(An open-ended problem). The one- and two-step ahead methods did not use the data as efficiently as possible, because the AR(m) model was built on the training set and never updated, as the model...

(An open-ended problem). The one- and two-step ahead methods did not use the data as efficiently as possible, because the AR(m) model was built on the training set and never updated, as the model moved forward through time. Assess whether the structure of the residuals changes over time, and revise the one-step ahead code to better model the noise. Does this improve the model fit? Explain.