Analysis of the composition of ancient glass objects based on cluster analysis and Bayesian discriminant methods

Abstract

The aim of this study is to investigate the open problem of classification laws and results. First, a comprehensive research method of quantitative analysis-Bayesian discrimination is used to reasonably and accurately investigate the classification laws of high potassium glass and lead-barium glass, obtaining the Bayesian discriminant for high potassium glass ????1 = 2.909????1 - 0.117????2 + 4.487 ????3 + 2.432????4 + 1.097????5 + 3.352????6 + 3.403????7 + 6.303????8 + 2.669????9 + 0.897????10 + 1.882????11 - 4.543????12 + 5.002????14 - 150.381 with the lead-barium glass Bayesian discriminant ????2 = 2.447????1 - 1.195????2 + 42.807????3 + 2.117????4 + 1.0971.534????5 + 3.221????6 + 3.299????7 + 4.830????8 + 2.513????9 + 1.226????10 + 1.590????11 - 5.628????12 + 3.087????13-111.481. Then, using the clustering process-Bayesian discrimination method, the number of sample clusters was roughly determined using systematic clustering, and the data were categorized in detail using K-Means clustering, resulting in the possibility of classifying each of the high-potassium and lead-barium glasses into Three subclasses were identified, and Bayesian discriminant functions were derived for each subclass, and the data were tested to justify the subclasses.