Durbin-Levinson Algorithm

To estimate this conditional correlation measure, a backward recursion (backward forecast) is applied, the so-called Durbin-Levinson-algorithm. Again, if the PACF value for a specific lag exceeds a critical bound, a significant AR-like correlation is to be... 

Although it is possible to solve this system of equations directly, computationally, it can be better to solve them using an iterative approach. The most common such method is the Durbin-Levinson algorithm (see Sect. 5.4.4.1 for an explanation of this method). 

The Durbin-Levinson algorithm is a useful iterative method for inverting and hence solving the prediction equations given by Eq. (5.100) for a large data set. The procedure can be summarised as follows ... 

The Durbin-Levinson Algorithm can be used to also compute the Yule-Walker parameters and the partial autocorrelation function, since both have a similar matrix form. 

Example 5.14 Using the Durbin-Levinson Algorithm to Obtain the Predictor... 

Use the Durbin-Levinson algorithm to obtain the one-step-ahead predictors for the case where there are 1,2, and 3 past data points. 

---