The digital transformation is increasingly determining everyday professional life. The increasing digitization of operational processes is taking place across all industries. The implementation of data digitization and company networking offers the possibility of optimizing operational processes using this technology. Technical implementation can increase the operational success of a company if it results in better decisions of the controlling and management. Such data-driven decision-making processes are the core of this thesis. First of all, this process includes the transformation of company data into decision-relevant information. This doctoral dissertation demonstrates that this is possible by using machine learning to develop prediction models to provide reliable predictions about future customer behavior. Predictions that allow a validated division of customers into customer groups enable target group-oriented customer relationship management. For this purpose, the explanatory variables were identified in the prediction models which have a significant influence on the target variable set for the respective model. In the next step, it could be proven that the consideration of the forecast models considerably improves the decisions of the controlling. This has set the basis for the development of stochastic decision models based on data science in this thesis. The marina industry in Croatia offered a suitable scenario as there is intense national and international competition in this industry. Decisions on berth allocation, improved predictability of expected revenues and better knowledge of who the company's customers really are will significantly increase the company's competitiveness.