The exam timetable problem with examination conflicts

In most of universities around the world, examinations are one of the most important ways adopted by the teachers to evaluate the performance of their teaching and the students always have several different exams at the end of each semester. After the exams have been assigned to the timeslots during the examination periods, examination conflicts appear frequently in practice. Therefore, this work studies a variant of the exam timetable problem, where a small number of examination conflicts are allowed. To solve the problem efficiently, an improved genetic algorithm is proposed for the problem and its performance is further validated by the instances in the real cases. The numerical results show that the improved algorithm demonstrates superior performance in problem-solving compared to the results generated manually and those by the traditional genetic algorithm. Furthermore, we find that allowing a few conflicts is able to reduce the frequency of students participating in consecutive and spaced exams.