基于BO-XGBoost的中小微企业综合质量动态画像方法研究
The Dynamic Portrait Method of Comprehensive Quality for Micro,Medium,and Small En-terprises Based on BO-XGBoost
周海霞 1曹丽娜 2陈进东3
作者信息
- 1. 北京信息科技大学经济管理学院,北京 100192
- 2. 华北科技学院经济管理学院,河北廊坊 065201
- 3. 北京信息科技大学经济管理学院,北京 100192;智能决策与大数据应用北京市国际科技合作基地,北京 100192
- 折叠
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.
Abstract
[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.
关键词
动态画像/中小微企业/综合质量/BO-XGBoost/综合质量服务平台Key words
dynamic portrait/MSMEs/comprehensive quality/BO-XGBoost/service platform of comprehensive quality引用本文复制引用
基金项目
国家重点研发计划(2019YFB1405303)
北京市市属高等学校优秀青年人才培育计划(BPHR202203233)
出版年
2024
NETL
NSTL
万方数据