10-31-2011 10:22 AM
If you had to forecast quarterly data, what SAS procedures would you start with? The data is somewhat seasonal meaning that it has peaks and then returns to a steady state but the time between peaks is random and the steady state that it returns to may differ from previous steady states. The quarterly data that I am trying to forecast is relatively new (8 years or so, with most activity 2008 on.)
We use proc arima to forecast other things here, but those things have a lot more history and they follow regular seasons. To use arima I would need 12 years of good data and it's just not there for this project.
What else should I look into?