Educational timetabling using computational intelligence algorithms

Educational timetabling is a fundamental part of the effective operation of schools and universities. Teaching quality, smoothness of school administration and operation cost-efficiency can all be affected negatively by problems with the timetabling. Educational timetabling problems (ETPs) can be categorised into three types: course timetabling problems, school timetabling problems and examination timetabling problems. Course timetabling allocates educational activities, such as lectures and tutorials, to timeslots, classrooms or other educational facilities, subject to the relevant constraints. School timetabling assigns school educators to a scheduled course timetable with the considerations of educators’ availabilities and specialisations. Examination timetabling prevents an individual student from taking more than one exam at the same time and optimises educational resource usage within an examination period. This thesis offers a model of educational timetabling using computational intelligence (CI) algorithms. A comprehensive literature review on CI algorithms applied to ETPs and their applications was conducted. Based on the literature review, two educational timetabling models were proposed. The online examination timetabling (OET) model targets examinational timetabling problem during the COVID-19 pandemic and the school timetabling problem (STP) model focuses on addressing the school educators’ allocations. The research proposed a conceptual model for the OET problem. A conflict table was introduced for constraint handling. A modified artificial bee colony (ABC) algorithm was applied to the OET model. The proposed approach was simulated with 16,246 exam items covering 9,366 students and 209 units and compared with the traditional constraint v programming method. The results suggest that the proposed approach can effectively provide more balanced solutions to the OET problem. An STP model considering school educators’ availabilities, preferences and expertise as a whole was also proposed. A virtual searching space (VSS) for dealing with a large solution pool was introduced. The proposed approach was simulated with a randomly generated large dataset, and the experimental results demonstrated that the proposed approach could solve the STP, and the VSS could significantly reduce the computational cost. The main contributions of this research, then, are that, first, a conceptual OET model was proposed to confront the challenge that the COVID-19 pandemic posed to ETPs, which is the first study in ETPs research field. A novel conflict table method was introduced for shrinking the solution pool. Second, an STP model respecting school educators’ preferences, availabilities and expertise was established. Generally, previous STP studies rarely considered educators’ preferences. The proposed STP model filled the gap. The VSS method enables regular computers to handle a large searching space. Third, in the comparison study, the experimental results indicated the ABC algorithm performed better compared to the constraint programming algorithm.