Discrimination between washed Arabica, natural Arabica and Robusta coffees by using near infrared spectroscopy, electronic nose and electronic tongue analysis

Background: The aim of this study is to investigate the feasibility of a 'holistic' approach, using near infrared (NIR) spectroscopy and electronic devices (electronic nose and electronic tongue), as instrumental tools for the classification of different coffee varieties. Analyses were performed on green coffee, on ground roasted coffee and on coffee beverage. Principal component analysis was applied on spectral and sensory data to uncover correlations between samples and variables. After variable selection, linear discriminant analysis was used to classify the samples on the basis of the three coffee classes: Robusta, natural Arabica and washed Arabica. Results: Linear discriminant analysis demonstrates the practicability of this approach: the external test set validation performed with NIR data showed 100% of correctly classified samples. Moreover, a satisfying percentage of correct classification in cross-validation was obtained for the electronic devices: the average values of correctly classified samples were 81.83% and 78.76% for electronic nose and electronic tongue, respectively. Conclusion: NIR spectroscopy was shown to be a very reliable and useful tool to classify coffee samples in a fast, clean and inexpensive way compared to classical analysis, while the electronic devices could assume the role of investigating techniques to depict the aroma and taste of coffee samples.