Use this URL to cite or link to this record in EThOS:
Title: Certain properties of bivariate generalized poisson distributions with special reference to the hermite distribution
Author: Papageorgiou, Haralambos
ISNI:       0000 0001 3466 9544
Awarding Body: University of Bradford
Current Institution: University of Bradford
Date of Award: 1977
Availability of Full Text:
Access from EThOS:
Interest in discrete distributions has increased markedly over the past two decades or so. However, although a very large number of univariate discrete distributions have been extensively investigated, relatively few bivariate versions of these have been studied in any detail. In this thesis we introduce bivariate Hermite, Poisson binomial, and Poisson Pascal distributions and go on to investigate certain structural and estimation properties of bivariate generalized Poisson distributions. We introduce the notation and terminology which provides the basis for this thesis in Chapter 1. In Chapter 2, following an approach due to Subrahmaniam [1966] we examine conditionality in bivariate generalized Poisson distributions with special reference to the bivariate Neyman type A models. We consider in Chapter 3 bivariate versions of two univariate estimation procedures known as 'the method of zero frequency' and '.the method of even points'. We continue by applying these to the bivariate Poisson, Negative binomial, and Neyman A type I, type II, and type III distributions. Holgate [1966] describes three ways of constructing bivariate distributions with Neyman type A marginals. Proceeding on similar lines in Chapter 4, we introduce bivariate Poisson binomial and Poisson Pascal distributions and we study various properties including conditional distributions and regression functions. We examine in Chapter 5 parameter estimation for bivariate Poisson binomial distributions using zero frequencies, even points, and moments. In Chapter 6, following Kemp [1972], we define two bivariate Hermite distributions, one with five and one with eight parameters via various straightforward extensions of the models leading to the univariate Hermite distribution. We continue by deriving recurrence relations for the probabilities and the central moments, bound for the correlation coefficient, conditional distributions, regression functions, and limiting forms.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available