Abstract
Nurse rostering is the process of creating a systematic working plan, showing working hours for the employees in the plan over a given planning horizon. Making rosters is an important activity in health care services. Today, highly qualified health personnel spend a lot of their time making and updating rosters manually. By using good, automatic tools for this work, a considerable amount of time will be freed for working with patients. An approach using CSP to create the initial solution and Iterated Local Search for optimization will be presented. Focal points are used to reduce the size of the neighborhoods. Both general and individual preferences are taken into consideration in addition to satisfying the Working Environment Act. The solution has been tested with good results on real world problems, and has been implemented in a commercial system.
