基于变系数模型的人口增长率分析(最终).doc

摘 要 人口预测是一个非常重要的理论和实践问题，人口的合理建模是人口预测、控制与管理的基础工作，是当前人口研究的热点问题之一. 为此，准确的预测人口信息，是制定未来人口发展目标和生育政策等有关人口政策的基础，对研究制定地方经济和社会发展决策具有重要的参考价值. 本文主要研究山东省人口自然增长率，分析预测其未来趋势，并对与之有关的因素进行分析.传统的线性模型不能反映人口数据中所存在的非线性特征，变系数模型不仅能克服线性模型的不足，而且能够克服高维数据“维数祸根”的问题. 本文分析方法如下： 首先，利用变系数模型局部线性拟合方法来估计每一年的模型参数的数值，在这一过程中，要用到最优带宽，通过交叉证实法来确定. 其次，利用得到的参数值求得每一年的人口自然增长率的预测值，由预测值与实际值之间的相对误差及二者关系图都可以看出，变系数模型对人口自然增长率的拟合效果是非常好的，并且根据变系数模型具有解释性这一特点得出各变量对人口自然增长率的影响大小. 最后，利用变系数局部线性估计方法对2007年的人口自然增长率进行预测，预测值3.9367‰,实际值3.58‰.可见，变系数模型对被解释变量的预测效果也是很好的. 关键词：变系数回归模型，带宽，局部线性估计，交叉证实法，人口自然增长率 ABSTRACT Population forecast is a very important theoretical and practical issue, a reasonable modeling of population is a basic work of population forecast, control and management, is one of the hot issues of the current population study. To this end, projecting the information on population accurately is the basic work for making the population policy, such as the goal of the future population development and reproductive policy, is also of great reference value of making the local economic and social development information. This paper studies the natural population growth rate of Shandong Province, analysis and forecast its future trends and the related factors for analysis. The traditional linear model can not reflect the non-linear characteristics which exist in the demographic data, variable coefficient model can not only overcome the the inadequacy of linear models, and be able to overcome the high-dimensional data ‘dimension curse’ problem. This paper is as follows: First of all, the use of local linear variable coefficient model fitting method to estimate model parameters for each year of the value, in this process, the optimal value of bandwidth is determined through cross-validation. Secondly, use the parameter values which have been obtained to estimate each year’s predictive value of natural population growth rate, by the relative error between predictive value and the actual value and the relationship