Ciência de dados na avaliação de amostras de café de qualidade extrema

Abstract

Demand for the quality of specialty coffees has driven the market and influenced the increased commercial value of coffee bags. In the Brazilian market, the state of Minas Gerais contributes a significant percentage of productivity, a fact that has been accompanied by quality coffee contests. This paper analyzed the first twenty samples of peeled cherry coffee ranked in the Concurso Mineiro de Qualidade do Café in 2013. Under the quantitative approach, an exploratory analysis was performed from the construction of the principal components with the 15 attributes. Subsequently, the prediction capacity of final grades was evaluated through a regression tree model. As a result, Linoleic and Palmitic acids were the attributes that most contributed to the construction of the first principal components. In addition, Linoleic acid was attributed as the root of the regression tree, which is a important attribute for the prediction of final scores.