Parameter Adjustment Approach Based on Distribution of Schedules in the Past for Staff Scheduling Problems

摘要

In a workplace where different workers do the jobs in relays around all day and night, e.g. a convenience stores and a hospital, a schedule which is suitable assignments of duty for a staff in each period of time is indispensable. In general, so far, the staff schedule is determined manually by considering the constraints which includes the original ones for each workplace. However, it is the case that the constraints are not completely clear because some constraints are to be given on an empirical basis. In this paper, first, the staff scheduling problem is formally described in a rather generalized form to apply a wide range of work types, and is formulated as an integer programming model. In this formulation, so many parameters concerning the desires of the workers as well as the conditions on works are included. While it is ideal that such parameter setting could be done directly by a manager, the number of parameters are too many and also some of the parameters are originally unclear. Considering these situations, in this research, an approach for adjusting parameters interactively is dealt with, and a way of estimating parameters by the use of past schedules is newly designed. Through some computational examples, the validity of the proposed approach is confirmed and its potential for future applications is shown.