Chemical compositions prediction of glass artifacts based on logistic regression and linear weighting

Abstract

Chemical composition analysis is of great significance for studying the category, origin and production age of the ancient glass. In order to consider the influence of weathering on chemical composition, this paper proposes a prediction method of glass chemical composition based on logistic regression and linear weighting. First, the chi-square test shows that whether the glass surface is differentiated is related to the type of glass. Secondly, the chemical composition of glass relics is divided into two categories by binary logistic regression: significantly affected by weathering and not significantly affected by weathering. Then, for the chemical components not significantly affected by weathering, combined with descriptive statistical results, the mean value was used as the chemical composition content before weathering. For the chemical components significantly affected by weathering, the linear weighting method was used to predict the content of chemical components before weathering. The final results show that this method can better reflect the influence of weathering factors on the chemical composition of glass, which is of great significance for the prediction of the glass composition.