Off-campus Eastern Washington University users: To download EWU Only theses, please use your EWU SSO (Single Sign-on) credentials. Clicking the blue “Download” button below will prompt you to log in in order to access the thesis document.

Non-EWU users: Please talk to your local librarian about requesting this thesis through Interlibrary loan.

Date of Award

Summer 2026

Rights

Access perpetually restricted to EWU users with an active EWU NetID

Document Type

Thesis: EWU Only

Degree Name

Master of Science (MS) in Applied Mathematics

Department

Mathematics

First Advisor

Dr. Frank Lynch

Second Advisor

Dr. Andrew Oster

Third Advisor

Dr. Stuart Steiner

Abstract

How should a university judge the quality of a course schedule? The usual answer checks only that it can be run: no instructor in two places, no room double booked, every section staffed. A schedule can pass every one of those checks and still stop the students it serves: a needed course is not offered that term, every offered section is full, or every open section meets at the same hour as another required course. Seats and sections cost money to add, while rearranging what already runs (who teaches, in which room, at which hour) costs nothing. This thesis holds the published offering fixed and asks what that free rearrangement is worth.

A mixed-integer linear program produces candidate schedules, considering only the instructor, room, and hour assignments with precedent in twenty-four years of one university’s published schedules. Three of the objectives are an even spread of sections across the teaching day, filled seats in the rooms chosen, and fewer time conflicts among courses students need together. Those eligibility rules cut the number of decision variables by a factor of ~4400 for the whole university in fall 2025. That cut brings department-scale instances within reach of a single machine, and lets the same machine build and price college-scale instances. An agent-based simulation of the student cohort then walks each candidate schedule term by term and counts every failed registration against its cause: not offered, section full, or time conflict. Without adding any seats or sections, the model’s best produced schedule cut the median count of registrations blocked by time conflict by about a third, and by about a quarter at the other two solver seeds, when compared to the schedule that the university actually ran. Every input (published schedules, catalog pages, federal completion records, and pass rates from the published literature) is public, so any institution can repeat the measurement.

Download

Share

COinS