Compositional Study of Ancient Glassware Based on K-Mean Cluster Analysis and Logistic Regression Algorithm

This paper focuses on the in-depth research on the chemical composition analysis and identification of ancient Chinese glass artifacts, and statistically analyzes and classifies and predicts the compositions of ancient glass artifacts using K-means cluster analysis and logistic regression algorithm. Firstly, some invalid data were excluded through data preprocessing, and the missing values were appropriately processed. In the analysis of the relationship between chemical composition and glass properties, the chi-square test was applied to find that there were significant differences between glass type and color and surface weathering. Secondly, by eliminating the SiO2 composition, which had a considerable influence on the clustering results, K-means cluster analysis was used to classify the subclasses of high-potassium and lead-barium glasses, and the reasonableness and sensitivity of the classification results were verified. Finally, the logistic regression algorithm was used to predict the type of unclassified glass artifacts, while the correlation between important chemical components and other components was explored by gray correlation analysis, which provided a new method and perspective for the compositional analysis and identification of ancient glass artifacts.