In this talk, we introduce a class of models for time series of count data that allow a flexible treatment
of time dependence as well as nonstationarity, while preserving marginal properties.
The models are based on stochastic processes derived from the convolution of compound Poisson random variables
with a kernel function. In the stationary setting, kernel functions are defined by probability
density functions and capture the underlying dependence structure in a computationally efficient manner.
We further develop regression models for count data where the shape of the kernels
depends on covariate information to obtain rich dynamics. The parametric specification
of the kernels we propose strikes a balance between tractability and the ability to capture
complex dependence structures, and also allows parameter estimation and prediction
based on bivariate distributions. We illustrate the construction by modeling real
data in crime analysis and epidemiology.
On modelling of time series of count data with marginal compound distributions
Carlo Gaetan (Università Ca' Foscari - Venezia)
Lieu
à BioSP