Mulberry leaf tea variety classification method based on machine learning of polyphenol components and mineral features

Abstract

The invention relates to the technical field of tea identification, and discloses a machine learning mulberry leaf tea variety classification method based on polyphenol components and mineral features, which comprises the following steps: S1, sample collection: collecting eight mulberry leaf tea samples, and measuring polyphenol and mineral content; s2, data cleaning: abnormal values are removed, and Z-score is used for detection and standardization; s3, dimension reduction processing: performing dimension reduction by using PCA, t-SNE and UMAP; s4, carrying out clustering analysis: clustering by adopting a K mean value; and S5, model training: training and evaluating the model by using a random forest, LightGBM and SVC. According to the method, the characteristics of polyphenol components and mineral substances are combined, the polyphenol components and mineral elements are combined, and classification is carried out by adopting a random forest algorithm from various algorithm models, so that the characteristic difference of tea varieties can be accurately captured, high accuracy is realized among various tea varieties, and the classification precision is effectively improved.

Images