Date of Award

2021-12-01

Degree Name

Master of Science

Department

Computer Science

Advisor(s)

Vladik Kreinovich

Abstract

In many real-life situations, we need to make decisions in situations when we do not have full information about the consequences of different decisions. In particular, instead of the exact values of the relevant quantities, we only know lower and upper bounds on these values – i.e., we know an interval that contains the actual (unknown) value. These interval estimates often come from experts. This fact naturally leads to the following important questions: How should we make decisions under such interval uncertainty? How to gauge the quality of the resulting decisions? And if this quality is not sufficient – because the original intervals were too wide – how can we improve the interval estimates so as to make better decisions? And if improvements are possible, why not do them from the very beginning, as a pre-processing of expert-provided intervals? In this thesis, we propose answers to these questions in several economically meaningful situations. We start, in Chapter 1, with a general description of how rational decisions should be made – according to decision theory. To make these decisions, we need to have some information about the corresponding quantities, information that often comes in terms of expert-provided intervals. In Chapter 2, we analyze how these intervals can be improved. In Chapter 3, we analyze how we can take interval uncertainty into account when gauging the quality of the existing decisions. Finally, in Chapter 4, we analyze how to make new decisions under interval uncertainty.

Language

en

Provenance

Recieved from ProQuest

File Size

67 p.

File Format

application/pdf

Rights Holder

Laura Adriana Berrout-Ramos

Share

COinS