Back to search

University of the Aegean

Mathematical modelling of categorical data with actuarial and financial applications

Abstract

dc:description

The main focus of this thesis is on model−based inferential statistics for contingency tables. The leading role in the search for an optimal modelling approach is played by divergence measures. Divergence measures can be used in statistical inference for estimating purposes, in the construction of test statistics for test−of−fit or in statistical modeling for the construction of model selection criteria. At the same time, in most scientific fields, the interest lies in statistical analysis of multivariate categorical data. A standard way for testing purposes is via the traditional Pearson’s chi−squared goodness−of−fit test statistic or by the likelihood ratio test statistic, where the unknown parameters are estimated through the maximum likelihood method. However, this procedure can lead to poor results in many circumstances and it is possible to acquire better results by considering general families of test statistics, as well as of general families of estimators. Under the set−up of contaminated data, regarding the estimation procedure, these general families of divergence measures impose a compromise between robustness and efficiency. Furthermore, in the situation where zero frequency cells occur, a modified version of these general families can lead to robust estimators that are more efficient than the classical ones. On the other hand, under the contaminated data regime, such general families can produce more stable test statistics regarding the size and the power of the test. In this thesis, we study and exploit the general (Φ, α)−power divergence family and prominent sub−families of it, like the BHHJ family of measures. Rigorous asymptotic results, regarding the estimators and the associated test statistics, are provided while their applicability is presented through extensive simulation studies. Note that, regarding the aforementioned methodology, there is a straightforward connection with the actuarial science, mainly associated with the collective risk model of general insurance businesses and the modelling of frequency and severity of claims, and in general with financial mathematics.

Degree

thesis:*
Grantor dc:publisher
University of the Aegean
Year dc:date
2022

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Meselidis, Christos
  • Μεσελίδης, Χρήστος

Subjects

dc:subject × 24

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/52497

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Meselidis, Christos; Μεσελίδης, Χρήστος. Mathematical modelling of categorical data with actuarial and financial applications. University of the Aegean, 2022. http://hdl.handle.net/10442/hedi/52497