摘要
文章根据ARIMA时间序列模型和BP神经网络模型分别在处理线性空间预测问题和非线性空间预过测对问北题京中市的居优民势消,费建价立格了指一数种(C基PI于)B序P列神的经实网证络分误析差证校明正了的该差组分合自预回测归模移型动相平对均(于A单RI一MA预)测组模合型预在测C模PI型预,通测中的有效性,并利用该模型预测了未来一段时间北京市CPI的走势。
Based on the advantages of ARIMA time series model and BP neural network model in dealing with linear space prediction problems and nonlinear space prediction problems respectively, this paper establishes a differential autoregressive mov-ing average (ARIMA) combined prediction model based on BP neural network error correction. And then through the empirical analysis of the Beijing consumer price index (CPI) series, the paper proves the validity of the combined forecast model compared with the single forecast model in CPI forecast, and also uses this model to predict the CPI trend of Beijing for a period of time in the future.
作者
吴晓峰
杨颖梅
陈垚彤
Wu Xiaofeng;Yang Yingmei;Chen Yaotong(Key Laboratory of Big Data Decision-making for Green Development of Beijing,Beijing Information Science and Technology University, Beijing 100192, China)
出处
《统计与决策》
CSSCI
北大核心
2019年第15期65-68,共4页
Statistics & Decision
基金
北京市教委科研计划一般项目(SM201611232002)
科研基地建设—绿色发展大数据决策北京市重点实验室专项建设经费资助(71F1810916)
