By group summary of distributed values in the variable

[webart999ARM]

Hello SAS community,

I came up with a quick and simple question at first glance.

But I am struggling with optimal solution.

Here is the dataset (part of it)

I have 5 groups (colvar).

My question is, how I can summarize the variable 'n' by group, summing from ">=21" to ">=3", so that for last observation here I would have the summary of all the distributed values above it? (e.g. new 'sum' variable for obs=7 would be  393 (9+42+56+80+98+75+33)).

I have  tried something like this

data want;
	retain sum1 0;
	set adsl5_2;
		by colvar;
		sum1=sum1 + n;
run;

But it is processing whole dataset and not by group

Thank you in advance for any help.

Sincerely

Artur

[PeterClemmensen · Accepted solution]

Try this. Here, we set sum1 back to zero when we hit a new by-group. 

data want;
   set adsl5_2;
   by colvar;
   if first.colvar then sum1 = 0;
   sum1 + n;
run;

. 

[webart999ARM]

Thank you.

[PaigeMiller]

You are using the word "summary" and "sum" as if they are interchangeable and mean the same thing. I wouldn't do that, but assuming that is what you intended, then this should work:

data want;
    set have;
    by colvar;
    if first.colvar then summary=0;
    summary+n;
run;

[webart999ARM]

This one works as well, thank you.
And I agree with your suggestion about "summary" and "sum", indeed, they are different things.