Your browser doesn't support javascript. This means that the content or functionality of our website will be limited or unavailable. If you need more information about Vinnova, please contact us.

Machine learning to reduce food wastage in restaurants

Reference number
Coordinator MAT OCH MÄTTEKNIK I UPPSALA AB - Green Innovation Park
Funding from Vinnova SEK 300 000
Project duration April 2017 - December 2017
Status Completed
Venture Innovative Startups

Purpose and goal

The project have produced a concept that uses machine learning to analyze food waste data in order to minimize the outcome of food that is cooked in vain. A significant factor to reduce food waste is to know how many guests that will dine and use this information in the cooking process. The project have also produced an algorithm that looks at kitchen data in order to categorize kitchens into different groups, this to make it visible which kitchens that succeed in their work with lowering food waste and in the long run spread this knowledge to other kitchens that might struggle in their work.

Expected results and effects

At the start of the project it was unclear about which parameters that had an impact on food waste and therefore which parameters that is important to catch in an algorithm. This was worked out together with SLU. In close cooperation with our partner municipality we selected to focus on predicting the number of guests. Data was collected from municipalities across Sweden that had the same characteristic as our test kitchens. The prototype succeeded with its predictions with a very satisfying result.

Planned approach and implementation

During spring we conducted a study about quantified parameters that effect food waste and in what proportion. This was done together with SLU. In the beginning of summer the algorithms and the concepts started to take shape. The prototypes were tested in our partner municipality during their food waste campaign weeks (2 weeks). Normally the campaign weeks are 4 weeks, which would have been better in order to obtain more data to validate how the algorithm deals with irregularities.

The project description has been provided by the project members themselves and the text has not been looked at by our editors.

Last updated 5 February 2019

Reference number 2017-00266

Page statistics