Prediction of the chemical composition content of ancient glass artifacts before weathering

Abstract

In this paper, the surface weathering, type, decoration, color and content of each chemical component of glass artifacts were studied and analyzed, and a K-means clustering model was established, using Spearman correlation analysis, chi-square test, and. It was solved to classify the glass types and analyze the change pattern of chemical composition of glass artifacts, and a better fitting effect was obtained. This paper characterized the problem as a prediction class, firstly, assigned values to four categorical variables: surface weathering, type, decoration and color of glass artifacts, and then used spss to perform Spearman correlation and chi-square test analysis to pre-process the data and eliminate invalid data. Then, it used descriptive statistical analysis to find that most of the chemical components of high potassium glass showed a decreasing trend after weathering, and most of the chemical components of lead-barium glass showed an increasing trend after weathering; finally, used Matlab matrix to derive a linear mapping relationship based on the changes of chemical components before and after weathering, and finally predicted the chemical components of glass artifacts before weathering.