06-22-2017 03:12 PM
Hi, I am trying to create an automated variable selection program that will filter through about 5,000 variables (transformations and such).
My problem: I need durbin watson test statistics. I know I can do ODS Output DWTest for autoreg, but this has a tendency to slow the program tremendously and great impact performance to the point of it being unusable. It also causes a flurry of errors and is a serious impediment to operation.
If I do noprint, there is no ODS output. Is there a way of getting DW test results (or ADF for that matter) without using ODS output? There will be 5,000 or so times 7 regressions (for each segment). Looking at each test result is not important, storing it is. Using ODS makes it almost unworkable and while I can do it, I estimate to prevent crashes and overwhelming the results I will have to divide it up and make it take a long time.
06-22-2017 03:39 PM
06-22-2017 03:45 PM
I think you didn't read the link I posted. The last example shows how to do this, but here's another brute force approach that isn't recommended. Note only the PRINT results go to the listing output. You can just turn off the ODS destination. For better approaches read Ricks post in the link above.
ods _all_ close;; proc means data=sashelp.class nway n median mean min max; ods output summary=want; run; ods listing; proc print data=want; run;
06-22-2017 03:49 PM
Rick's last statement in that post: