Timetabling Problems and the Effort Toward Generic Algorithms: A Comprehensive Survey

Abstract

The timetabling problem, a well-known NP-Hard optimization challenge, spans multiple domains such as education, healthcare, sports, and transportation. Due to its computational complexity, heuristic methods have become the dominant method for solving these problems. However, a lack of consistent comparisons across studies has led to difficulties in evaluating the effectiveness of these algorithms. This paper provides a comprehensive survey of 64 studies published between 2012 and 2022, focusing on timetabling algorithms and their performance on established benchmarks. The algorithms are categorized into metaheuristics, hybrid metaheuristics, and hyper-heuristics, with their efficacy evaluated across six major benchmarks. The analysis highlights the superiority of single-solution-based metaheuristics, with Simulated Annealing emerging as the most effective algorithm, particularly for educational timetabling problems. Additionally, the challenge of developing generic algorithms capable of performing across different timetabling problem domains is addressed. Despite advances in the field, cross-domain adaptability remains a critical area for further exploration. This survey serves as a guide for future research, providing insights into algorithmic strategies that can enhance the efficiency and generalizability of solutions for timetabling problems.