Seminars 2008 — Abstracts
Friday, November 21
Speaker:
Andy Tremayne,
Melbourne
Title:
Some Alternative Models for Time Series Of Count Data
Abstract: The paper surveys some regression
and time series models suitable for the analysis of time series of
count data. In particular, the parameter- driven extension to the
static Poisson regression model proposed by Zeger (1988) is considered
as it is frequently used in empirical applications. Moreover, the
recently introduced parameter-driven autoregressive conditional Poisson
(ACP) model class, as well as integer-valued autoregressive (INAR)
models, are discussed in a comparative manner. The models, including
corresponding diagnostic tests, are illustrated using the famous
U.S. polio incidence count data set.
