基于ARIMA-RF组合模型的CPI预测  

CPI Prediction Based on ARIMA-RF Combined Model

作　　者：曾令麒 ZENG Lingqi(School of Mathematical Sciences,South China Normal University,Guangzhou 510631,China)

机构地区：[1]华南师范大学数学科学学院,广东广州510631

出　　处：《现代信息科技》2023年第13期13-17,共5页Modern Information Technology

摘　　要：居民消费价格指数(CPI)是一个重要的宏观经济变量,反映了国家的通货膨胀水平、居民的消费水平和生活成本,它与国家、社会和个人有密切的联系。基于Savitzky-Golay平滑滤波去噪后的2002年1月至2021年12月的CPI月度数据,构建ARIMA-RF组合模型,对CPI序列进行预测并与单一的ARIMA和RF模型进行比较。结果表明,ARIMA-RF组合模型的预测效果和稳定性均优于单一模型。Consumer price index(CPI)is an important macroeconomic variable,which reflects the inflation level of the country,the consumption level of residents and the cost of living.It is closely related to the country,society and individuals.Based on the monthly CPI data from January 2002 to December 2021 after Savitzky-Golay smooth filtering and denoising,an ARIMA-RF combined model is constructed to predict the CPI sequence and compare it with a single ARIMA and RF model.The results show that the prediction effect and stability of ARIMA-RF combined model are better than that of single model.

关 键 词：CPI Savitzky-Golay平滑滤波 ARIMA 随机森林 ARIMA-RF 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TP18[自动化与计算机技术—计算机科学与技术]