Near- and mid-infrared spectra of a set of 56 lyophilized coffee samples were collected. The sample set comprised 29 Arabica and 27 Robusta coffees and was prepared in the laboratory. Each spectral collection was used separately to develop mathematical models for varietal authentication of the coffees using factorial discriminant analysis and partial least squares techniques. Subsequently, data from both spectral regions were combined and the chemometric approaches repeated. The relative success of the separate and combined approaches is discussed, as is the basis for the observed discriminations.
- food authentication