Proceedings of the International Conference on Industrial Engineering and Operations Management
Pilsen, Czech Republic, July 23-26, 2019
© IEOM Society International
Teaching Assistantship Assignment Optimization using
Hungarian Algorithm – A Case Study
Chandra Mouli R. Madhuranthakam
Chemical Engineering Department,
Abu Dhabi University
Abu Dhabi, United Arab Emirates P.O.Box. 59911
chandra.mouli@adu.ac.ae
Mukhtar Al-Ismaily
1
and Ali Elkamel
1,2
1- Chemical Engineering Department,
University of Waterloo,
Waterloo, Ontario, Canada
mukhtar.al-ismaily@uwaterloo.ca
aelkamel@uwaterloo.ca
2- Chemical Engineering Department,
Khalifa University of Science and Technology,
Abu Dhabi, United Arab Emirates, P.O.Box. 2533
ali.elkamel@ku.ac.ae
Abstract
Teaching assistantship (TA) assignment is typically a time-consuming and enduring process due to conflicting
constraints and multiple objectives. This project involves assigning TA assignments to a large pool of candidates
over a smaller set of offered courses, all the while ensuring that fairness is maintained during the assignment. A
program was developed to address and optimize based on the given constraints and objectives. It begins with a data
collection phase where all relevant information on the candidates and courses is acquired and then a software
program is used to optimize the TA assignment using the Hungarian algorithm. The model was initially formulated
as a multi-objective function consisting of three objectives viz., Course/Instructor Preference, Student Preference
and Income deviation among candidates. It was then scalarized into a single weighted-sum objective function. The
optimizer was tested on candidates applying for teaching assistantship towards three separate semesters, yielding an
average of 60-70% matching with the manual assignments that were previously carried out by the department. The
statistical dispersion in optimization cost and matching extent towards the manual assignment was observed under
the different set of weights. Higher assignment deviations would extend the Pareto domain, which also results in
more favorable optimization costs.
Keywords
Multiobjective optimization, Hungarian algorithm, Data analysis, Conflicting constraints.
1393