Prediction of Chemical Composition Ratio of Ancient Glass before Weathering Based on LASSO Regression

Abstract

Ancient glass is highly susceptible to weathering by the environment in which it is buried. During the weathering process, the internal elements of the glass undergo a large exchange with the environmental elements. This phenomenon leads to a change in the proportion of its composition. By analysing the chemical composition of ancient glass before and after weathering, the relationship between the changes can be identified, which means the original chemical composition can be predicted from the chemical composition of the glass after weathering. This is of great significance to the exploration and study of ancient artefacts. This paper first uses Spearman and Kendall correlation coefficient to identify the most influential factor on the weathering condition of glass from ornamentation, type and color, which eventually points to the glass type. This finding is secondly verified using a chi-square test. So the glass is divided into two types: high-potassium and lead-barium. By using the box plot to compare the chemical composition of high-potassium and lead-barium glass before and after weathering, it is found that there is a certain linear relationship between the chemical composition of glass before weathering and after weathering.LASSO regression was eventually applied to find the correspondence between the chemical composition of glass before and after weathering, thus enabling the prediction of the chemical composition of the glass before weathering.