Abstract :
The paper has two major goals: to define a staff scheduling problem for a heterogeneous workforce with many realistic constraints extracted from the real world, and to investigate its solution using a customized genetic algorithm featuring a group of operators which combine stochastic behavior and heuristics. After formalizing the problem, schedules for the whole workforce are represented by integer chromosomes of fixed dimension. Violations of constraints and problem requirements are reflected by cost increases, and the operators act stochastically but tend to decrease such costs. Although the operators interact with each other, they were designed in an independent way for the sake of simplicity and modularity. Overall, the action of these stochastic heuristic operators resembles a sophisticated mutation operator biased to improve schedules by reducing the costs of constraint violations. Experiments show that high quality workforce schedules can be obtained in reasonable time even for large problems
