摘要
【目的/意义】为了准确刻画中小微企业质量动态变化、提升综合质量服务水平,开展中小微企业综合质量动态画像方法研究。【方法/过程】从综合质量概念出发,结合综合质量服务内容和中小微企业特点,设计画像维度;基于聚类等方法进行静态标签提取,并提出基于贝叶斯优化的XGBoost(BO-XGBoost)模型画像标签分类算法,实现中小微企业综合质量画像标签自动提取与动态更新;构建中小微企业综合质量画像标签体系,开展动态画像方法的应用研究。【结果/结论】通过与逻辑回归、随机森林、KNN等方法比较,基于BO-XGBoost的综合质量动态画像方法的标签分类准确率达到95.71%,模型整体性能最优。【创新/局限】本文提出面向综合质量服务平台的动态画像方法与流程,能够高效分析中小微企业综合质量特点,提升综合质量服务平台服务效率。
【Purpose/significance】In order to accurately depict the quality of medium,small,and micro enterprises(MSMEs)and im⁃prove the comprehensive quality service level,the comprehensive quality dynamic portrait method research of MSMEs is carried out.【Method/process】Based on the concept of comprehensive quality,combined with the content of quality services and the current situa⁃tion of MSMEs,design the portrait dimensions;The static labels of enterprises are extracted based on clustering and other methods,a portrait label classification algorithm based on Bayesian optimized XGBoost(BO-XGBoost)model is proposed,and the automatic ex⁃traction and dynamic update of the comprehensive quality portrait labels of MSMEs are realized;Finally,a comprehensive quality por⁃trait label system for MSMEs is constructed,and the application research of dynamic portrait algorithm is implemented.【Result/con⁃clusion】Compared with Logistic Regression,Random Forest,KNN etc.,the prediction accuracy of BO-XGBoost reaches 95.71%,and the overall performance of BO-XGBoost is the best.【Innovation/limitation】The dynamic portrait algorithm and process for the quality service platform in this paper can effectively analyze the comprehensive quality characteristics of MSMEs,and improve service effi⁃ciency of the comprehensive quality service platform.
作者
周海霞
曹丽娜
陈进东
ZHOU Haixia;CAO Lina;CHEN Jindong(School of Economics and Management,Beijing Information Science and Technology University,Beijing 100192,China;School of Economics and Management,North China Institute of Science and Technology,Langfang 065201,China;Beijing International Science and Technology Cooperation Base of Intelligent Decision and Big Data Application,Beijing 100192,China)
出处
《情报科学》
CSSCI
北大核心
2024年第3期89-99,共11页
Information Science
基金
国家重点研发计划课题“基于数据驱动的中小微企业综合质量精准服务技术研究”(2019YFB1405303)
北京市属高等学校优秀青年人才培育计划项目“中小微企业综合质量智能服务与优化技术研究”(BPHR202203233)。
