A study of the composition of ancient glassware based on focal logistic regression models

Abstract

Glass has been developed in China for over two thousand years and was an important commodity in the economic and trade exchanges between China and abroad along the Silk Road. Since the Han Dynasty, glassware from the West began to be imported into China via the Silk Road, and Chinese glassmaking techniques have taken on foreign technologies, thus making ancient glass made in China similar in appearance to foreign glass, but the main chemical composition of the two is different due to the different regions and the different materials used. This paper focuses on the types of ancient glass artefacts that make up ancient glass, the patterns of content of each major chemical component and the specific mathematical relationships within them. This paper establishes a correlation analysis model based on information entropy using the information gain and information gain in the decision tree algorithm to correlate the weathering of glass artefacts with their color, decoration and type, and ultimately concludes that only the decoration of high potassium glass has a high correlation with its weathering. Then a nearest neighbor analysis model was established based on the idea of KNN nearest neighbor algorithm, and a joint focal logistic regression model was used to complement the weathering of the sampling points.