双回归（AR—R）预测模型之新探——及其在中国人口中短期预测上的应用

构造一种称为双回归的时间序列预测新方法。本文作者利用与因变量自身相关性紧密的因变量前几期取值,综合前一次的自变量构建了双回归(时间序列自回归-多元线性回归)预测组合模型。这种方法克服了以往时间序列预测只是自身拓展而不考虑多项因素(变量)的不足,也弥补了回归分析预测法必须已知同时期各个自变量值才能预测的缺陷。并用这种新方法对中国的人口总量进行预测,预测效果比较理想。

Research on Double Regression Model: Its Application is Short-term Prediction of China's Population

The author establishes a new method to predict time-series which is named as double regression.Furthermore,the author applies frontal values of dependent variable(its' autocorrelation value is high) and independent variable values in order to establish double regression model.This theory overcomes such disadvantages that time-series prediction only involves dependent variable but not independent variables and makes up such disadvantages that multi-regression must know the same period's independent variables to predict.This new theory is used to predict China's total population,with relatively ideal prediction results.