Solving an application of university course timetabling problem by using genetic algorithm
Generating timetables for academic institutions is a complex problem. This is due to many constraints involved whether they are vital or desirable, which are known as hard and soft constraints. The problem becomes more complicated and difficult to solve as the number of courses increase. Moreover, g...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English |
| Published: |
2022
|
| Subjects: | |
| Online Access: | https://etd.uum.edu.my/10137/1/s821004_01.pdf https://etd.uum.edu.my/10137/2/s821004_02.pdf https://etd.uum.edu.my/10137/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Generating timetables for academic institutions is a complex problem. This is due to many constraints involved whether they are vital or desirable, which are known as hard and soft constraints. The problem becomes more complicated and difficult to solve as the number of courses increase. Moreover, generating manual timetables is challenging and time-consuming, particularly to meet lecturers’ preferences. Thus, it is crucial to establish an automated course timetable system. Many efforts have been made using various computational heuristic methods to acquire the best solutions. Among the approaches, genetic algorithm (GA), constructed based on Darwin's theory of evolution, becomes the renowned approach to solve various types of timetabling problems. Therefore, this study produces the best timetable using GA to solve clashed courses, optimize room utilization and maximize lecturers’ preferences. Data of 41 course sections from 17 courses offered in semester A172 were taken from Decision Science Department, School of Quantitative Sciences (SQS). The phases in GA involves a number of main operators which are population initialization, crossover and mutation. The best parameter setting for GA was determined through combination of different mutation rate, population and iteration. The simulation results of GA show that this method is able to produce the best fitness value that satisfied all hard and soft constraints. There are no clashes either between lecturers or lecture rooms, and lecturers’ preferences were satisfied. The system can help SQS or any other academic schools or institutions to easily develop course timetabling in the coming semesters. |
|---|