Origin Identification of Buckwheat from Different Areas in Loess Plateau by Chemical Analysis and Data Mining Techniques

In this study, the quality characteristics of mineral elements, vitamins and amino acids of 39 common buckwheat samples from the eastern, central and western regions of Loess Plateau were analyzed and compared and then effective indicators were selected for discrimination geographical origins. The linear discriminant analysis (LDA) was used to determine the geographical origin of common buckwheat according to three chemical families and their combinations each other. All models showed different correct classification rate and cross-validation rate. Among them, the correct classification rate and cross-validation rate of model based on the relative content of amino acid (the percentage of one amino acid in total amino acids) in combination with mineral elements were both 97.4%. The results demonstrated that the established method using multivariate analysis and data mining techniques was very effective and feasible for identification origin of buckwheat in the Loess Plateau.