Maximum likelihood estimation of phase-type distributions

Luz Judith R Esparza

    Research output: Book/ReportPh.D. thesis

    3885 Downloads (Pure)

    Abstract

    This work is concerned with the statistical inference of phase-type distributions and the analysis of distributions with rational Laplace transform, known as matrix-exponential distributions. The thesis is focused on the estimation of the maximum likelihood parameters of phase-type distributions for both univariate and multivariate cases. Methods like the EM algorithm and Markov chain Monte Carlo are applied for this purpose. Furthermore, this thesis provides explicit formulae for computing the Fisher information matrix for discrete and continuous phase-type distributions, which is needed to find confidence regions for their estimated parameters. Finally, a new general class of distributions, called bilateral matrix-exponential distributions, is defined. These distributions have the entire real line as domain and can be used, for instance, for modelling. In addition, this class of distributions represents a generalization of the class of matrix-exponential distributions.
    Original languageEnglish
    Place of PublicationKgs. Lyngby, Denmark
    PublisherTechnical University of Denmark
    Publication statusPublished - 2011
    SeriesIMM-PHD-2010-245

    Fingerprint

    Dive into the research topics of 'Maximum likelihood estimation of phase-type distributions'. Together they form a unique fingerprint.

    Cite this