LAUSR

Abstract The geographical origin of sea cucumber Apostichopus japonicus plays a key role in affecting its economic value. To quickly and effectively identify the geographical origin of sea cucumbers, Fourier transform near infrared (FT-NIR) spectroscopy coupled with machine learning methods (random forest, gradient boosting decision tree, light gradient boosting machine) was applied and compared in present study. The results showed that a light gradient boosting machine (lightGBM) model achieved the best performance by proper sampling and preprocessing techniques. The mutli-class logloss during the model training can reach as low as 0.36. The accuracy, precision, recall, F1 score and the area under curve for the test sets prediction was 0.91, 0.92, 0.91, 0.91, 0.98, respectively. These indicators showed the lightGBM model established has good robustness and strong generalization ability. The results proved that NIR spectroscopy combined with lightGBM could be used as a rapid and effective technique for tracing the geographical origin of sea cucumbers.