Improving craft beer style classification through physicochemical determination and the application of deep learning techniques

Abstract

The consumption of craft beer at fairs and festivals is a phenomenon that keeps growing in the world. For this reason, it is important to control the quality characteristics of the different styles. This study aimed to analyze the different styles of beer, classify them according to their physicochemical parameters, and propose a predictive pattern-based model known as deep learning that best defines the styles that are presented at festivals. Physicochemical analyses of final gravity, color, alcohol, bitterness, and α-acids were carried out on eight styles of beer. The first four parameters are those that characterize the styles according to the Beer Judge Certification Program style guide. The incorporation of the α-acid determination allowed a more realistic classification that considers the brewers' new tendencies. This study will lay the foundations to improve local recipes, implement standardization, and provide training to local brewers.