摘要
本文主要研究了古代玻璃制品的成分分析与鉴别问题。选取相关数据,对缺失值处理和异常数据剔除后,进行定类与定类之间关系分析,选择差异性分析中的卡方检验判断玻璃表面风化与纹饰、颜色、类型之间的关系,再使用多独立样本Kruskal-Wallis和单因素方差分析进行显著性差异判断,得到相应的统计规律。并依据风化点检测数据,建立基于参数检验和非参数检验的化学成分预测模型,并将数据合并处理后采用随机森林对高钾玻璃和铅钡玻璃分类规律进行分析,根据玻璃利用类别,利用主成分分析法进行玻璃的亚类划分。再根据得到的载荷矩阵热力图结合因子载荷象限,并进行降维处理,得到最终的划分结果。
This article mainly studies the composition analysis and identification of ancient glass products.After dealing with the missing data and removing the abnormal data,the relationship between definite and definite data was analyzed.The cubic test in the difference analysis was used to determine the relationship between the glass surface and the decoration,color and type.After that,Kruskal-Wallis with many independent samples and the single factor analysis of deviation were used to make a significant difference judgment and get the corresponding statistics.According to the data of the weathered point,a prediction model of chemical composition based on the data of the paramedics and the non paramedics was established.After the data was combined and processed,a random forest was used to analyze the classification rules of the high K glass and the lead-ore glass.According to the type of glass used,the secondary division of glass was carried out by the main element analysis method.Then,according to the heat map of the obtained load array and the factors in the load quarters,the dimension was reduced and the final division result was obtained.
作者
肖锦娟
庞金祥
陈文卓
Xiao Jinjuan;Pang Jinxiang;Chen Wenzhuo(School of Shipping and Ship Engineering,Chongqing Jiaotong University,Chongqing,China)
出处
《科学技术创新》
2023年第14期37-40,共4页
Scientific and Technological Innovation
关键词
古代玻璃
随机森林
卡方检验
化学成分
参数检验
ancient glass
random forest
square test
chemical composition
testing of data
