An evolutionary algorithm based subject allocation system

Abstract

In the School of Electrical and Electronic Engineering (EEE), Nanyang Technological University (NTU), allocation of subjects according to preferences indicated by students during registration is a complex optimization problem. We propose an evolutionary algorithm based system that attempts to maximize satisfaction for the overall student group. The system is able to improve the satisfaction level from 61% to 74% even though some of the classes may be heavily oversubscribed. A novel mutation scheme, known as Directed Window Mutation is used and it is shown to be effective. Results of testing on a sample student group have been positive. The system will greatly reduce the level of frustration and unhappiness of students during the subject registration. In addition, it will result in significant cost and time saving for the School's Administrative body. The feedback on this proposed system has been positive from both the students and administrators.

Key words:

• evolutionary algorithms
• subject allocation
• directed window mutation

Notes

Corresponding author. (Email: [email protected])