Application of an Electronic Nose as a New Technology for Rapid Detection of Adulteration in Honey

This work demonstrates the application of an electronic nose (e-nose) for discrimination between authentic and adulterated honey. The developed e-nose is based on electrodes covered with ionogel (ionic liquid + gelatin Fe3O4 nanoparticle) films. Authentic honey samples were submitted to analysis, capacity sensors was evaluated using principal component analysis (PCA) average relative response data. From PCA biplot, it possible note two well-defined clusters no intersection observed. To evaluate data as input autonomous classification, different machine learning algorithms evaluated, namely instance (IBK), Kstar, Trees-J48 (J48), random forest (RF), multilayer perceptron (MLP), naive Bayes (NB), sequential minimal optimization (SMO). Considering data, highest accuracy obtained Kstar: 100% (k-fold = 3). Additionally, this algorithm also compared regarding its sensitivity specificity, both being features. Thus, due rapidity, simplicity, methodology, technology e-noses has potential be applied quality control.