During the ancient Silk Road period, glass played a significant role in witnessing cultural integration. However, glass was highly susceptible to environmental and weathering effects. This article aims to explore the changes in elements that occur during the weathering process of glass and propose a method to identify and classify glass based on corresponding characteristics. To begin, an in-depth examination and classification of the components of ancient glass artifacts were conducted. Logistic regression models and ensemble learning techniques, specifically classification tree ensemble learning, a machine learning algorithm, were utilized to improve the understanding of the factors influencing glass properties. These methods enabled the training and optimization of two different types of ancient glass. Additionally, sensitivity analysis was carried out, revealing the significant impact of barium content on ancient glass. Finally, examples of the two glass types were analyzed, and the predicted results from the models were compared. This process led to the determination of an optimal classification model that exhibits excellent applicability, accuracy, and simplicity. The research presents innovative ideas for the identification and authentication of cultural relics such as glass.