Classification of Ancient Glass Based on Logistic Regression and K-means Clustering Algorithm

Abstract

Ancient glass is easy to weather, after a large number of internal and external elements exchange, its composition ratio will change, thus affecting the correct judgment of the type of glass. In order to classify the types of glass, this paper first preprocessed the data, analyzed the classification rules of high-potassium glass and lead-barium glass by using logistic regression, and got a classification model that could identify the types of glass according to the content of chemical components. Next, K-means clustering algorithm is adopted to divide lead, barium and high potassium into A and B respectively according to the content of silica. After that, the rationality test was carried out on the clustering results. The test index was the belongingness of the sample to its category, and the clustering results passed the rationality test. Finally, the sensitivity analysis of the clustering results was carried out.