Authentication of the Geographical Origin of Shandong Scallop Chlamys farreri Using Mineral Elements Combined with Multivariate Data Analysis and Machine Learning Algorithm

Geographical traceability of seafood is a global concern for both consumers and importers. It is urgent to develop a scientific approach for identifying the geographic origin of seafood to combat labeling fraud. This study verified 14 mineral elements as a tracer for identify the geographic origin of scallops in Shandong Province of China. Multivariate data analysis and machine learning algorithm including linear discriminate analysis (LDA), k-nearest neighbor (KNN), random forest (RF) and support vector machine (SVM) were used to evaluate their performance in terms of classification or predictive ability. Thirteen elements in scallop samples with different regions showed significant differences (p < 0.05), which proved that the elemental composition was an effective tool for distinguishing the origins of scallops. The overall discrimination accuracy and predictive accuracy obtained from the LDA, KNN, RF, and SVM analysis was over 98.96% and 97.78%, respectively. Among these models, LDA model was the most recommended for the origin identification of scallops based on its high discriminant accuracy rate (100%), cross-validated accuracy rate (100%), and predictive accuracy rate (100%). Present results indicated the feasibility of element fingerprints combined with multivariate data analysis and machine learning algorithm in authenticating the geographical origin of scallops in China.