Forecasting changes production in the Flemish food sector

The Flemish food industry is at a turning point. Research from Hogeschool VIVES shows how data-driven forecasting and AI can help achieve more efficient production and reduce food waste. For a long time, planning relied heavily on experience and intuition, but the supply chain is becoming more complex, consumers have higher expectations, and societal pressure is increasing.

By combining historical data with real-time sales figures and external factors, companies gain a clearer picture of what, when, and in what quantities to produce. This leads to more efficient production, less waste, and a smaller ecological footprint.

Planning, waste, and practice

In many companies, production planning is still largely based on gut feeling. Excel and internal spreadsheets often serve as the starting point, supplemented by the experience of planners. Safety buffers are usually determined manually, even when data is available. Food waste is recorded, but the level of systematization and accuracy varies greatly. Sometimes tracking happens daily, other times only after a production cycle. Causes are diverse, including overproduction, unpredictable demand, or returns. Companies employ a range of strategies to limit waste: reprocessing surplus, donating to food banks, selling at a discount, or composting. Yet many organizations still lack insight into underlying patterns, making structural reduction difficult.

AI, VIVES research, and collaboration

Within the TETRA project Smart Meal Planning, Hogeschool VIVES works with partners to investigate how AI can help reduce food waste and improve operational efficiency. This hands-on research combines data from production, orders, and external factors to improve demand forecasting. AI can uncover patterns that planners might miss, such as the impact of school holidays or temperature on the demand for certain meals, and continuously adjust forecasts using real-time data. Leading companies are testing these AI-driven forecasting modules so that data supports human decision-making rather than replacing it. Collaboration with research institutions, technology partners, and other companies helps share insights, make forecasting more reliable, and cut down on waste.

Vives.be