Vis-NIR Spectroscopy and PLS-Da Model for Classification of Arabica and Robusta Roasted Coffee Bean

Abstract

Visible-Near Infrared (Vis-NIR) spectroscopy combined with partial least squares discriminant analysis (PLS-DA) was used to classify Arabica and Robusta roasted coffee beans. The number of coffee beans analyzed was 200 samples consisting of 5 origins (Flores, Temanggung, Aceh Gayo, Jawa, and Toraja). Reflectance spectra with a wavelength of 450-950 nm were used to build two types of models, namely single-origin and general models. Single-origin Flores, Temanggung, Aceh Gayo, and Toraja models performed very well to classify coffee beans samples from the same origin with Sen, Spe, Acc, and Rel of 1, as well as TFN and TFP of 0. General PLS-DA model with baseline correction pretreatment yields Sen, Spe, Acc, and Rel of 0.97, as well as TFN and TFP of 0.04. Based on this paper, it was concluded that Vis-NIR combined with PLS-DA perform well in classifying roasted coffee beans based on the variety.