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

Abstract

Ancient glass is highly susceptible to weathering by the burial environment, with a large number of internal elements exchanging with environmental elements, resulting in changes in the proportions of their composition. In order to analyse the classification patterns of High potassium-rich glass and Lead and barium enriched glass, and to classify each type of glass into appropriate subcategories, Decision tree was first chosen to broadly classify the decision factors for the glass types, and then lasso-logistic was chosen for the exact solution. K-means clustering was used in combination with Factor Analysis Method to achieve subcategories for both types of glass. Finally, using the above model, the chemical composition of the unknown glass is used to identify the type.