Department of Statistics, University of Ibadan, Ibadan, Nigeria
Ojo, J.F., Department of Statistics, University of Ibadan, Ibadan, Nigeria
Full autoregressive moving average model are always characterize by many parameters and this is a problem. Some of these parameters are always close to zero and there is the need to eliminate these parameters and this can be done through subsetting. In this paper subset autoregressive moving average models is considered using the proposed algorithm and as well compare with subset autoregressive models. The autoregressive models of order p and autoregressive moving average models of order p and q from which we can have various subsets is represented respectively thus: Xt = φ1Xt-1 + φ 2Xt-2 + φpXt-p + εt 1 (1 - a1 β - a2 β 2 - ⋯ - ap βp) Yt = (1 + b1 β + b2 β 2 + ⋯ + bp βp) εt (2) The best model among the subsets autoregressive models and subsets autoregressive moving average models is selected using the Akaike Information Criteria. The best models in the subsets autoregressive models and the subsets autoregressive moving average models are compared using residual variance. Results revealed that the residual variance attached to the subset autoregressive moving average model is smaller than the residual variance attached to the subset autoregressive model. We conclude that the subset autoregressive moving average model perform better than subset autoregressive model. © EuroJournals Publishing, Inc. 2007.
