Abstract
This article will discuss China in the future the population quantity forecast and the appraisal question. Based on certain supposition, establishes the mathematical model, makes the forecast to the population, is advantageous in the construction harmonious society.
This article from short, neutral carries on the forecast for a long time separately, has used three different algorithms. In the short-term predict that uses Malthus and the Logistic differential equation population model. A Yuan linear regression model, can carry on the short-term forecast quite precisely. Finally forecast separately to these three kind of models the data and the official data carry on the comparison. In the intermediate stage forecast that we have used in the gray forecast model GM(1,1) model, this model by “partial informations known, partial informations unknown” “the small sample”, “the poor information” the uncertainty system is the object of study, although the information content are quite few, but to the short and medium term forecast that also has the good effect. But the long-term forecast uses in the multi-dimensional linear regression model the stepwise regression method to carry on the predict that and has considered the birth rate and the mortality rate to the population quantity influence. Finally we also use the multivariable auto-adapted return transect, its algorithm is similar for the model construction strategy to the forward gradually linear regression, this model fits the data excessively, for this reason uses a backward delete procedure. The multivariable auto-adapted return transect through selects the cross different year section to realize the short intermediate stage predict that the effect is very good, tallies with China's population growth tendency, because the data are few, the algorithm cannot realize the long-term forecast.
model improvement: The differential equation model must obtain the better forecast precision, should divide the crowd, for instance a minute age group, the city towns and villages group, the sex group and so on, the differential treatment, considers the respective general character. In the multi-dimensional linear regression model's stepwise regression method may consider more variables.