Class schedule generation using genetic algorithms

Abstract

The class scheduling problem is an NP-hard problem and belongs to the multi-objective optimization problem (MOP) that concentrates on establishing an optimum allocation of subjects onto a limited accessible number of slots and spaces. It is a challenging issue faced by universities across the world. Every academic institution has a dilemma while planning courses and test plans. There are several limits imposed when setting timetabling. Multi-objective evolutionary algorithms (MOEAs) are one of the methods to solve the MOP, including the class scheduling problem above. Besides, evolutionary algorithms (EAs) have four main classes such as genetic algorithms (GAs), evolution strategies (ESs), differential evolution (DE), and estimation of distribution algorithms (EDAs). Especially, GA is no longer weird after it has been employed a lot in the issue of constructing schedules. But there are a few disadvantages to executing genetic algorithms. GA may be difficult to debug and can be computationally costly. Additionally, GA may be sensitive to the starting circumstances and occasionally converge to local optima. That is why various alternative algorithms were invented based on GA to tackle those restrictions such as the Non-Dominated Sorting Genetic Algorithm II (NSGAII). In this research, the class schedule made by MOEAs displays the class schedule as well as the use of MOEAs in practice. Then, make comparisons between MOEAs to observe the difference in the approaches of each algorithm.