多元回归与年龄移算法在老龄人口研究中的整合分析  

The Meta-analysis of Multiple Regression and Age Shifting Algorithm in the Aging Population Research

作　　者：吴启凡[1] 

机构地区：[1]河海大学公共管理学院,南京211100

出　　处：《西北人口》2015年第4期6-10,共5页Northwest Population Journal

基　　金：江苏人口老龄化对经济社会发展影响及其对策研究;江苏省高校哲学社会科学基金重大项目(项目编号:2010ZDAXM004);社会养老服务体制机制创新研究--基于江苏苏南;苏中;苏北的调查与比较;教育部规划基金项目(批准号:13YJA840008)

摘　　要：我国人口老龄化问题日趋明显,现阶段对人口老龄化的模型研究依然存在问题,在对我国人口老龄化情况的研究过程中,单纯运用多元回归的方法需考虑多重共线性问题,为避免此问题则要优选变量,但在逐步回归过程中又会将对其可能造成显著性影响的偏相关扰动项忽略,而且单纯运用回归模型进行预测将在长时间序列中造成较大误差,为此,结合年龄移算法对回归因子进行单项细度预测,再运用回归方程进行宏观计算,将大幅提高预测的精度。本文以男性人口、女性人口、城市人口、乡村人口等因素进行动态研究,先根据相关性分析,初步筛选影响因素,再通过多元线性回归找到人口老龄化与人口结构中相关因素的数量关系,这里通过逐步回归出恰好出现了偏相关扰动项无法接受检验的情况,我们运用两种标准化方法结合Mann-Whitney U检验进行验证分析,最终运用年龄移算模型和回归矩阵预测人口老龄化发展趋势,并根据预测结果进行相关分析,给出相应评价。The problem of our aging population has become more evident,the model for the study of population aging is still a problem at this stage. In the case of China＇s aging population of the study,the issues of using a simple method（multiple regression multicollinearity） is to be considered,To avoid this problem may lead to the Multicollinearity,however they will be the likely cause of a significant impact which can be easily ignored. And use the simple regression model to predict the result in the long sequence may also give rise to more errors,so we need to combined with age-shift algorithm to return the individual factors fineness forecast,then use the macro regression equation to calculate,which will significantly improve the prediction accuracy. In this paper,According to correlation analysis,initial screening factors from the male population,the female population,urban population, rural population dynamics. Then find the relationship between population aging and population structure of the related factors by multivariate linear regression. We use two kinds of standardized methods combined with MannWhitney U test to make the confirmatory analysis. Finaliy use the age shift operator model and regression matrix to predict the development trend of population aging, and give the analysis and evaluation by the forcasting results.

关 键 词：多元线性回归 Mann-Whitney U检验 年龄移算模型 

分 类 号：C921.2[社会学—人口学]