Potenziale und Herausforderungen von Machine Learning im Bereich Efficient Consumer Response (ECR) und insbesondere Collaborative Planning, Forecasting and Replenishment (CPFR) in der Milchindustrie

Abstract

The dairy industry is facing a variety of challenges characterised by volatile markets, rising consumer demands and increasing competitive pressure. In view of the prevailing dynamics, it is necessary for companies to optimise their supply chain strategies in a continuous process. Concepts such as Efficient Consumer Response (ECR) and Collaborative Planning, Forecasting, and Replenishment (CPFR) offer promising approaches for increasing efficiency and competitiveness. The integration of machine learning into these concepts enables companies to make more precise forecasts, make better decisions and utilise resources more efficiently. Nevertheless, there are considerable uncertainties regarding the practical application of machine learning in the dairy industry. This thesis is divided into a theoretical and an empirical part. In the theoretical part, the basics of ECR and CPFR as well as the specific challenges and potentials of the dairy industry are analysed in detail. Particular attention is paid to the role of machine learning, which could be used in planning and forecasting in the future. In the empirical part, a qualitative research method in the form of expert interviews was used. The aim of the interviews was to gain practical insights into the implementation and use of machine learning in the context of CPFR. The combination of these methods allows a comprehensive analysis and the formulation of practical recommendations. The results of the master's thesis show that the implementation of machine learning in the CPFR process in the dairy industry brings with it both considerable potential and various challenges. On the one hand, the use of machine learning can improve the accuracy of demand forecasts, better incorporate seasonal and weather-related fluctuations and thus increase efficiency in the supply chain. In addition, the use of machine learning allows processes to be optimised, resulting in long-term cost savings. On the other hand, significant challenges have also been identified, including ensuring data quality, high implementation costs, technical infrastructure requirements and guaranteeing the acceptance of the workforce. To summarise, careful consideration of costs and benefits and careful selection of the use case are crucial to the successful use of machine learning in CPFR processes. The analysis shows that machine learning is not the optimal solution in all scenarios and that traditional static methods can still offer a valid and cost-efficient alternative depending on the context.

Date of Award	2024
Original language	German (Austria)
Supervisor	Matthias Winter (Supervisor)