Abstract

ARX model is considered as one of the important models in time series analysis, and there are several and serial steps for building this model, one from those steps is orders identification. In this paper a comparison has been conducted between four criteria: Akaike information criterion (AIC), Minimum description length (MDL), Hannan and Quinn information criterion (HQIC), and the median between MDL and HQIC and denoted (MMH). The comparison carried out in time domain by using simulation, standing to number of success for each criterion in true order selection of model and for four ARX models with different orders (1,1,1), (2,1,1), (1,2,1), (2,2,1) in different samples sizes. And to become obvious that the criteria MDL, HQIC, MMH efficient in order selection, but in general MDL criterion is the best one.

DOI:10.52113/6/2013-3-7/1-40