View article

A variety of neighbourhood operators have been used in local search and metaheuristic approaches to solving nurse rostering problems. We test and analyse the efficiency of these neighbourhoods on benchmark problems which are taken from real world scenarios. A variable depth search is then developed based on the results of this investigation. The algorithm heuristically chains together moves and swaps which define the more effective search neighbourhoods. A number of heuristics for creating these chains were developed and we present experiments conducted to identify the best ones. As end users vary in how long they are willing to wait for solutions, a particular goal of this research was to create an algorithm that accepts a user specified computational time limit and uses it effectively. When compared against previously published approaches the results show that the algorithm is very competitive.