Automated University Timetabling with Robustness

Timetabling and scheduling problems occur widely in both private and public sectors. (In)effective
schedules can have a significant effect on profits and the efficient use of resources along
with consumer and employee satisfaction.

Alongside the effectiveness of a given timetable/schedule, the robustness/fragility of
a deployed solution when coping with changes in demand/resource availability is often also an
important concern.

A prototypical timetabling problem is university course timetabling. With hundreds of rooms,
hundreds of modules, hundreds of staff and thousands of students, the search domain is far too
vast to be exhaustively enumerated to find an optimal solution satisfying the wide range of
constraints needing to be satisfied. Instead, heuristic search algorithms are typically employed by
a team of staff to develop the final timetable to be used.
As a university grows in size, and the choice of modules available to a student both within and
outside their department expands, so the complexity the timetabling task also increases. Beyond
this, developing a timetable which has flexibility to cope with staff illness, and deal with variation
in projected enrolments, complicates the task further.

This project will develop and apply timetabling algorithms for the university timetabling domain,
using e.g. hyper-heuristics, converting the constraints to be satisfied by a solution as criteria in a
many-objective optimisation problem, and developing criteria to effectively assess the robustness
of a solution.