Date of Award

May 2023

Degree Type

Dissertation

Degree Name

Doctor of Philosophy

Department

Mathematics

First Advisor

Richard Stockbridge

Committee Members

Chao Zhu, David Spade, Jeb Willenbring, Gabriella Pinter

Abstract

In this dissertation, possible methods for multiple change point detection on Markovchain processes are studied. Related works for oine and online change point detection are discussed and their applicability on sequential multiple change point detection for several regimes is evaluated. We develop a method for a multiple change point detection for a process having three regimes. Its eciency is then evaluated on simulated Markov chain data by looking into dierent scenarios such as processes that signicantly dier between each other or probability distributions that are slightly similar. This approach is then applied on Covid- 19 hospital data. Therefore, the data is modeled into three dierent Markov chain processes and then used to successfully apply the derived change point detection method. In the end, the possible enhancements and its applications in other real world examples are discussed.

Share

COinS