Fairness in nurse rostering problem
Student thesis: Doctoral Thesis
Many Operational Research (OR) problems like scheduling and timetabling, areassociated with evaluating the distribution of resources in a set of entities. This set of entities can be defined as a society having some common traits. The evaluation of the distribution is traditionally done with a utilitarian approach, or using some statistical methods. In order to gain a more in depth view of distributions in problem solving new measures and models from the fields of Computer Science, Economics, and Sociology, as well OR are proposed. These models focus on 3 concepts: fairness (minimisation of inequalities), social welfare (combination of fairness and efficiency) and poverty (starvation of resources). A Multiple Criteria Decision Making (MCDM) model, combining utilitarian, fairness and poverty measures is also proposed. These measures and models are applied to the nurse rostering problem from a central decision maker point of view. Nurses are treated as a society, trying to optimise nurse satisfaction. Nurse satisfaction is investigated independently from the hospital management, forming two conflicting criteria. The results from different measures cannot be evaluated using cardinal measures, so MCDM methods and Lorenz Curves are used instead of a numerical, cardinal measure.
1 MB, PDF document