| With the development of computer technology and the refinement of the social division of labor,enterprises have increasingly high requirements for the management of human resources.Employees have become the primary strategic capital for creating value,and the human resource management model has entered the era of digital human resource management.Human resource management departments need to transform their thinking from auxiliary and support functions to value creation and frontline thinking,and analytical and predictive capabilities are the core competencies of digital human resource management systems.This thesis aims to develop a human resource management system with predictive capabilities,the core of which is to use machine learning algorithms to establish an employee turnover prediction model,predict whether employees have a tendency to leave,and encrypt and decrypt employees’ identification numbers using SM2 to prevent the leakage of critical information.Firstly,data processing operations such as data integration,missing value analysis and filling,data replacement,data standardization,and feature selection were performed on the dataset.Secondly,by analyzing the balance of the training data and selecting appropriate classification evaluation indicators,preliminary employee turnover prediction models were established using three machine learning algorithms: random forest,XGBoost,and Light GBM.Then,the three machine learning algorithms were optimized for their parameters,and it found that the XGBoost model had the highest F1 score after hyperparameter optimization,and the AUC,balanced accuracy,and geometric mean of the optimized Light GBM model were superior to those of the random forest and XGBoost models.Finally,the optimized three machine learning algorithms were combined with the Stacking method to propose an employee turnover prediction model based on LXR-Stacking.Through the comparison of four evaluation indicators:F1 score,balanced accuracy,geometric mean,and AUC,it found that the employee turnover prediction model based on the LXR-Stacking fusion model outperformed other prediction models,with an F1 score,AUC,balanced accuracy,and geometric mean of 99.29%.Based on the LXR-Stacking fusion model,this thesis designed and implemented a human resource management system.SM2 algorithm is adopted to encrypt the ID number in order to protect the security of employee ID number information.In the process of system construction,open-source frameworks such as Spring Boot,Layui,Mybatis,My SQL,and ECharts were used.The system’s functions were tested to ensure the usability of the system after deployment.The system implements employee evaluation,auxiliary decision-making,and statistical analysis,and can provide employee turnover warning from multiple dimensions of employee data,ensuring the regular employment of the enterprise and providing decision support. |
