云南民族大学学报（自然科学版）2025，Vol.34Issue(5)：597-610,14.DOI:10.3969/j.issn.1672-8513.2025.05.013
基于CTGAN和逻辑回归的企业员工流失预测及影响因素研究
Research on employee turnover prediction and influencing factors in enterprises based on CTGAN and logistic regression
摘要
Abstract
Market competition is becoming increasingly intense,and high employee turnover rates seriously impact the sustainable development of enterprises.This study focuses on analyzing the key factors influencing employee attrition and constructing a Conditional Tabular Generative Adversarial Network-Logistic Regression(CTGAN-LR)predictive model.First,an open-source human resources dataset from a certain company is selected and preprocessed.Due to the limited number of turnover samples,Conditional Tabular Generative Adversarial Network(CTGAN)is used for oversampling to address the issue of data imbalance.Secondly,employee turnover prediction is performed on both the original dataset and the balanced dataset using various machine learning algorithms,including logistic regression,decision trees,random forests,and gradient boosting trees.The results show that the CTGAN-LR model achieves the best performance across all metrics.Finally,the study examines the main factors influencing employee turnover and validates their significance through feature importance analysis and causal inference.At the same time,survival analysis provides companies with a dynamic perspective to help formulate more effective human resource management strategies.The findings offer empirical evidence for developing targeted employee retention strategies and emphasize the importance of enhancing employee satisfaction and optimizing compensation structures.关键词
员工流失/不平衡数据集/条件表格对抗生成网络/逻辑回归Key words
employee turnover/imbalanced dataset/conditional tabular generative adversarial network(CTGAN)/logistic regression分类
信息技术与安全科学引用本文复制引用
金艺鸥,王宁若,唐昊,王淼..基于CTGAN和逻辑回归的企业员工流失预测及影响因素研究[J].云南民族大学学报（自然科学版）,2025,34(5):597-610,14.基金项目
大连市社科项目(2024dlsky095) （2024dlsky095）
河南省科技攻关计划项目(242102211009) （242102211009）
河南省高等学校重点科研项目(24A520049). （24A520049）
