- Defining the Problem:
The National University of Singapore (NUS) is a prominent higher education institution in Singapore, currently enrolling more than 36,000 undergraduate and postgraduate students from all over the world.- NUS changed to a modular academic course structure in 1993, and the enhanced flexibility resulted in increasing difficulties when scheduling exams, with exam dates often overlapping across faculties. The inefficient scheduling system soon translated into longer examination periods and augmented room rental costs.
2. Developing a Model:
NUS developed a new centralized system, called University Timetable Scheduler Software Exam (UTTSExam),
in 1999 to address its problem. UTTSExam was an application of the constraint satisfaction optimization
problem (CSOP), an LP model that combines two different problems—one of constraint satisfaction and another of optimization.
3. Acquiring Input Data:
Data and constraints were derived directly from the university system. Essential information included student numbers, the time and length of examinations, and venue booking details.
4. Developing a Solution:
NUS had previously adopted both centralized and decentralized systems for exam timetabling, but both
systems had severe limitations. UTTSExam represented a hybrid approach, combining the best of the two
systems, allowing for a flexible but centrally managed solution that offered substantial improvements.
5. Testing the Solution:
UTTSExam was implemented in the first semester of the 2001–2002 school year. A total of 27,235 students were involved, each student sitting at least one of a total of 1,350 examinations. Eleven venues were
used, with a total of 4,654 available seats.
6. Analyzing the Results:
The gains obtained using the new system were clealy evident. In the original system, the exam scheduling
procedure was cumbersome and difficult to modify. With UTTSExam, once data and the constraint set
were defined, the scheduling procedure took less than five minutes. Late changes could also be
accommodated easily, and the system was virtually conflict free.
7. Implementing the Results:
When UTTSExam was fully implemented, it solved the problems of allocating a large set of examinations
over a short period and ensured that this was done in the best possible configuration for students. It
meant that there could be three sessions of examinations a day rather than two sessions. This reduced the
overall examination period from one month to two weeks and enabled large savings.
3 Comments
This post really helped me understand the value of queuing theory in retail. Thanks for the practical examples!
ReplyDelete"The real-world examples made this so much more relatable!"
ReplyDeleteThat’s the goal—appreciate your feedback!
Delete