The catering service Stattküche in Münster produces around 22,000 portions of lunch every day, providing them to schools and nurseries. A portion of pre-ordered meals, however, is routinely not gathered, because cancellations show up too late or are forgotten entirely. For health reasons, this food typically has to be dealt with.
Therefore, in collaboration with the Center for Entrepreneurship and Transfer (CET) at TU Dortmund University and Stattküche, Dr. Ina Dormuth and Teacher Markus Pauly developed a forecasting technique using Expert system (AI), so that production can be flexibly adjusted to actual day-to-day demand. To attain this, they first processed comprehensive historic order information and integrated it with specifications such as public holidays, climate condition, and data on influenza waves. They then checked different statistical and machine learning techniques, eventually designing a tool that analyses historic order patterns and supports future portion planning by predicting expected cancellations as specifically as possible. In this method, food waste is to be reduced, resources used better, and expenses lowered. The digital application has actually given that been integrated into Stattküche’s day-to-day workflows and supports the large-scale kitchen in its decision-making in a transparent, data-driven manner.
Methodologically Noise and Practically Relevant
The obstacle for the task team lay in establishing a method that was both methodologically sound and almost suitable: How can one gain from previous order and cancellation patterns? How can various forecasting methods be compared? And how can the outcomes be made functional for individuals who prepare meals on a daily basis? “Our objective was to release clinical methods in such a way that they truly assist in the daily running of a large-scale kitchen. The award shows that data-driven options can make a concrete contribution to greater sustainability,” says Professor Markus Pauly of the Department of Statistics, who is likewise a researcher at the Research Center Trustworthy Data Science and Security of the University Alliance Ruhr. Dr. Ina Dormuth, who carried out the core analytical work as a previous doctoral and postdoctoral scientist at Teacher Pauly’s chair, has actually been working as an Information Scientist in the Research and Development department of the Wilo Group considering that 2026.
The job “AI Rather of Waste” was supported by the European Digital Development Hub Dortmund (EDIH-DO). This consortium, with the participation of CET and TU concept GmbH, helps small and medium-sized business to expand their digital proficiencies and to carry out transfer projects from research into practice. The Center is co-financed by the EU and kinds part of the network of European Digital Development Hubs.
The “EDIH Network Award for Quality 2026” existed in early June at the EDIH Top in Brussels. With its innovative method, its tangible impact, and its transferability to other fields of application, the EDIH-DO and “AI Instead of Waste” were able to dominate the 2 other finalists. A total of 69 projects had been submitted.
Contact individual for inquiries: