ABSTRACT
Coffee is one of the most consumed hot beverages worldwide. The beverage is prepared from roasted and ground coffee beans, which, in their raw state, are harvested from coffee plants all around the world. Aroma patterns were evaluated by the Sammon mapping technique, allowing for both the separation of Arabica and Robusta coffees into two distinct groups and their discrimination by the respective geographic origins. The ANN correctly classified and tested 100% of the samples by their respective geographic origins. Multivariate statistical models were applied to the NMR data set and were able to identify significantly different levels of 14 metabolites of green coffee beans, including sucrose, caffeine, CGAs, choline, amino acids, organic acids, and trigonelline, allowing for the discrimination of both species and geographic origins. The combined use of color parameters and hydrosoluble compounds content ratio was considered suitable for the differentiation of coffee species in blends of roasted samples.
