@ChrisNZ :  True. The reason I've mentioned the regexen's purported efficiency is that once on SAS-L someone (unfortunately, don't remember who or the circumstances) showed a regex - which I then tested - that worked much faster than the combos of string functions anyone else (including me) had offered. Could be a fluke ... but got seared in my memory nonetheless.   ... View more

@u13161556U Actually you can make it 3-4 lines shorter and easier to read. Edited code is as below.   DATA Intermediate; SET HAVE; IF Balance > 0 THEN OUTPUT; RUN; DATA Want (DROP=StartBalance Balance); SET Intermediate; BY Account_Num; RETAIN StartBalance 0; IF FIRST.Account_Num THEN StartBalance=Balance; IF LAST.Account_Num THEN DO; Difference=Balance - StartBalance; OUTPUT; END; RUN; ... View more

If the 'obs=max' post solved your issue, please mark that message as the solution. So that others know exactly message resolved your issue, if they encounter it as well. Thanks ... View more

Thanks! I'm having difficulties setting -LOGPARM "WRITE=IMMEDIATE" because the program is launched as a job in the SASJobExecution (so it's like a batch but there isn't a real command line where I can put parameters) and inside I redirect the log to a file with proc printto. How to set the LOGPARM using SASJobExecution? ... View more

Based on the way you expect the output to appear: Without having each employee ID spread out horizontally with the count of records for each, i just want the distinct total of employees per department with records.   It doesn't sounds like you want a two-way frequency table.   Using this dummy data: DATA WORK.HAVE; FORMAT Dept $20. Employee $20.; INPUT Dept Employee; INFILE DATALINES DLM='|' DSD; DATALINES; Human Resources|Adam Apple Accounting|Bill Bobington Accounting|Chuck Cooper Accounting|Donna Dillard Human Resources|Edward Ekins Sales|Fred Flintstone ; I used this PROC FREQ logic to generate a list of all of the departments and a distinct count of employees for each department. PROC FREQ DATA=WORK.HAVE; TABLES Dept /LIST MISSING NOPERCENT NOCUM; RUN; And the resulting output that printed was this: Dept Frequency Accounting 3 Human Resources 2 Sales 1 A list of all of the departments from the dataset and a distinct count of the total employees from that department.   If this isn't what you were looking for, please provide some additional details (like dummy data that aligns to your actual dataset, a visual representation of what your expectations are for the final output, etc).   Hope this helps.   ... View more

@tsap : Yup, that is it. ... View more

To achieve the outcome you requested on my own system, I created a libref for the Teradata database connection and then executed a 'PROC CONTENTS' datastep:   LIBNAME Tera TERADATA AUTHDOMAIN="TeradataAuth" TDPID=APROD MODE=Teradata; PROC CONTENTS DATA=Tera.Hello; RUN; This is what I mocked up from the information provided in your post.   Hope this helps. ... View more

There are a couple of things that confuse me in your post: 1. Does the ID_Payment not play as a factor in the sorting order?           - There are two payments between '14JUN2019' and '17JUN2019', then nine payments between '18JUN2019' and               '01DEC2019'.           - Since we are calculating a running calculation for how much debt was payed off at what point and what debt was remaining,             wouldn't we need to sort across ID_Payment? (assuming that wasn't just a random number, which by looking at it, does not               appear to be random).   2. If I am reading this correctly it appears as though the field, 'Payment_Sum' is the amount paid in that specific transaction.           - If that is the purpose of that field, why would we want the first observation in your results to show 'payed' as 0 if the             'payment_sum' was 59,079.99?           - I would think that the first observation would populated 'payed' with 59,079.99 and then the following 'to_pay'(remaining debt)                would be 920.01   Understanding these items a little better might help me understand what needs to occur to accomplish the results you need. ... View more