When executing a causal forecast there are times when serial correlation or autocorrelation is detected using the Durbin-Watson measure of fit. In the help files it states, "High autocorrelation means that MLR using the ordinary least squares method is not a suitable forecasting technique for this data." The user I am working with currently uses another software package that performs a correction for this autocorrelation. The correction is done using an autoregressive error model to correct for autocorrelation, and the generalized autoregressive conditional heteroscedasticity (GARCH) model and its variants are used to model and correct for heteroscedasticity.
Do you know if there is the same ability in Demand Planning? We are on SCM 2007 currently.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.