文档介绍：居民消费价格指数的季节ARIMA模型及短期预测
摘要
本文是以居民消费价格指数(CPI)的短期预测为主线,采用定量的时间序列分析方法,建立季节自回归综合移动平均模型(季节ARIMA模型)对CPI进行定量分析,并用图、文、数相结合的方式来揭示经济变量CPI与时间变量之间的变动规律。
本文的研究思路是运用季节ARIMA模型, 并提供使用这类模型进行建模及预报的一般过程,即:对数据进行平稳性检验、平稳化处理、模型的差分阶数识别、参数估计,建立时间序列模型,并对模型进行检验,确定较适合季节自回归综合移动平均模型,然后再进行预测分析。
本文基于时间序列分析,对以1990为基期的我国CPI定基比数据进行了实证分析(时间长度为从1990年到2007年的216个月),—2009年CPI作出预测并与实际值比较,结果表明相对误差均在1%之内,, 季节ARIMA模型在短期内的预测结果是可以接受的, 平均相对误差e=, , 预测的误差将逐渐增大, 反映在预测置信区间的宽度上, 表现于随着预测期的延长, 预测置信区间的宽度越大, 因此, 该模型适合短期预测。
但是,、利率水平、汇率水平、货币供应量等内在因素的影响, 而这些突然变化的因素在季节ARIMA模型中只能以随机扰动项来表示, , 利用季节ARIMA模型进行预测时应和实际情况结合综合考虑和分析。
关键词:CPI;时间序列分析;季节ARIMA模型;数量研究
Seasonal ARIMA Model for CPI Short Forecasting
ABSTRACT
This article,adopting quantitative time series analysis approaches,establishing the Seasonal Auto-regressive Integrated Moving Average Model (Seasonal ARIMA) for quantitative analysis of the Consumer Price Index(CPI) and using graphs,characters and algebra to reveal variation law between the CPI and the time variable, concentrates on the short-term forecasting of the CPI.
Research ideas of this article not only use the Seasonal ARIMA model but also provide the use of such models the general process of modeling and prediction,namely:stability test of the data,establishing the time series models,and testing the model to find out the more suitable parameters for Seasonal ARIMA model,then the forecast.
Based on time series analysis,this paper makes the empirical analysis(length of time from 1990 to 2007,216 months total) on the basis of the fixed base ratio CPI data whose base time is in 1990,and ultimately determine the more suitable Seasonal ARIMA model ARIMA(0,2,2)(0,1,1).Through using this model to forecast CPI for 2008-2009 two years pared with the actual values,results show that the relative error is within 1%,thus the prediction model is