Analysis and research on the chemical composition of ancient Chinese glass

Abstract

The purpose of this study is to investigate the classification rules for lead-barium and highpotassium glass. First, we noticed that the glass types differed considerably for the five components. After analyzing the data and establishing the independent T-test model, We developed the Logistic regression classification model using these components as independent factors and the glass types as dependent variables, and we derived the classification function for the glass types. Second, high-potassium glass, was divided into two subclasses based on calcium and aluminum content using the K-means++ model and associated literature on the chemical composition of ancient Chinese glass [1][2]. These were grass-ash high-potassium glass and nitrate high-potassium glass, as well as two types of lead-barium glass: lead-sodium glass and lead-calcium glass. Finally, the cluster center coordinate values were perturbed by -10% and 10%, verifying the classification of each sample under different perturbations, and the resulting subclass divisions remained unchanged, confirming the low sensitivity of the classification results.

Keywords

• logistic regression
• k-means++ model
• sensitivity analysis