 
Summary: UNIVERSITY OF REGINA
Department of Mathematics and Statistics
Graduate Student Seminar
Speaker: Xiaojing Wu
Date: 31 August 2005
Time: 10.00 o'clock
Location: College West 307.18 (Math & Stats Lounge)
Title: Modelling Negative AR(1) Processes
Abstract: Modelling discrete time series data is one of the most challenging areas in the
time series study. In the first part of my work, I will briefly introduce several different
approaches to modelling discrete nonGaussian time series data. Then one of them, a
class of model that is based on thinning operation, will be discussed in more detail.
Geometric AR(1) and Poisson AR(1) will be mentioned about as two typical examples of
the modelling method based on thinning operations. In the second part of my work, two
different AR(1) models proposed to deal with Negative Binomial varieties will be discussed
in great detail. As the main part of the discussion of the Negative Binomial AR(1) models,
the derivation of their Moment Generation Function and Correlation Function will be
presented and their Joint Cumulantes will be calculated in this section. This is followed
by a brief discussion on the method which will be employed to estimate the parameters
of the two Negative Binomial AR(1) models. The estimation problem and the further
