Classification of Arabica Java Coffee Beans Based on Their Origin using NIR Spectroscopy

Abstract

Abstract The NIR spectroscopy for characterization and classification of intact Java arabica coffee beans based on their origin is proposed. Commonly, NIR spectroscopy is used to determine composition of agricultural products including coffee. Three kinds of Java arabica coffee beans namely Arabica Java Preanger, Arabica Bondowoso and Arabica Malang were used in this research. Three hundred samples, each consisted of 100 g coffee beans were prepared. The coffee beans of 100 g were placed in petri dish and the light reflectances of intact coffee beans were measured by FT-NIR spectrometer at the wavelength of 1000-2500 nm. After reflectance measurement, the samples were subjected to composition analysis using proximate and Liquid Chromatography Mass Spectrometry. The reflectance and absorbance spectra were processed by five spectra data pretreatments (smoothing, first derivative, second derivative, Standard Normal Variate (SNV), Multiple Scatter Correction (MSC)) and then Principle Component Analysis (PCA) were carried out. Discriminant analysis (DA) of Principle Components were developed to classify coffee beans based on their origin. The results show that the spectra data treatments of SNV and MSC of reflectance spectra and PC analysis using PC1 and PC2 gave the best results for discriminating three kinds of coffee beans. The DA of three principle components of reflectance data processed by SNV and MSC could classify arabica coffee beans accurately (100%). This results shows that NIR spectroscopy can be used as a nondestructive method to classify arabica coffee beans based on their origin.