Data-driven logistics : improving the decision-making process in operational planning by integrating a supervised learning model

This research introduces a supervised learning modelin the existing planning software of a large logistic service provider. The goal of this research is to decrease the total replanning time by the human planners. Our solution design compares four learning models on obtained replanning data of a two-month period. After parameter tuning and evaluation, the Random Forest model is implemented as an Artifact to calculate its impact on the planning performance. This resulted in a decrease of replanning time of 30.6%. Therefore, this thesis substantiates the use of an artificial intelligence algorithm in a practical environment.