Abstract
Several methods have been proposed with the aim of improving the precision of quantitative measurements of biological components (baseline correction, classification, elimination of unwanted components, etc.). In this context, we propose a classification method of biological samples (raw sugar cane juices) before sucrose content prediction is performed. The method consisted of isolating the two most dissimilar individuals from a large calibration family of mid-FT-IR spectra, and, by successive principal component analysis (PCA) and principal component regression (PCR), a family composed of a few individuals was constituted. Each individual from this family represented the first spectrum of the corresponding classes that were ultimately formed. The classification of the remaining samples from the calibration family was carried out by the mobile centers method, that is, by the measurements of the Euclidian distances. This procedure improved the precision of the predictions. The mean and standard deviation (SD) of the differences between predicted and reference values were, respectively, -1.62 x 10<sup>-3</sup> and 0.308 before classification and 2.38 x 10<sup>-3</sup> and 0.254 after classification. The procedure developed in this paper first allowed a qualitative classification of spectra without knowledge of their chemical composition, and second, improved the precision of the quantitative predictions.
PDF Article
More Like This
Cited By
You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
Contact your librarian or system administrator
or
Login to access Optica Member Subscription