人口预测的投影寻踪自回归模型构建与实证  

Construction and Empirical Study of Projection Pursuit Autoregressive Model of Population Prediction

作　　者：于晓虹 徐海燕[1] 楼文高[3] Yu Xiaohong;Xu Haiyan;Lou Wengao(School of Economics and Management,Nanjing University of Aeronautics and Astronautics,Nanjing 211106,China;School of Humanities,Shanghai Business School,Shanghai 200235,China;Shanghai Lixin University of Accounting and Finance,Shanghai 201209,China)

机构地区：[1]南京航空航天大学经济与管理学院,南京211106 [2]上海商学院人文学院,上海200235 [3]上海立信会计金融学院,上海201209

出　　处：《统计与决策》2022年第23期38-42,共5页Statistics & Decision

基　　金：国家社会科学基金资助项目(16BXW031)。

摘　　要：文章根据全国和上海市1980—2020年两种完全不同人口变化规律的时间序列数据,采用自回归分析法,确定了3个预测因子,分别建立投影寻踪自回归(PPAR)模型,建模样本的数据拟合精度和验证样本的预测精度均显著高于GM、Logistic模型,完全能够满足人口预测的需要。实证结果表明:全国人口2011—2020年验证样本的平均绝对误差MAE和平均绝对百分比误差MAPE分别为160.0万人和0.116%,最大相对误差仅为0.26%;2021—2030年的人口预测结果表明,2021—2027年,人口增速越来越慢,并于2027年出现人口达峰,达峰人口为14.15亿人。上海市人口2016—2020年验证样本的MAE和MAPE分别为6.9万人和0.28%,最大相对误差仅为0.53%;2021—2027年人口增速越来越慢,并于2027年达峰,达峰人口为2499.9万人。According to the time series data of two completely different population change rules in the whole country and Shanghai from 1980 to 2020, this paper uses the autoregressive analysis method to determine three predictors and establishes the projection pursuit autoregressive(PPAR) model respectively. The data fitting accuracy of the modeling sample and the prediction accuracy of the verification sample are significantly higher than that of the GM and Logistic models, which can fully meet the needs of population prediction. The empirical results show that the mean absolute error(MAE) and mean absolute percentage error(MAPE) of the sample in China from 2011 to 2020 are 1.6 million and 0.116%, respectively, and that the maximum relative error is only 0.26%.The population prediction results from 2021 to 2030 are shown as below: The population growth rate will be slower and slower from 2021 to 2027, and the population will peak in 2027 with 1.415 billion people. The MAE and MAPE of the verified samples of Shanghai population from 2016 to 2020 were 69,000 and 0.28%, respectively, and the maximum relative error is only0.53%.From 2021 to 2027, the population growth rate will be slower and slower, and will peak in 2027 with a population of 24.999million.

关 键 词：人口预测 投影寻踪自回归模型 时间序列数据 人口达峰 

分 类 号：F224.9[经济管理—国民经济] C924.24[社会学—人口学]