Comparison of Machine Learning Models for Employee Turnover Prediction
Qing Yin · Applied and Computational Engineering · 2023
In the era of big data, companies are able to more easily record, analyze, and utilize data, making it feasible to predict employee turnover. In the current economic situation, how to predict employee turnover has become particularly important. The core of this article is the establishment of a forecasting model based on real data, and then through the application to reflect the use and importance of the forecasting model. This paper uses XGBoost, LGBM, random forest and decision tree models, uses IBM's real data to construct prediction data, and evaluates the models to select the model that is most suitable for the lost data. Additionally, the role of Synthetic Minority Over-sampling Technique (SMOTE) in improving classification accuracy was explored. The results of this work indicate that: 1) balancing the data set using SMOTE oversampling technique is more effective than using the original data set; and 2) The XGBoost algorithm performs better, with higher precision and greater flexibility.