Fast recognition of the harvest period of Porphyra haitanensis based on mid-infrared spectroscopy and chemometrics

The quality of seaweed is greatly influenced by its harvest period. Hence, the development of efficient and precise analytical techniques for identifying the harvest period is significant from a practical standpoint. In this study, mid-infrared spectroscopy was utilized to perform quantitative analysis of nutritional components, evaluate the quality grades, and determine the harvest period of Porphyra haitanensis. While the outcomes of one-factor analysis of variance and correlation analysis indicated a general decrease in P. haitanensis quality with each subsequent harvest, the nutritional contents of samples obtained from the first and fifth harvest were similar, suggesting that the harvest period alone could not fully represent the quality of seaweed. To create discriminant models of the harvest period, decision tree, K-nearest neighbor, partial least squares-discriminant analysis, and adaptive boosting algorithms were adopted based on band integral areas and a small sample size (30 per harvest period, 6 periods in total). The adaptive boosting algorithm provided the most comprehensive results with regards to forecast accuracy (96.7%), macro precision (0.971), macro recall (0.965) and RMSE (0.183). The findings demonstrated that band integral areas of the mid-infrared spectroscopy not only provided quantitative information on the nutrients but also offered effective prediction variables for classification algorithms. Thus, the proposed method can serve as an essential instrument to evaluate the seaweed quality swiftly and precisely. [GRAPHICS] .