Author

Sawyer Welden

Date of Award

2019

Document Type

Thesis

Degree Name

Bachelors

Department

Natural Sciences

First Advisor

Doucette, John

Area of Concentration

Computer Science

Abstract

Ordinal preferences indicate an agent's preferences over items in terms of alternative items, rather than on a quantitative scale. Modelling ordinal preferences is useful for the analysis of ranked choice voting for social choice and resource distribution as well as for the micro economic choice of consumer goods by an agent. This thesis uses a new methodology to compare the goodness of fit of preference models and applies it to the Mallows and Plackett-Luce model. Our findings show the Plackett-Luce model to have better goodness of fit in general, but the Mallows model to have a better score under certain information criterion. This thesis also proposes a new preference model using generative adversarial networks to create synthetic preference rankings given an arbitrary set of preferences. While this model was found to not output preferences, it may be useful with a more restrictive data representation.

Share

COinS