Your code added a zero, so your middle range is between 1M and 2M. Also, you mixed the conditions around for middle. 1M will never be greater than 2M. Use simple code that you understand such as: and sales >= 200000 then ... View more

Seriously?  You're subsetting on: where t2.email is null; But then you want to also check: lowcase(t1.email)=lowcase(t2.email That's really the same as saying: where t1.email is null and t2.email is null; This looks like code that was written to confuse someone rather than a real attempt at writing a program.  Can you explain in words what you are trying to make happen? ... View more

If the date always appears at the beginning of CONCAT, subsetting is easy: data want; set have; if concat =: "202406" then delete; run; If 202406 might appear in the middle of CONCAT instead of the beginning, it is only slightly more complex. Which is it? ... View more

This type of question has actually been around for 40 years, ever since the WHERE statement was invented.  Either interpretation seems reasonable: Consider observations 5 through 15, and select only the males Consider only the males, and of those take the 5th through the 15th In fact, originally SAS couldn't decide which interpretation to use.  For a year or two, using WHERE when OBS/FIRSTOBS was in effect would generate an error message.  Finally SAS decided that the second interpretation would be the more useful one. ... View more

Years ago, I wrote a DATA step to find all combinations without using any functions.  I had to stop after about 25 values, because the program took so long to run.  Let me try to explain why. Each value  you add doubles the number of possible combinations.  That value can either be omitted, giving you the same number of  combinations you had without the value, or included, giving you another equally sized set of combinations.  So with 100 values, you can have 2**100 combinations (I guess you could subtract one because you don't need to include the empty set of all values omitted.) If you had a machine that could process 1M values per second, it could handle 20 values easily since 2**20 is roughly 1M.  Once you get up to 40 values, you would need 1M seconds (almost 12 days if I did the math right).  If you wrote a program to process 100 values, you would not live long enough to have it process every combination.   ... View more

Presumably, you would like to have someone read your post and explain what happened. For that to happen, you need to post the log from your program. ... View more

There are at least two errors that make me question whether you have posted your actual code. There's a missing semicolon before the SET statement after data calc_lsd. And there are many missing percent signs. For there is no such thing as a mend statement. It should be %mend; instead. And there id no Rotation statement in SAS. it should be %rotation. ... View more

You're missing the easy way. For a variable that is always 1 or 0, there's an easy way to calculate the percentage of times it is 1. Simply take the mean of the variable, something which tabulate can do easily. ... View more

Nothing takes a long time to run.  Your data set is small.   There might be a way to write a shorter program that doesn't list all the names.  But you would have to explain the structure that you have and the structure that you want.  That's about the same amount of work as using a RETAIN statement.  But if you spell out the before and after structure, I can probably give you a short cut. ... View more

If these are the only file names that begin with "A2024" then the task is easy and does not require macros.  For example: data want; set A2024: indsname = month; run; If some version of MONTH is already part of each data set,  you can skip the INDSNAME option. ... View more

The posters you have been working with all understand the process perfectly. My only contribution here will be to simplify the process to highlight answers to some of the questions you ask. Study the results and see what makes sense. data have; do cust =1 to 5; do obs =1 to 8; n + 1; output; end; end; run; That gives you a simple data set you can play with; So here is one way to play. data want; do until (last.cust); put _all_ "before"; set have; by cust; put _all_ "after"; end; run; Hope this helps rather than confuses. ... View more

An interesting approach, which would work with a slight tweak:   data want; merge have (where=(credit^=.)) have (drop=credit where=(debit^=.)); by accountid month; run;   It also suggests another possibility:   data want; update have (where=(credit^=.)) have; by accountid month; run;     ... View more

Given that  your data set is sorted this would be one way: proc summary data=have; var credit debit min_pay; by accountid month; out=want (drop=_: ) max=; run; ... View more

You're correct about what +1 does. But we have no way to know whether this is a smart thing to do. We can't see the data or the purpose of the program. ... View more