Analysis And Prediction Of China’s Employment Population Based On ARIMA-XGBoost Model

Employment is the biggest livelihood and the cornerstone of social stability.It is of great significance to economic development and even the prosperity of the country.China has always attached importance to employment and has achieved considerable results in the past few decades.However,we should still be aware that under the current environment,the employment situation is still severe,and the total number of employed people is a direct reflection of the employment situation.The study of the total number of employed people can provide strong support for the formulation of relevant measures and policies,and is of great significance for the proper solution of the employment problem.This thesis establishes three models for the prediction and analysis of the total number of employed people in China,evaluates and compares the effects of the models,and verifies the effectiveness of the fusion model constructed in this thesis.The first model is ARIMA(0,2,1)model based on the time series data of employment population,which can better fit the historical time series data.However,when predicting the future,it shows the characteristic that the prediction error increases significantly with the increase of prediction periods,and can only be used for short-term prediction.The second model takes the influencing factors of the employment population as the input characteristics,and establishes the XGBoost model for the total employment population.The model has good fitting effect in the training set,and the prediction error of the test set is lower than that of ARIMA model.In addition,its prediction error does not increase significantly with the increase of prediction periods,so it can achieve better performance in the long-term prediction.The third model not only takes the influencing factors of the employment population as the input characteristics,but also inputs the predicted value of ARIMA model into XGBoost model as a new feature.Finally,ARIMA-XGBoost fusion model is established for the total number of employed people.This model extracts the time series information of the employment population and the information of various influencing factors at the same time.It fits well on the training set,and the prediction error of the test set is smaller than that of any single model,it shows its good accuracy and robustness.Empirical analysis shows that ARIMA-XGBoost fusion model has better modeling effect than single ARIMA model or single XGBoost model in the prediction of employment population.It can be used to study the problem of employment population and provide a reliable reference for the proper solution of employment problem.