RESUMO
This work aims the study chemometric methods for the classification of the origin of coffee samples. Samples of finely pulverized coffee grains were analyzed by synchronous molecular fluorescence spectroscopy to carry out the classification. The spectral data of the samples were obtained in triplicate in two offsets: 10 nm (with emission wavelengths from 240 nm to 600 nm) and 40 nm (from 240 nm to 560 nm), all with 1 nm resolution. Different strategies were performed using the spectra obtained with the offsets of 10 nm and 40 nm and fused data at mid-level (10 nm + 40 nm). The performances of linear and nonlinear methods were compared, the best results were obtained from the raw data from the fusion at low-level of the 10 nm and 40 nm offset spectra with the Pareto optimization criterion.
