A Greedy Algorithm for the Nurse Rostering

Abstract

Abstract : The Nurse Rostering Problem (NRP) concerns the complex task of scheduling nurses’ shifts in hospitals, taking into account many constraints and restrictions. The challenge of manually creating these schedules is considerable due to the multitude of scenarios and variables involved. In this paper we deal with a special variant of the NRP, where nurses are allowed to perform extra-shifts in order to fulfill minimal staff demands. We propose two variants of a pseudo-greedy algorithm to tackle the problem, with an extra component of randomness. This component implies that different runs output different solutions, giving an iterative dimension to the classical greedy template. Computational experiments on artificially created instances indicates that the algorithm is a viable option to be used on hospitals, providing its low total running times and fast solution convergence