Study on Ancient Glass Classification Based on Fisher Linear Discriminant Analysis and Systematic Cluster Analysis

Abstract

This paper discusses the classification of high potassium glass and lead-barium glass, and puts forward the model establishment and solution method using Fisher linear discriminant analysis and system cluster analysis. This PAPER points out THE obvious changes of chemical composition in glass cultural relics, and explores the classification rules of glass according to the different weathered states. The establishment of the model includes the steps of index selection, training group and test group division, Fisher linear discriminant function construction and cluster analysis. Through independent sample t-test and data analysis, the chemical components affecting the classification of glass were selected, and the Fisher linear discriminant function was successfully constructed, which reached 100% classification accuracy on the training group data. In the systematic cluster analysis, the sub-categories of high potassium glass and lead-barium glass are realized by distance calculation and grouping merging. The results of the model have been verified and sensitivity analysis show that its high reliability and robustness have the potential to accurately classify glass in practical applications. These methods provide effective analytical tools and strategies to solve the problem of ancient glass classification.