Oke so I couldn't resist myself. Here is one way to go about it:   data have; input var1-var7; cards; 1 2 3 4 5 6 8 0.0 0.1 0.2 0.3 0.4 0.5 0.6 1.1 1.2 1.2 1.3 1.3 1.4 1.4 ; run; data want; if _n_=1 then set have(rename=(var1=day1 var2=day2 var3=day3 var4=day4 var5=day5 var6=day6 var7=day7)); array days{*} day1-day7; array var{*} var1-var7; do until(eod); set have end=eod; id+1; do i=1 to 7; day=days{i}; value=var{i}; output; end; end; keep id day value; run; proc print data=want; run; You can run proc reg by id on dataset "want" and have reg create an output dataset.   And again, there is no way this can be done on your dataset as it is. You need to rearrange it.   Hope this helps, - Jan. ... View more

Hi Nicholas,   You will have to normalize your data before trying to do anything with it. No proc will ever be able to make sense of the current shape it's in. Regression is performed on values in rows, not columns. It is one of the common mistakes described in the excellent article "3 common messy data problems and how to tidy them in SAS".   Use a datastep to combine two consecutive rows and create an identifier for that pair of rows. Take care to rename the variables on the two reads so not to overwrite values. Then output every variable pair (from row1-var1, row2-var1 etc) plus the identifier to a new dataset. Then you can run your regression on that dataset using a BY on the identifier. If row 1 is not repeated (please provide sample data) then retain that first row and loop over the rest. It could in fact make the datastep easier. But we'd have to know. Please provide sample data before I can have a shot at coding.   And I do assume that what you call a SAS data sheet is in fact a SAS dataset.   Regards, - Jan. ... View more

What have you tried sofar and what happened. Can you provide logs?   If your code works outside of EG you could paste it as-is into a program node inside EG and run it from there.   Keep in mind: depending on your EG configuration your code inside EG may run on a different server and the files may not exist on that machine. Copy them over if that's the case. Work with your IT staff/SAS support people if needed.   Hope this helps, - Jan. ... View more

Instinctively I'd go with @ballardw and say id wouldn't make sense to support DELETE in FCMP. And indeed using it results in a run-time (not compile time!) error:   ERROR: Parsing error generating code. However, I tried to find support for this in the docs and couldn't find it. In fact, the 9.1 pdf doc of FCMP states that DELETE is in fact supported. So go figure.   Anyways the use case of @KafeelBasha makes it a trivial problem. Which does not preclude the existence of use cases that make it desirable to be able to delete observations.   - Jan.   ... View more

You have eloquently phrased exactly my gripes. Did you not want to add that the SAS Support site is slow as a dog and very unforgiving when it comes to typos in the search phrase?   - Jan. ... View more

Hi,   Short of using SOUNDEX (a really horrible approximation of matching words on how they sound) I think this is good case for the SAS Data Managment Advanced suite. The matching capabilities of the (formerly known as) Dataflux data quality products are very advanced and wil be your best bet to solve this.  And yes, if you do not have licenced it yet you will have to part with a bit of cash. But that *should* be oke if the benefits outweigh the cost.   Regards, - Jan. ... View more