial literature. As mentioned in Cameron and Trivedi (1998), an important
milestone in the development of models for count data is reached by the emergence of
Generalized Linear Models (GLMs). The Poisson regression is a special case of GLMs that
was first developed by Nelder and Wedderburn (1972) and detailed later in the papers of
Gourieroux et al. (1984a, 1984b) and in the work on longitudinal or panel count data models
of Hausman et al. (1984). Within non-life insurance context, McCullagh and Nelder (1989)
demonstrate that the usage of the GLMs techniques, in order to estimate the frequency of
claims, has an a priori Poisson structure. Antonio et al. (2012) present the Poisson
distribution as the modelling archetype of claim frequency.