Drawing up the nurses
duty roster is one of the main key issues that are faced by hospital managements,
this study focuses on Nurse Rostering Problem (NRP), an NP-hard problem, that
is difficult to solve for optimality. Harmony Search Algorithm (HSA) refers
to the meta-heuristic algorithm inspired by the improvisation of Jazz musicians.
Due to the problem of slow convergence of the basic HSA, this study attempted
to enhance basic HSA (called EHSA). This is done by using a semi cyclic shift
patterns in the initialization step to generate the initial harmonies (population)
rather than using a fully random mechanism in basic HSA. Furthermore, a dynamic
mechanism was employed in EHSA to update the parameter values of harmony memory
considering rate and pitch adjusting rate instead of fixed values in basic HSA.
A real world dataset from large hospital in Malaysia was used to evaluate the
performance of EHSA. Results showed that EHSA can produce high quality rosters
in shorter execution time compared to basic HSA. A comparison between EHSA and
Adaptive Harmony Search (AHS) is also presented to demonstrate the performance
of the proposed method. Better results have been obtained by EHSA compared to