Portal de Periódicos da CAPES

Machine Learning for Cup Coffee Quality Prediction from Green and Roasted Coffee Beans Features

2020; Springer Science+Business Media; Linguagem: Inglês

10.1007/978-3-030-61834-6_5

1865-0937

Javier Andrés Suarez-Peña, Hugo Fabián Lobatón García, José Ignacio Rodríguez Molano, William Camilo Rodríguez-Vásquez,

Advanced Chemical Sensor Technologies

Coffee is one of the main exported products of Colombia. It is grown in different regions throughout the territory and is recognized worldwide for its flavor and freshness. Its quality is evaluated by professional tasters, who taste the coffee drink obtained from roasted coffee beans. They qualify it according with the platform or method requested by customers. This study proposes the use of different Machine Learning (ML) algorithms for the prediction of cup coffee quality, based on a set of measurements made to almond and roasted coffee beans. The data was obtained with the support of Almacafé, a company belonging to the National Federation of Coffee Growers (FNC) of Colombia. The classification results with the validation set, showed a higher accuracy with the Neural Network algorithm, with an average score of 81% for a 10-fold stratified cross validation. This work demonstrates the possibility of qualifying cup coffee quality with ML algorithms.