Is it possible to remove observations from an output created by a proc tabulate, by applying a condition on one of the aggrigative variables I've created during the proc tabulate?
for example, lets asy I am creating a variable that displays number of customers, and I want to remove all observations with less then 100 customers.
No, PROC TABULATE does not let you act on summarized statistics. You'll have to "pre-summarize" the data -- with PROC MEANS, PROC SQL, PROC REPORT, or even PROC TABULATE -- somehow to determine which rows are non-conforming. At that point, you can either eliminate the group to which the rows belong -- before your TABULATE -- or adjust the values for purposes of the report (like lump a bunch of small categories into an "OTHER" category). (If your elimination criteria is based on a simple count of obs for each group/category, you could even get that info out of PROC FREQ.)