Untargeted classification for paprika powder authentication using visible - Near infrared spectroscopy (VIS-NIRS)

This paper describes a non-destructive screening method for authentication of paprika belonging to the Spanish Protected Designation of Origin (PDO) "Pimenton de La Vera". Different multivariate classification models were developed in order to differentiate PDO and non-PDO samples, using visible-near infrared spectra as fingerprint for each paprika sample. Sample treatment was not required. Principal component analysis (PCA) was applied in different spectral ranges: 400-2500, 400-800 and 800-2500 nm. In all spectral ranges, PCA was largely able to differentiate PDO from non-PDO samples. Partial least-squares discriminant analysis (PLS-DA), PCA-linear discriminant analysis (LDA) and PCA-quadratic discriminant analysis (QDA) were used as classification methods in the different spectral ranges. All methods were able to differentiate PDO from non-PDO samples, with error rates (ER) lower than 0.15. The best models were those obtained with PLS-DA in the NIR range (800-2500 nm), showing ERs lower than 0.07 and error indexes (IERROR) (false positives) lower than 0.05.