Analysis And Forecast Research On The Influencing Factors Of China’s Real Unemployment

With the expansion of colleges and universities and the aging of population,the attention of employment and related information is different in the past.In western developed countries and China,which is now a developing country,the employment problem is always worrying.If we want to quantify the employment situation,the commonly used indicator is the unemployment rate.If a country’s unemployment rate is too high,that is,the unemployment crisis will lead to instability and panic in all aspects of society.From the micro perspective,the high unemployment rate indicates that the country’s labor surplus or resources have not reached the optimal operation.Such an economy is a laggard and lifeless economy First of all,it will lead to the imbalance between supply and demand,which is bound to force government regulation and increase expenditure,resulting in a vicious circle;from a macro point of view,the unemployment rate exceeds a certain threshold,leading to the collapse of the whole system.With the popularity of the Internet,the gap between the rich and the poor is becoming more and more significant.In recent years,hatred for the rich is overwhelming.If we do not solve the unemployment problem from the root in time,we can reduce the loss Employment rate,social stability and harmony may be seriously damaged.In the new era,with the popularity of machine learning algorithms and the exponential improvement of computer computing power,"preparing for a rainy day is no longer a fantasy.Through objective massive data,scientific and effective algorithms and persistent training,the establishment of mathematical models can explain and predict the laws of our world.This paper aims to use a variety of machine learning prediction algorithms to predict the real unemployment rate in China,and finally choose a single method or a combination of methods with the highest accuracy to determine the final prediction model: the first is to use two benchmark models,the autoregressive lag(ARIMA)model and the unobservable component random volatility(UCSV)model,to make the initial prediction of the unemployment rate There are many indicators that affect the change of unemployment rate,and the indicators may have different degrees of multicollinearity and different time periods of data,and their importance is not the same.Based on the benchmark model,this paper uses shrinkage methods: ridge regression(RR),Lasso algorithm(lasso)and elastic network(elnet)to construct a penalty function to obtain a more refined prediction model,retaining the advantages of subset shrinkage Secondly,the mathematical model of factor analysis is used to extract the main factors and reduce the dimension,so as to grasp which indicators play a major role in the prediction of response variables.In order to improve the accuracy,some ensemble methods are used Methods: complete subset regression(CSR),bagging algorithm Methods)and random forest,combine several different machine learning algorithms and different parameters of an algorithm to complete the prediction of unemployment rate;finally,establish China’s unemployment prediction model,make a quantitative analysis of the current unemployment problem,and make reliable and credible policy recommendations according to the corresponding prediction variables.