<![CDATA[Coffee taste quality is a key factor influencing customer satisfaction and reflecting the professionalism of baristas in coffee service operations. However, manual coffee taste assessment remains highly subjective and often leads to perceptual differences between professional assessors and consumers, resulting in inconsistent flavor quality. To address this challenge, this study aims to develop and evaluate a machine learning-based system for predicting coffee taste quality, which supports the sensory assessment process in operational coffee business environments. The proposed system utilizes the Naive Bayes algorithm and is implemented as a web-based application to facilitate structured data management, automated classification, and informed decision-making during the espresso dialing-in process. The research data were obtained from coffee taste assessments involving baristas as trained panelists and customers as consumer panelists. Sensory attributes, including coffee bean origin, brewing time, resting period, water temperature, and extraction results, were collected and processed as classification input features. The evaluation involved comparing assessment patterns between baristas and customers to analyze consistency across key sensory attributes. The results show high agreement between barista and customer assessments for attributes such as body and balance, while noticeable differences were observed in acidity and sweetness. At the system evaluation stage, the model was tested on a limited dataset consisting of seven samples and correctly classified six instances, achieving an accuracy of 85.71%. These findings indicate that the proposed system has strong potential to support objective and consistent coffee taste quality evaluation in real-world operational settings.]]>
Copyrights © 2026
