%0 Journal Article
%T 结合LightGBM与SHAP的家政服务员离职预测方法研究<br>A Study on the Combined LightGBM and SHAP Approach for Predicting Domestic Helper Turnover
%A 刘峰涛
%A 刘雅琦
%A 王扶东
%J Service Science and Management
%P 37-47
%@ 2324-7916
%D 2023
%I Hans Publishing
%R 10.12677/SSEM.2023.121005
%X 针对互联网环境下家政服务员人力资源管理场景的变化，本文将LightGBM算法与SHAP模型结合，形成解决互联网背景下家政服务员离职问题的集成方法。以企业真实数据为研究对象，经数据预处理后建立LightGBM模型进行预测，并与KNN、逻辑回归、决策树、随机森林和GBDT算法对比，结果表明，LightGBM模型的准确率、F1值与AUC值分别为81.23%、84.41%和86.5%，优于其他算法。最终使用SHAP模型分析影响员工离职的重要因素，以此增强模型的可解释性，为企业管理者进行决策提供依据。<br />
In response to the changes in the human resource management scenario of domestic helpers in the Internet environment, this paper combines the LightGBM algorithm with the SHAP model to form an integrated approach to solve the problem of domestic helpers’ leaving in the Internet environ-ment. Using real data from enterprises as the research object, the LightGBM model was established for prediction after data pre-processing, and compared with KNN, Logistic Regression, Decision Tree, Random Forest and GBDT algorithm, and the results showed that the accuracy, F1 value and AUC value of LightGBM model were 81.23%, 84.41% and 86.5% respectively, which were better than other algorithms. Finally, the SHAP model is used to analyze the important factors influencing helper turnover, thus enhancing the interpretability of the model and providing a basis for corporate managers to make decisions.
%K 机器学习，LightGBM，SHAP，员工离职，Machine Learning
%K LightGBM
%K SHAP
%K Helper Turnover
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=60469