z-logo
Premium
An attempt to classify the botanical origin of honey using visible spectroscopy
Author(s) -
Lorenc Zofia,
Paśko Sławomir,
Pakuła Anna,
Teper Dariusz,
Sałbut Leszek
Publication year - 2021
Publication title -
journal of the science of food and agriculture
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.782
H-Index - 142
eISSN - 1097-0010
pISSN - 0022-5142
DOI - 10.1002/jsfa.11176
Subject(s) - principal component analysis , honeydew , pollen , pattern recognition (psychology) , nectar , artificial intelligence , mathematics , data reduction , computer science , preprocessor , biological system , botany , data mining , biology
BACKGROUND The production of honey, and especially the unifloral varieties, is limited by factors such as weather conditions or the availability of nectar flow and honeydew. This results in a deficit in supply leading to the adulteration of honey. If they are not properly labeled, customers cannot distinguish artificial / synthetic products from real honey. Currently, the basic, commonly used method for determining the varieties of honey (botanical origin) is palynological analysis. However, this procedure is quite difficult owing to the dearth of experienced staff in the field of melissopalynology. RESULTS A method for identifying and classifying natural honey accurately based on its botanical origin has therefore been proposed. This analysis would rely on the visible light spectra transmitted through a relatively thin layer of the substance of interest, regardless of deviations in thickness. We present algorithms for analyzing the transmittance spectra‐parametrization based on polynomial approximation (PMA) and applying a method for spectra selection and reduction (SSR) and a classical classification model (decision tree). A comparison is presented of the classification of four varieties of honey, confirmed by pollen analysis, obtained from the analysis of optically measured transmittance spectra of the samples. The algorithms that are compared contain a decision tree that uses raw data, data reduced by principal component analysis (PCA), and data after calculations based on the proposed algorithms alone (PMA and SSR) and together with the PCA method. CONCLUSION This novel method produced outstanding results in comparison with the standard PCA method and is helpful in identifying the botanical origin of honey effectively. © 2021 Society of Chemical Industry.

This content is not available in your region!

Continue researching here.

Having issues? You can contact us here