SAS and time series

[noetsi]

If anyone knows any good material on this, I have the documentation from PROC Autoreg and the books SAS publishes on time series, I would love to hear about it.

I have spent weeks looking for good material with limited success. Something as basic as how you interpret slopes with time series, I know it is different than regular cross section data, and how you select which lags of a variable to use is something I have not found - let alone the associated code.

 

[hlsmith]

Let me know when you find that magical book. I need to figure out time series some dauseuseIs part of the problem that SAS probably has multiple procedures that you can use?

 

[noetsi]

I keep ordering ones that look good, but when they show up they are all equations with little explanation in English. Not good for a beginner. I have spent several years reading it, so I know the basics, but I need to move beyond that point.

SAS has lots of PROCS for this and I have wasted a good deal of times on ones of limited value (at least for me) such as the Almon lag.

Selecting lags for SAS is complex at best.

 

[hlsmith]

Is it like selecting covariance structures?

 

[noetsi]

I don't know anything about that - although that might be behind some of the comments on it. It is tied to theory about how many periods ahead a given predictor influences Y. Covariance structures are not raised. Alternately you can estimate all lags and use tricks like the Almon lag to calculate the results.

 

[hlsmith]

Yeah, they seem to be associated. If you think about autoregressive covariance structure, etc.