Self Adaptive and Simulated Annealing Hyper-Heuristics Approach for Post-Enrollment Course Timetabling

Abstract

One of timetabling problem in education field is Post-Enrollment Course Timetabling (PE-CTT). The challenges faced in the PE-CTT are differences in type of problems, a number of limitations, and requirements that differ from one university to another. It is difficult to find common and effective solutions. State-of-art method that can develop more general systems by using cheaper methods and still being able to solve problems is Hyper-Heuristic approach. The Self-Adaptive Strategy is used as a strategy for selecting Low-Level-Heuristics (LLH) and Simulated Annealing as a Move Acceptance (MA) strategy to solve the course timetabling problems. Self-adaptive can increase the level of convergence to the optimal value in the optimization process. Simulated Annealing can accept solutions that are no better so that the final solution is not trapped in the local optima solution. The contributions of this paper are state-of-art hybridization of Self-Adaptive and Simulated Annealing Hyper-Heuristic approach to solve post-enrollment course timetabling problem. The Self Adaptive and Simulated Annealing Hyper-Heuristics then compared with Simple Random and Simulated Annealing Hyper-Heuristics. Tests will be carried out on Socha dataset. The hybridization of Self Adaptive and Simulated Annealing shows competitive results.

Optimization in University Student Timetables: A Comprehensive Literature Review

“…[28] General framework with four layers model with Layer 1 to Layer 3 is called Stemming Phase involving preprocessing steps, and the Layer 4 is called Solution Finding Phase with enhanced simulated annealing-based search [29] Model leveraged by artificial bee colonies (ABC), cloud theory-based simulated annealing (CTB-SA), and genetic algorithm (GA) [18] Two phase hybrid local search algorithm with Tabu Search with Sampling and Perturbation with Iterated Local Search (TSSP-ILS) and Simulated Annealing with Reheating (SAR) with two preliminary runs (SAR-2P) [30] Two-stage heuristic algorithm consists of Lecturer Grouping Stage and Group Allocation Stage [33] Two hybrid variants of flower pollination algorithm (FPA) which were Jaccard FPA (JFPA) which uses the Jaccard index and a greedy selection mechanism, and Dragonfly FPA (DFPA) which incorporates the navigational traits of the dragonfly algorithm (DA) [34] Hybridization of Self-Adaptive and Simulated Annealing Hyper-Heuristic approach [35] Sequential constructive algorithm and Fuzzy Logic [36] Multi-Agent System (MAS) incorporating Integer Programming (IP)…”

“…Kartika and Ahmad [35] presented a hybridization of the Self-Adaptive and Simulated Annealing Hyper-Heuristic approach to tackle Post-Enrollment Course Timetabling (PE-CTT). The approach involved employing a Self-Adaptive Strategy to select Low-Level-Heuristics (LLH) and using Simulated Annealing as a Move Acceptance (MA) strategy to enhance optimization for solving PE-CTT problems.…”

“…Reference Dataset description [11] UCSI University, Malaysia [12] Eleven datasets obtained from Naresuan University, Thailand [13] Semester 1 2017/2018, Universiti Malaysia Terengganu (UMT), Malaysia [14] Semester 1 2018/2019, Universiti Malaysia Sabah Labuan International Campus (UMSLIC), Malaysia [15] Semester 2016/s6 and semester 2016/s2, i-Cats University College (ICATS), Malaysia [16] ITC 2019 [17] ITC 2007 Semester 1 2019/2020 and semester 2 2019/2020, Faculty of Computer Science and Information Technology, Universiti Malaysia Sarawak (UNIMAS), Malaysia [18] Lecturers-courses assignment simulation [19] Spring semester of 2020 for Computing Fundamental Department, FPT University, Vietnam [20] Semester 1 2019/2020 and semester 2 2019/2020, Faculty of Computer Science and Information Technology, Universiti Malaysia Sarawak (UNIMAS), Malaysia [21] Faculty of Information Technology, Nong Lam University (NLU), Vietnam [22] Even semester 2018/2019, Informatics Engineering Study Program of Universitas Catur Insan Cendekia Cirebon, Indonesia [23] ITC 2019 [24] Generate 4 datasets based on assumptions [25] ITC 2002 and ITC 2007 [26] Department of Mathematics XYZ University [27] Semester 1 2018/2019, semester 2 2017/2018, semester 1 2017/2018, semester 2 2016/2017, University of the East, Philippines [28] Eleven datasets obtained from Naresuan University, Thailand [29] Socha STKIP (a medium-size Indonesian college from the year 2014 to the year 2016) Lewis Carter ITC 2007 [30] ITC 2002 ITC 2007 Socha Hard [31] Semester 2 2014/2015 and semester 1 2015/2016, Universiti Malaysia Sabah Labuan International Campus (UMSLIC), Malaysia [32] Odd and Event semester, Operational Division of Universitas Teknologi Yogyakarta, Indonesia [33] Semester 1 2017/2018 and semester 2 2017/2018 from Faculty of Cognitive Sciences and Human Development, Universiti Malaysia Sarawak (UNIMAS), Malaysia [35] Socha [36] Semester 1 2016/2017, Universiti Malaysia Sabah Labuan International Campus (UMSLIC), Malaysia [37] Semester 2 2014/2015 and semester 1 2015/2016, Universiti Malaysia Sabah Labuan International Campus (UMSLIC), Malaysia…”

Bibliometric Analysis of University Timetabling Using Publish and Perish

“…Several studies describe the wide variety of complexity. For example, the consequences of improper scheduling for students with different semesters can cause schedules to clash [14]. There are several choices of algorithmic approaches that can be applied to solve the university timetabling problem.…”

Automatic Design of Metaheuristics for Practical Engineering Applications

“…Likewise, Garza-Santisteban et al proposed a high-level solver based on SA for selecting the best heuristic sequence that fulfills the requirements of multiple instances of Job Shop Scheduling problems [23]. Moreover, several hybrid approaches incorporating SA-based processes are easily found in the literature [24], [25], [26]. For example, Mosadegh et al presented a hyper-heuristic based on Q-Learning and Simulated Annealing for dealing with the mixed-model sequencing problem, including stochastic processing times in a multi-station assembly line [27].…”