Imperative Selection Intensity of Parent Selection Operator in Evolutionary Algorithm Hybridization for Nurse Scheduling Problem

Abstract

Flexibility of shifts assignment in real-time condition is complex because it must consider multiple important aspects such as nurses’ diverse requests and nurse ward coverage. In fact, creating nurses’ work schedule is a time-consuming task and the created schedule may not be effective due to considerable dependence of the process on the head nurse’s capability and working experiences. Thus, hospitals are becoming increasingly interested in the deployment of technology to solve the nurse scheduling problems. In the current research, three classifications of constraints namely the hard, semi-hard and soft constraints were implemented technically to refine undesirable work schedule in nurse scheduling. To deal with heavy constraints handling, this research implemented an enhancement of Evolutionary Algorithm with Discovery Rate Tournament parent selection operator (D_rT) to minimize constraints violations. Selection intensity resulted from hybridizing discovery rate of Cuckoo Search and tournament elements were used for exploration and exploitation. Correspondingly, three parent selections were tested, and D_rT parent selection was found to achieve the best accuracy which gives way to obtaining better quality schedule with lowest fitness value. In particular, the superiority of D_rT parent selection suggested that selecting elite parents and ensuring diverse characteristic of the selected parents in a population are especially useful in small-sized population.