基于CART和聚类分析的古代玻璃分类预测模型研究
Classification and Prediction Model for Ancient Glass Based on CART and Cluster Analysis
邵光明 1夏贤齐 1殷何杰1
作者信息
- 1. 安徽中医药大学(安徽 合肥 230012)
- 折叠
摘要
该文针对古代玻璃文物的化学成分,借助CART决策树、聚类分析和主成分分析方法对其进行分析和预测.通过对不同特征进行系统和K-means聚类,结果表明将玻璃制品分为高钾类和铅钡类两大类和三个亚类为最佳.利用主成分分析法对其化学成分分析发现,当特征根选为前六个时,效果最好,对解释原有变量贡献率较大.根据以上结果构建CART决策树预测未知玻璃制品类型,预测准确率较高.
Abstract
This paper focuses on the chemical composition of ancient glass relics and uses CART decision tree,cluster analysis,and principal component analysis methods to conduct a series of analysis and predic-tion.Based on conducting systematic and K-means clustering on different features,the results show that it is the best to classify glass products into two major categories of high potassium and lead barium with three subcategories.Using principal component analysis to analyze chemical components,it was found that when the feature roots were selected as the first six ones,the effect was the best,and the contribution rate of explaining the original variables was relatively large.Based on the above results,a CART decision tree is constructed to predict unknown types of glass products,with high prediction accuracy.
关键词
CART决策树/K-means聚类/聚类分析/主成分分析Key words
CART decision tree/K-means clustering/cluster analysis/principal component analysis引用本文复制引用
基金项目
安徽省高校重点项目(KJ2019A0438)
安徽省高校重点项目(SK2020A0255)
出版年
2024
NETL
NSTL
维普
万方数据