Glass classification and identification based on logistic regression model and K-means clustering analysis

Abstract

In order to investigate the effect of chemical composition on weathering, all post-weathering data were transformed into pre-weathering, for which the basis for the classification of high potassium and lead-barium glasses was inferred using a support vector machine algorithm. Afterwards, a detailed subclass differentiation method was determined for the two main classes of glass based on the K-means cluster analysis method. Further classification is then carried out on the basis of what is known about the classification. To identify unknown glasses, logistic regression was also carried out on the composition of weathered samples to predict the pre-weathering content.