Analyzing Coffee Maturity to Determine Caffeine Content Using KNN

This study analyzes the caffeine content in various types of coffee powder using a color profile approach as a non-destructive identification method. Regulating caffeine consumption within safe limits is crucial, especially for individuals who monitor their daily intake. The validity of data in this research is supported by several key assumptions. First, each coffee variety possesses a unique volatile compound profile that can be identified using aroma sensors. Second, roasting level influences the color variation of coffee powder, which correlates with caffeine content. Third, digital image acquisition must be conducted consistently in terms of lighting and distance to ensure accurate RGB values. Prior to caffeine analysis, coffee types were identified by measuring gas concentrations such as methane, alcohol, hydrogen sulfide, ammonia, volatile organic compounds (VOCs), and butane. The aroma sensor (E-Nose) was employed to distinguish Arabica, Robusta, and Liberica coffee based on emissions released during roasting and grinding processes. A total of 390 samples were tested, consisting of three coffee types across three roasting levels. The results indicated that caffeine levels ranged from 0.4% to 0.7%, still within safe consumption thresholds. A significant negative correlation was observed between color intensity and caffeine concentration. The K-Nearest Neighbors (KNN) model achieved training accuracy of 82.23%. These findings demonstrate that combining aroma sensor-based coffee identification with color profile analysis offers a practical and accurate method for caffeine standardization in the food and beverage industry, pharmaceuticals, and other related sectors.