Research On the Classification Patterns of High-Potassium Glass and Lead-Barium Glass Based On K-Means Clustering Analysis

Abstract

This study employs the K-means clustering analysis algorithm to conduct a comprehensive investigation into the classification patterns of high-potassium glass and lead-barium glass. Initially, by selecting two cluster centers for analysis, significant differences in composition between high-potassium and lead-barium glass are revealed through the merging and simple processing of data samples from forms one and two. Subsequently, extensive analysis and statistics with multiple cluster centers reveal a suitable cluster ratio of 1:3, leading to the determination of four cluster centers for further analysis, resulting in the identification of four subcategories. Detailed analysis indicates one as a high-potassium type and three as lead-barium types. Through algorithmic computations and validation using the silhouette coefficient model, we summarize the composition characteristics of each category, demonstrating the rationality of the classification. Furthermore, fitting classification is applied to eight cultural relic samples, and the validity and sensitivity of the classification results are further verified through the silhouette coefficient model, ensuring the reliability of the classification.