RESUMO
ATR-MIR spectroscopic analysis was used to classify sliver carp surimi and surimi products adulterated with different levels of hairtail surimi. Five chemometric methods, including SIMCA (soft independent modeling class of analogies), KNN (K-nearest neighbor), SVR (support vector machines regression), PLSDA (partial least squares discriminate analysis) and ID3 (interative dicremiser version 3) Decision tree were used to build the classifying models. And the performances of the models were compared. Results showed that for both cooked and uncooked mixed surimi samples, better classifications were obtained using SIMCA model, the percentage of the correct classification reached 96.59% and 96.43%, and the corresponding RMSECV were 0.185 7 and 0.189 8, r value were 0.988 0 and 0.994 1 respectively. The results of this study demonstrated for the first time that ATR-MIR spectroscopy combined with chemometrics method can be used to classify sliver carp surimi and surimi products adulterated with different levels of hairtail surimi.