Cromatografía de gases ultrarrápida acoplada a técnicas de machine learning para predecir el nivel de adulteración en miel de azahar y girasol
- Punta -Sánchez, I. 1
- Dymerski, T. 2
- Calle J. L. 1
- Ruiz -Rodríguez, A. 1
- Ferreiro-González, M. 1
- Palma, M. 1
-
1
Universidad de Cádiz
info
-
2
Gdańsk University of Technology
info
- Fernández-Daza Álvarez, Carmen (coord.)
Publisher: Centro Universitario Santa Ana
ISBN: 84-7930-113-9
Year of publication: 2024
Pages: 239-259
Congress: Congreso Agroalimentario de Extremadura (5. 2023. Almendralejo)
Type: Conference paper
Abstract
Honey adulteration is a major problem in the food industry,and detection of these fraudulent practices isessential to ensure product quality and authenticity. Ultra-fast gas chromatography (ultra-fast GC) is a fast andsensitive analytical technique for detecting adulterationin honey and, in combination with machine learning algorithms,has proven to be an effective tool for developingaccurate and reliable models to detect adulteration.In this study, several machine learning techniques werecompared to predict the level of adulteration in orangeblossom (OB) and sunflower (SF) honey using ultrafastGC. The OB and SF honey samples were adulteratedusing a mixture of other honeys from different floralorigins as adulterant. Adulterated samples were obtained with a range of honey purity that oscillated between50% and 95%. It was found that the supported vector regression(SVR) showed the best performance with an R2of 0.9086 for the data set containing orange blossom andsunflower honey. To improve the accuracy of the regressionmodels, it was proposed to classify the honey samplesbased on their botanical origin and then apply theregression models separately. All the regression modelstested on orange blossom and sunflower honey separatelyobtained superior performance. The least absoluteshrinkage and selection operator (LASSO) turned out tobe the best to predict the properties of orange blossomand sunflower honey, with an R2 of 0.9987.