Abstract

Tricholoma matsutake is an expensive product in the global edible fungus market due to its nutritional and medicinal properties. As the price of T. matsutake increases, some adulterated and fraudulent products have also emerged in the market. It is difficult to detect the fraudulent products with similar shape, and the unfair competition often happens. Discriminant methods combined near infrared spectroscopy with chemometrics analysis have been used in many fields. However, due to the high correlation between the spectral data, it is difficult to construct an effective model using original spectra. In this work, a discriminated model developed by the elastic net algorithm and near infrared spectroscopy is presented to determinate the adulterated and fraudulent products of T. matsutake. First, the difference of protein and aspartic acid contents between T. matsutake and three products with similar shape were analyzed. Then, the information variables selected from near infrared spectroscopy using the elastic net were used to establish quantitative models. And, the prediction performance of developed models was evaluated by using the validation set. Finally, the Monte Carlo experiment based on the test set demonstrated the effectiveness of the proposed method. Compared with least absolute shrinkage and selection operator and partial least square regression models, the developed model has a great prediction accuracy and robustness, which can be served as a new discriminant method for T. matsutake adulteration determination.

© 2020 The Author(s)

PDF Article

References

You do not have subscription access to this journal. Citation lists with outbound citation links are available to subscribers only. You may subscribe either as an OSA member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access OSA Member Subscription

Cited By

You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an OSA member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access OSA Member Subscription