【摘要】 基于古代玻璃制品的化学成分数据分析其分类规律以及亚类划分方法.首先对原始数据进行中心对数比变换后分析风化和无风化数据的转换关系；然后，通过构建有监督特征选择方法探索高钾、铅钡类别规律，进而基于无监督特征选择方法探索亚类划分方法，重点构建了过滤式特征选择结合特征R型聚类和封装式特征迭代选择两种亚类划分方法，第二种方法可以给出多种划分方案和结果.更多还原

【Abstract】 Analyze the classification rules and subclass classification methods of ancient glass products based on their chemical composition data. Firstly, perform a central logarithmic ratio transformation on the original data, and then analyze the conversion relationship between weathered and unweathered data;. Secondly, by constructing a supervised feature selection method to explore the rules of high potassium and lead-barium categories. Thirdly, based on unsupervised feature selection methods, we explore subcategory partitioning methods. We have focused on constructing a subcategory partitioning method using filtered feature selection combined with feature R-type clustering, as well as a subcategory partitioning method based on wrapper feature iterative selection. The second method can provide multiple subcategory partitioning schemes and results.更多还原