Species and geographical origin authenticity of green coffee beans using UV–VIS spectroscopy and PLS–DA prediction model

Abstract

Coffee with different regions and species tends to have different chemical compositions and, therefore, gains variable commercial values. This means that the authenticity of green coffee beans, the raw material for coffee beverages, is necessary for coffee supply chains. This paper focused on the application of ultraviolet–visible (UV–VIS) spectroscopy combined with the Partial Least Squares with Discriminant Analysis (PLS–DA) model to authenticate Vietnamese green coffee beans. Spectra of 147 samples were used to build prediction models. PLS–DA model combined with the Standard Normal Variate (SNV) and second order of derivative performed excellently in predicting coffee species with a percentage of cross–validation accuracy at 100.00%. With the treatment of SNV and the first order of derivative, percentages of cross–validation accuracy for the PLS–DA model authenticating the origin of samples from all provinces varied from 88.23% to 95.80%. The results showed a high potential of the UV–VIS spectroscopy techniques in the authenticity of green coffee beans.