Research On The Prediction Model Of Employee Turnover

Talents,especially excellent talents,as the core part of an enterprise,play a key role in the stable development of the enterprise.In the information age,the enterprise platform has accumulated a large number of employee data,which can be used to analyze the factors that affect employee turnover,and select appropriate prediction methods to predict employee turnover,which has certain guiding significance for enterprises to take measures in advance to prevent brain drain.First of all,this paper analyzes the original data visually,and obtains several factors that affect employee turnover.It uses one-hot coding and sub box processing to preprocess the classification features and continuous features respectively,and reprocesses the features with strong correlation through the correlation coefficient matrix.On this basis,it uses recursive feature elimination method to choose features,and combines the results of visual analysis,the experimental data set needed to build the model is determined.This paper studies the prediction model,selects the random forest model,lightgbm model and logistic regression model to predict employee turnover,and adjusts the parameters.By comparing the AUC values of the models under different parameters,the single model with the best prediction effect is selected as the final single prediction model,and the three single prediction models after adjustment are modeled with the determined experimental data set.The experimental results show that the single prediction model of logical regression is the best,but there are still some differences among the three models.Therefore,the model can be further combined to improve the accuracy of prediction and AUC value.This paper studies the combined forecasting model,and designes SRLL model based on stacking combination strategy and voting model,uses the combined model to predict employee turnover,and compares the accuracy rate and AUC of all models.The results show that the prediction effect of the two combination prediction models is better,and the prediction effect of the SRLL combination model designed according to the stacking combination strategy is best.Therefore,the SRLL prediction model is used as the final employee turnover prediction model.