Variable neighborhood descent search based algorithms for course timetabling problem: Application to a Tunisian University

University course timetabling problem refers to schedule a set of lectures, tutorials and practical works to a limited number of teachers, classrooms and time slots over a planning horizon while satisfying a set of hard and soft constraints. In this paper, we investigate Variable Neighborhood Descent approach to tackle a specific course timetabling problem related to the case of the Faculty of Economics and Management Sciences of Sfax in Tunisia. The objective is to minimize the total number of holes and the number of isolated lessons for all student groups. We have developed eleven specific neighborhood structures: six of these are designed for correcting the holes, while the other five are designed for adjusting the isolated lessons. Computations are made on real instances and numerical results show that our approach outperforms the existing solution with the elimination of 52.47 % of holes and isolated sessions on average.