Prediction and Classification of Ancient Glass Types Based on Logistic Regression Models

For the prediction and classification of the chemical composition of ancient glasses, we provide a brief analysis of the weathering mechanisms of high-potassium glass and lead-barium glass presenting descriptive statistics of their chemical composition before and after differentiation. In addition, the content of the chemical composition of the glass prior to differentiation is predicted. Logistic regression models are established to classify the glass artifacts separately, and systematic clustering models based on the elbow rule are established to cluster the samples. Finally, the experimental results confirm the efficacy of the logistic regression model in predicting the chemical composition of ancient glasses, demonstrating its effectiveness in the classification and clustering of glass artifacts.