• Journal of the Korean Statistical Society
• /
• 제26권1호
• /
• pp.47-56
• /
• 1997

The asymptotic distribution of the extended Yule-Walker estimates of a mixed ARMA process is derived. Also, a recursive algorithm is derived to calculate the extended Yule-Walker estimates recursively, which is a generalization of the Levinson-Durbin algorithm.

• Communications for Statistical Applications and Methods
• /
• 제3권3호
• /
• pp.47-58
• /
• 1996

The generalized partial autocorrelation (GPAC) method of Woodward and Gray (1981) and the 3-pattern method of Choi (1991) have been used for identifying ARMA processes. The methods are based on the extended Yule-Walker equations. The purpose of this paper is to show the 3-pattern method is superior to the GPAC method through theoretical analysis and computer simulations.

• Journal of the Korean Statistical Society
• /
• 제16권1호
• /
• pp.1-9
• /
• 1987

Recently a new form of the extended Yule-Walker equations for a mixed autoregressive moving-average process of orders p and q has been proposed. It can be used to obtain p+q+1 parameter values from the first p+q+1 autocovariance terms. The autoregressive part of the equations is linear and can be easily solved. In contrast the moving-average part is composed of nonlinear simultaneous equations. Thus some iterative algorithms are necessary to solve them. The iterative algorithm presented by Choi(1986) is very simple but its convergence has not been proved yet. In this paper a Newton-Raphson solution for the moving-average parameters is presented and its convergence is shown. Also numerical example illustrate the performance of the algorithm.