Rapid Origin Identification of Chrysanthemum Morifolium Using Mid-Infrared Spectroscopy and Chemometrics

Abstract

The nutritional values and chemical compositions of chrysanthemum can vary depending on their geographical origin, which can significantly impact their final price. Conventional chemical analysis techniques are usually time-consuming and have the limitation of high consumption of organic reagents. Therefore, this study aimed to apply mid-infrared (MIR) spectroscopy combined with chemometrics for accurate and rapid classification of chrysanthemum from nine different origins. Various data pre-processing methods and variable selection methods were assessed, and the standard normal variate combined with the least absolute shrinkage and selection operator algorithm (SNV-LASSO) was ultimately chosen as the optimal processing method. The random forest (RF), K-nearest neighbors (KNN) and linear discriminant analysis (LDA) models demonstrated excellent classification performance, and the prediction accuracy on the validation set was 100% for all models. The established LASSO-LDA model for the prediction set achieved a high prediction accuracy of 98.6%. Hence, this work showed that the origin identification of chrysanthemum can be realized effectively based on MIR spectroscopy combined with chemometrics.

Keywords: Chrysanthemum, Origin identification, Mid-infrared spectroscopy, Chemometrics, Rapid analysis