Bayesgil VAR Modelinin Gerçek Zaman Dizileri için Kestirim Amaçlı Kullanılması

Abstract

In this paper, it has been done the application of the forecasting for real time series using the Bayesian vector autoregressive (BVAR) that is improvised by Litterman [1,2]. So, for the data, the performance of forecasting for BVAR according to VAR and the univariate (Box-Jenkins) [3] model has been compared by the known measurement that is RMSE (root mean square error). Time series that are used for the analysis are the annual (1925-1999) series of the population, the export for every person, the import for every person and the ratio of GNP (gross national product) for export of Turkey. As a result of this study, it may said that the BVAR models can be used as a method to produce appropriate forecasts on time series that have different fluctuations.