A New Forecasting Approach for Generating Stochastic Knowledge from Past Request Information and Utilizing the Stochastic Knowledge

In this chapter, a new forecasting approach is presented in order to extend the deterministic real-time control approach to a pro-active real-time control approach. The proposed forecasting approach uses a new segment-based clustering approach which makes it possible to generate stochastic knowledge out of the available past request information. We elaborate on how our forecasting approach generates and subsequently integrates stochastic knowledge into the deterministic real-time control approach thereby creating the proposed pro-active real-time control approach. Several quality criteria are introduced in order to ensure that the generated stochastic knowledge can be advantageously utilized. Furthermore, the handling of the generated stochastic knowledge in the real-time control approach during the execution of the transportation process is described. This includes how relevant parameters of dummy customers are updated during the execution of the transportation process and how the decision about when to remove dummy customers from further consideration is made.