Algorithms for Scheduling Teaching Assistants - A comparison focused on preference satisfaction, workload fairness, and computational performance

Abstract

Efficient scheduling of Teaching Assistants is a complex combinatorial optimization problem faced by many universities, involving constraints such as course requirements, availability, working hours, and fairness. A survey was conducted among Teaching Assistants at the Department of Computer Science and Engineering at Chalmers University of Technology and the University of Gothenburg to determine schedule preferences and system requirements. These preferences and requirements were used to create soft constraints and hard constraints for the scheduling algorithms.

This thesis investigates four algorithms: a simple greedy algorithm; a simple genetic algorithm; Tabu Search; and Simulated Annealing. Results reveal that although Greedy Algorithm was the quickest, it performed poorly on constraint satisfaction. Both Tabu Search and Simulated Annealing found high-quality solutions, but since Tabu Search was more computationally efficient, it was concluded to be the superior algorithm.