Identification of Traditional Chinese Medicine Based on KNN Algorithm and Random Forest

Abstract

Medicinal materials are a system of components and complex mixtures, and spectroscopic principles can provide in-depth analysis of the composition mechanism of traditional Chinese medicinal materials through the determination of their material structures. Chinese medicinal materials of different origins and varieties exhibit different spectral characteristics due to differences in chemical composition and organic matter. However, spectral features have high-dimensional attributes, so PCA principal component dimensionality reduction is considered to condense spectral features into representative feature variables. Then, a hierarchical clustering method with a clear hierarchy can be used to classify multiple Chinese medicinal materials into three categories based on their spectral characteristics; At the same time, in order to solve the problem of origin identification, based on the situation that there are many classifications of origins, the KNN algorithm is combined to achieve the requirements of origin identification; Using both mid infrared and near infrared spectral data and using KNN algorithm to verify the origin of Chinese medicinal materials is more accurate; Due to the small number of categories, the prediction of the types of Chinese medicinal materials is implemented using commonly used random forests. The realization of the above methods demonstrates the feasibility of identifying Chinese medicinal materials through infrared spectroscopy, and it is also worth further exploration and research.