Recommendation of Roast Curves for Speciality Coffee

Thesis Type Master
Thesis Status
Student Jonas Stock
Thesis Supervisor
Research Field

Speciality coffee enjoys rising popularity, as more and more people tend to choose their groceries consciously. But still, roasters of quality coffee mostly rely on their experience to determine how the green coffee must be roasted to receive a certain result. This thesis explores a way to improve this process by automatically recommending a roasting profile based on the green coffee characteristics, external influences (e.g., weather) and the desired outcome. To achieve this goal, we apply current recommendation and classification techniques on a unique knowledge database, which was collected over years by a software platform for specialty coffee, and which contains detailed information on thousands of roasting processes.