Elemental fingerprinting combined with machine learning techniques as a powerful tool for geographical discrimination of honeys from nearby regions

Discrimination of honey based on geographical origin is a common fraudulent practice and is one of the most investigated topics in honey authentication. This research aims to discriminate honeys according to their geographical origin by combining elemental fingerprinting with machine-learning techni...

Full description

Bibliographic Details
Authors: Mara, Andrea, Migliorini, Matteo, Ciulu, Marco, Chignola, Roberto, Egido, Carla, Núñez Burcio, Oscar, Sentellas, Sonia, Saurina, Javier, Caredda, Marco, Deroma, Mario A., Deidda, Sara, Langasco, Ilaria, Pilo, Maria I., Spano, Nadia, Sanna, Gavino
Format: article
Status:Published version
Publication Date:2024
Country:España
Institution:Varias* (Consorci de Biblioteques Universitáries de Catalunya, Centre de Serveis Científics i Acadèmics de Catalunya)
Repository:Recercat. Dipósit de la Recerca de Catalunya
OAI Identifier:oai:recercat.cat:2445/206171
Online Access:https://hdl.handle.net/2445/206171
Access Level:Open access
Keyword:Mel d'abelles
Taxonomia botànica
Espectrometria de masses de plasma acoblat inductivament
Honey
Botanical taxonomy
Inductively coupled plasma mass spectrometry
Description
Summary:Discrimination of honey based on geographical origin is a common fraudulent practice and is one of the most investigated topics in honey authentication. This research aims to discriminate honeys according to their geographical origin by combining elemental fingerprinting with machine-learning techniques. In particular, the main objective of this study is to distinguish the origin of unifloral and multifloral honeys produced in neighboring regions, such as Sardinia (Italy) and Spain. The elemental compositions of 247 honeys were determined using Inductively Coupled Plasma Mass Spectrometry (ICP-MS). The origins of honey were differentiated using Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), and Random Forest (RF). Compared to LDA, RF demonstrated greater stability and better classification performance. The best classification was based on geographical origin, achieving 90% accuracy using Na, Mg, Mn, Sr, Zn, Ce, Nd, Eu, and Tb as predictors.

Similar Items