时间序列模型选择的灰色关联法及其应用

为了解决传统的模型选择准则在实际应用中的不足,提出一种时序模型选择的灰色关联法,在传统的选择准则的信息量基础上再包括序列平均值、标准差、峰度、偏度等统计参数构建模型评价指标体系,运用灰色关联分析法对模拟序列进行选择。结果表明此法能对多种适用模型作进一步筛选,这对于经济时间序列建立模型来说,有一定的意义。应用实例证明该法可行性和有效性。

Grey relation method of model selection in time series and its application

To resolve the shortcoming of traditional model selection criteria in practical application,a grey relational method about model selection in time series is presented.An evaluation index system based on traditional model selection criteria information and including statistical parameters such as mean,standard deviation,kurtosis,skewness is constructed to select simulated time series by applying grey relational method.The results show that this method can select further for a variety of appropriate model,which is certain significant for economic time series to build model.An application example is given to illustrate and verify the feasibility and effeitness of the proposed method.