中国社会消费品零售总额的时间序列分析  

Time Series Analysis of Total Retail Sales of Consumer Goods in China

作　　者：陈鹏蕾 连雨欣 王美 武业文[1] 

机构地区：[1]南京信息工程大学数学与统计学院,江苏 南京 [2]吉林建筑大学经济与管理学院,吉林 长春

出　　处：《运筹与模糊学》2023年第2期919-932,共14页Operations Research and Fuzziology

摘　　要：本文基于2000年1月~2022年12月的社会消费品零售总额数据,通过时序图和确定性因素分解初步了解该序列具有明显的季节性和趋势性。ARIMA模型为时间序列数据分析中最常用的模型,在对模型添加季节因素后可以很好地对具有该性质的数据进行建模,Holt-Winters三参数指数平滑模型中的指数包含了季节项和趋势项,也可以很好地分析和预测该数据,因此本论文采用ARIMA乘积季节模型和Holt-Winters三参数指数平滑模型对社会消费品零售总额数据进行建模,并根据建立的模型预测未来五个月(2022年8月~2022年12月)的社会消费品零售总额,探究不同模型在处理该类数据时的优劣。从结果来看,Holt-Winters三参数指数平滑模型优于ARIMA模型,我国短时间内社会消费品零售总额仍然将保持增长状态。Based on the total retail sales data of social consumer goods from January 2000 to December 2022, this paper preliminarily understands that the sequence has obvious seasonality and trend through time series diagram and deterministic factor decomposition. ARIMA model is the most used model in the analysis of time series data. After adding seasonal factors to the model, data with this nature can be well modeled. The index in the Holt-Winters three-parameter exponential smoothing model contains seasonal and trend items, which can also analyze and predict the data well, therefore, this paper uses ARIMA product seasonal model and Holt-Winters three-parameter exponential smoothing model to model the total retail sales of social consumer goods, and predicts the total retail sales of social consumer goods in the next five months (August 2022~December 2022) according to the established model, and explores the advantages and disadvantages of different models in processing such data. From the results, the Holt-Winters three-parameter exponential smoothing model is better than ARIMA model, and the total retail sales of consumer goods in China will still maintain growth in a short time.

关 键 词：ARIMA模型 Holt-Winters三参数指数平滑模 趋势预测 

分 类 号：F72[经济管理—产业经济]