A Multivariate Time Series Analysis of Some Flour Price Data
本文开发并比较了多种多元时间序列模型，用于分析美国三个地点的面粉价格指数，通过AIC准则评估不同模型拟合效果，探讨复杂建模方法是否优于简单方法。
In this paper we develop and compare several multivariate models for some multiple time series data. The data are indices of the price of flour at three sites in the USA and have been used for illustration in a recent methodological paper. The models all come from the vector autoregressive moving average class so that comparisons between them can easily be made using criteria such as Akaike's information criterion. It is particularly interesting to compare the models produced by relatively complicated model specification procedures with those developed by using more straightforward techniques to see whether we gain any worthwhile irnprovements in fit.