Today, food waste work is often about taking care of food afterwards. Now we want to test if we can use our food data to plan away food waste in advance.


Large parts of the food produced today are thrown away - a third of it to be exact. In our school canteens, it is thrown away much less, but it is still a lot of food that ends up in the garbage. After two years of intensive work with food waste within the SmartMat Hbg project, the City of Helsingborg's schools have reduced their food waste by 48%. We have gone from throwing 240 tons of food a year to almost 120 tons. During these two years, we have collected a lot of data about our food waste as the chefs have every day reported kitchen waste, serving waste and plate waste in a database.

Now we wonder: Can we use this information to plan out future food waste? Can we predict wastage before it occurs?


By linking food waste data with other types of data (for example from menus, food purchases and absenteeism statistics), we investigate whether an AI solution can help us predict food waste before it occurs.

The idea is being tested and different types of data have been collected and started to be used to predict what food waste will look like in advance. The next step is to test whether it is the right data we use and whether the predictions can be used for planning.

Data we use:

  • Waste registrations (Managed report 460)
  • Menu from Skolmaten.se  (open platform)
  • Prescription
  • We are also looking at being able to use: food invoices, absence data, food queue measurement, weather data.

AI features:

The algorithms behind the solution classify various parameters that can affect wastage and find patterns between, for example, fish sticks and breaded fish. With the help of historical wastage data, the solution then provides predictions at school level.


It has been difficult to get an automatic flow from the registration of food waste to the city's central handling of data, something that during previous tests has proven to be important for us to be able to apply prediction models in a good way. Now, however, there is an API for registration and management and we are waiting for the connection to the system for self-control management in which we will register food waste in the future. In parallel, we are developing a prototype for a food waste calculator that can serve as an interface for children and students who want to understand how different parameters can affect food waste.

At the H22 City Expo:


Impact target

Further reduced food waste


Name: Cecilia Larsson
E-mail: cecilia.larsson@helsingborg.se

Name: Mathias Andersson
E-mail: mathias.andersson3@helsingborg.se