About DSUM

Data Science for University Management (DSUM) is a Grand Solutions project funded by Innovation Fund Denmark and MaCom A/S. The project includes research and development of a decision supporting tool for university timetabling. The research is done by two PhD students, Rasmus Ørnstrup Mikkelsen and Dennis Søren Holm supervised by Assoc. Prof. Thomas Stidsen located at DTU Management. The research projects are related to Quality Timetable Recovery and Strategic University Timetabling

The timetable is the production plan of the university

The problem of timetabling is not solved with a full-automated timetable generator. Afterall, the timetable is the production plan of the university. If the management is not in control of the production plan the outcome of the company will be affected by it. Several times a year the production plan is made, most of this process can be (and should be) automated. Generating a feasible high-quality timetable is a very difficult task to do manually, therefore, it is urgent to have a state-of-the-art automated timetable algorithm. The algorithm itself is not enough, the input data needs to be of high quality to give the timetable algorithm the best working environment and the output timetable from the algorithm needs to be validated by a human. It is thus this symbiosis between human and computer that generates the best possible timetables.

The project focuses on the different operational levels that are associated with timetabling.

A possible situation on the operational level is that something has happened such that the current timetable can no longer be used. In this case it is most beneficial to provide multiple solutions to account for this disruption. With multiple solutions and the information about consequences of each solution, the human can make well-informed decisions to account for such disruptions.

On the strategic level we want to change the amount the resources such that it is possible to have a decent quality timetable. But the weight of the timetable quality contra the “cost” of a resource is not easy to understand for a algorithm unless you give very specific inputs. But very specific inputs also leads to very little information about alternatives. Therefore, it is better to provide the full spectrum of solutions in a quality contra resource manner (pareto-front). Again, the human is provided with several solutions for the problem and can make well-informed decision about future resources. That is why we aim to develop a decision supporting tool for university management.