06-07-2012 04:58 AM
I am going to use Proc UCM for a series of sales in order to forecast them in future periods.
Before that i would like to validate my model (s) by telling SAS not to use the last n observations in the data set when creating the model, so that SAS would forecast them and therefore i could compare the actual vs the forecasted.
So i was wondering which option of UCM could help me with this validation?
Thanks in advance