Chinese Population Prediction Based On Lasso-FGM Model And Improved GM(1,1)Model

In line with the data of China Statistical Yearbook,we can see that the number of Chinese population has revealed a declining trend in recent years.Given that China’s population development is currently affected by domestic and foreign challenges and the novel coronavirus pneumonia.So that better grasp the population development trend,scientific and accurate population prediction has certain practical significance for the formulation of population development decisions.In order to achieve accurate prediction of the population of China,this article explores the various factors affecting the population of China.Based on the Lasso model,Lars and Glmnet methods are called to screen variables.By comparing the results and searching related literature,the main factors affecting the population of China are determined by the Lasso model of Lars.Furthermore,this paper introduces the multivariable fractional grey model to forecast the population of China,determines the differential order by comparing the error of the predicted data of different orders from 2005 to 2017,and puts the data of 2018-2020 into the model for verification,in order to make a more accurate model,and then predicts the population in the next ten years.The prediction results show that by 2030,China’s total population will drop to 1348,3740,which is a certain gap from the number predicted by the national population planning policy.For the calculation of developmental gray number and endogenous control gray number,the traditional GM(1,1)model adopts the least square method to fit the development trend of data with a linear model.However,for data points with volatility,there is no way to simply use a linear model to fit,otherwise the fitting error will increase significantly.However,non-parametric kernel regression can better solve this situation.It can fit a line suitable for the overall trend,and also deal with smoothing problem well.Based on the population data,R language is used for empirical analysis.Through fitting and data comparison,the results show that the improved model fits well,so as to predict the future population of China.