| Resumo: | In this work, a methodology based on near-infrared spectroscopy (NIRS) was exploited in order to discriminate between commercial coffee brands. The main advantages of this approach compared to other strategies (e.g., wet chemistry methods) are its lower cost, less labor, and lower time per analysis. Two commercial brands were discriminated among several others present in the Portuguese market. The chemometric method used to estimate discriminant models was partial least squares discriminant analysis (PLSDA). Results show that it is possible to discriminate coffee brands using this strategy with a correct classification of 100 %. The spectral region, more favorable to discrimination of roasted coffee brands, can be related with differences in the concentrations of compounds, such as, chlorogenic acid and sucrose, and also due to differences on lipid fraction. This methodology is adequate for field implementation, namely, adopting handheld NIRS instruments.
|
|---|