Forecasting consumption of cereals in Greece: an application of univariate ARIMA and transfer function models
1995
Ncib, S.
The scope of this study is essentially concentrated on building a forecasting model for consumption of and consumer expenditure for cereals in Greece, using time series data. Two classes of models were used: the univariate Box-Jenkins ARIMA model which has been shown to be accurate and flexible since the predicted value seems to satisfy the properties of ARIMA forecast, and the Transfer Function Model which describes the dynamic structure between output and input. The latter procedure was carried out for forecasting consumer expenditure for cereals and fulfilled in two small sections. The first is by using a single input variable (price of cereals) identified by prewhitening cross correlation functions. The fitted model is discovered to be less accurate. The second is by adding national disposable income as a second input to the first one. This methodology, unlike the previous one, was identified by a regression method which has a wide application in multi-input transfer function models and is based on the corner method. The evidence of comparative criteria of model selection, to come out with the perfected multi-input transfer function model in the present case study
显示更多 [+] 显示较少 [-]