Improvement of Task Pattern Generation in Multi-task Call Centers

Staff scheduling problem in call centers is very important, since effective staffing can improve work efficiency, reduce costs, and provide sufficient service for customers. In this research, we deal with the daily shift scheduling (rostering) problem in call centers with multiple tasks. It is known that the set covering model can be applied to obtain the optimal solution for this kind of problems. However, the size of the model increases explosively as the numbers of staff, shifts, and tasks increase, because it requires to enumerate all combinations (patterns) of shifts and tasks. Hence we need some methods to reduce the patterns. In a previous study, we proposed a heuristic algorithm to obtain sub-optimal solution by automatically generating necessary task patterns. But there are still some timeslots where understaffing and overstaffing occur simultaneously. Therefore in this study we aim to investigate whether task patterns can be further improved and to reduce the under-staffing through analysis of understaffing and overstaffing performance.