A Levinson-Durbin Recursion for Autoregressive-Moving Average Processes
提出一种能按阶数递推计算自回归滑动平均模型参数的算法，推广了纯自回归情形的莱文森-德宾递推，有助于提高模型阶数自动选择等估计过程的计算效率。
We discuss an algorithm which allows for recursive-in-order calculation of the parameters of autoregressive-moving average processes. The proposed procedure generalizes the recursion of Levinson (1946) and Durbin (1960), which applies in the pure autoregressive case. We use ideas similar to the multivariate autoregressive case. Our results suggest how estimation procedures for autoregressive-moving average parameters, which include an automatic choice of model order, for example, the one proposed by Hannan & Rissanen (1982), may be made computationally more efficient.