I need to build up a model to predict sales using several variabls (including info about date, promo, competition, etc.). How can I combine the pattern in date (time series) and other input variables in SAS miner?
Another question is about partitioning the data into training and validation. How can I partition the data while preserving the patterns of time series? I've seen some articles on SAS Miner website. Some articles say I can do that only by using SAS code (which is a little difficult for me to understant). Others say I can do that with the partition node's user-defined sampling. But I do not see "user-defined" but just "cluster" sampling. Are they the same?
Hope those experts could help me out of this. Thanks!