Long term planning of skill set assignment

Capacity planning is a hard problem these days. Employees with their capabilities must be assigned to the needed demand. In this thesis, a call center is taken as example. Currently an inefficient heuristic is running to determine the needed amount of agents to fulfill the demand. The needed demand is based on historical data and the service level the call center wants to reach. Based on this heuristic, a model is developed to get an overview of the long-term planning skill set assignment problem. According to this model several heuristics are developed. These heuristics are compared to the current heuristic. For small problems, we can also compare the solution and computation times to the optimal solution determined by the model solver. In the heuristics the employees or skills are sorted according to a rule that is different for every heuristic. The company wants the heuristic to be very fast and as good as possible. The planning in the application is made for all 78 weeks ahead, with a rolling horizon. All weeks are independent of each other, so if there is demand left at the end of the week, this demand is not taken into account for the next week. In this thesis an extension is made to handle this left demand. We call this inventory demand. E-mail is an example of inventory demand, while we can not store calls to handle them next week. To test the heuristics, data is simulated with different settings. Simulations with a different amount of agents, skills and weeks, both random and real case data are generated. All heuristics perform much better than the current heuristic both in quality and computation time. The best heuristic is offered to the organization to be implemented in the current application.