Document Type : Research paper
Authors
1
Master Program of Agronomy, Faculty of Agriculture, Universitas Padjadjaran, Sumedang 45363, Indonesia
2
Department of Agricultural Engineering, Faculty of Agriculture, Universitas Syiah Kuala, Banda Aceh 23111, Indonesia
3
Department of Agricultural Engineering and Biosystem, Andalas University, Limau Manis, Padang, 25163, Indonesia
Abstract
The improvement of research on the prediction of coffee beans quality by near infrared spectroscopy (NIRS) offers the benefit of high efficiency and precision. The objective of this study was to utilize NIRS to estimate caffeine and chlorogenic acid (CGA) levels in coffee beans sourced from two distinct geographical areas. While models were typically customized for specific products, broadening their applicability had the potential to increase productivity. This multivariate analysis used a total of 50 samples, including both Arabica and Robusta varieties. The results showed a caffeine model with coefficient of correlation in calibration set (Rcal), root mean square error in calibration set (RMSEC), coefficient of correlation in cross-validation set (Rcv), root mean square error in cross-validation set (RMSECV), and ratio of prediction to deviation (RPD) values of 0.85, 0.30, 0.82, 0.31, and 2.21, respectively. Simultaneously, the CGA model produced values of 0.88, 0.61, 0.88, 0.65, and 2.18. The wavelengths at 1122, 1452, 1682, and 1950 nm showed a close association with caffeine and water, while 1415, 1718, and 1909 nm correlated with CGA. The investigation showed that the accuracy of the model was satisfactory. The application of NIRS for predicting caffeine and CGA in coffee beans held significant potential as an alternative to conventional laboratory analysis methods.
Keywords
)