基于残差修正GM(1,N)的成都市社会消费品零售额预测  

Retail Sales Forecast of Consumer Goods in Chengdu Based on Residual Correction GM(1,N)

作　　者：王洪平 Wang Hongping

机构地区：[1]华中师范大学工商管理学院

出　　处：《阿坝师范学院学报》2018年第4期56-62,共7页Journal of Aba Teachers University

基　　金：国家自然科学基金面上项目"机器学习与信息融合技术研究"(41271197)

摘　　要：通过介绍灰色GM(1,N)模型基本原理和改进措施,提出残差修正GM(1,N)模型。运用残差修正GM(1,N)模型对成都市社会消费品零售额进行预测。平均预测误差为10. 66997%,比非残差修正的灰色GM(1,N)模型的平均预测误差14. 60739%减小了26. 95499%,近期预测误差更小,2017年预测误差仅为1. 76773%。研究还发现,对成都市社会消费品零售额影响由大到小的因素为城镇化率、城镇居民人均可支配收入、地区人口数量、农村居民人均纯收入。The residual correction GM(1,N)model is proposed by introducing the basic principle and improvement measures of grey GM(1,N)model.It was used to predict the retail sales volume of social consumer goods in Chengdu where the average prediction error was 10.66997%,which was 26.95499%lower than that of the non-residual modified grey GM(1,N)model-14.60739%.Actually,the short-term prediction error is even smaller when the prediction error in 2017 is only 1.76773%.The study also finds that the factors influencing retail sales of consumer goods in Chengdu are urbanization rate,per capita disposable income of urban residents,regional population and per capita net income of rural residents.

关 键 词：成都市 社会消费品零售总额 预测 GM(1 N) 残差修正 

分 类 号：F724.2[经济管理—产业经济]