DEVELOPMENT OF AN EXAMINATION TIMETABLING SYSTEM USING GENETIC ALGORITHM

Abstract

The aim of this study is to apply genetic algorithm to the optimization of the timetable scheduling problem for the examination timetable based on the size of the registered students, the number of courses and the carryover courses of the students. This was achieved by eliciting knowledge on identifying the requirements of the system, formulating the genetic algorithm model and the simulation of the genetic algorithm model. A structured interview was conducted with the timetable representative in order to obtain quantitative data for the purpose of testing the system and the data collected consisted of sets of school configurations like the number of examination halls, the capacity of each hall, the number of timeslots per day etc. Hard and soft constraint of the genetic algorithm were formulated based on the limitations and feedbacks of timetable representative. Anaconda Navigator was used as an Integrated Development Environment (IDE). The system was simulated using comma separated values (CSV) file which served as a storage capacity for all the quantitative data gotten. The program was written using the Jupyter notebook with python being the interpreter. After representing the different constraints using mathematical modelling and the simulation of the GA using the python interpreter, the result has shown that the system is capable of providing useful solutions. It does not, however, fully automate the process. There are still some circumstances when the operator will need to make changes to some of the entries in order to achieve a flawless result. The enormous number of possible combinations for testing in order to arrive at an appropriate assessment for the application has proven to be impossible. However, considering the number of constraints set on the system, it can be inferred that the system was able to provide findings that, despite being imperfect, are valid and acceptable. In conclusion, the system was developed to improve timetabling scheduling in general, this study suggested that the approach/technique to tackling the problem of genetic algorithms should be used. Although the experimental results show that a more efficient and dependable schedule can be reached with a properly constructed genetic algorithm. This offers good review schedules without conflicting examinations and in a much quicker time.