A model for predicting geographic origin of honey from the same floral source
S Karabournioti, A Thrasyvoulou, E P Eleftheriou
A method for predicting geographic origin of honey of the same floral source coming from different areas is presented. For this purpose, Greek thyme honey was selected, and melissopalynological analysis of 180 samples produced in different areas was carried out. Discriminant analysis was applied to the results and a discriminant model was made by using absolute pollen grains frequency. The analysis showed a high degree of discrimination for honeys examined and a significant discriminative power for the model. The simulation performance of the model was estimated from an external testing set with verification from 90% to 100%, according to geographic area. Emphasis was given to the sampling method which limits the influence of beekeeping handling and accidental contamination and increases validity of the predictive model.
honey, melissopalynology, geographical origin, discriminant analysis, predictive model