Well, there may be a more elegant way to do it, but you could use multiple arrays, two at a time.  The first would hold the 1 or 0 for the category and the second would hold the value to be averaged.  As you loop through the 1/0 array, the vale of the second array is added to an accumulator, and then at the end of the iteration for a given month, the average is computed.  Subsequent categories would be computed in like manner, all results being stored in a results array.   Something like the below.  See results below code.  Is this along the lines of what you're looking for? DATA Have; FORMAT Date YYMMD7.; INFILE DATALINES DSD DLM='09'X MISSOVER; INPUT Date : ANYDTDTE11. Cat1 Cat2 Cat3 Cat4 Cat5 Cat6 Cat7 Value; DATALINES; Jan-18-2021 1 0 0 0 1 1 0 0 Jan-18-2021 0 1 1 0 0 1 1 0 Jan-18-2021 0 1 0 1 0 1 1 1 Jan-18-2021 1 1 0 0 1 0 0 Jan-18-2021 1 0 0 0 0 0 0 1 Jan-18-2021 0 1 0 1 0 0 1 0 Jan-18-2021 0 1 0 1 1 0 1 1 Jan-18-2021 0 0 1 0 0 1 1 1 Jan-18-2021 1 1 1 0 0 0 1 1 Jan-18-2021 1 0 0 0 1 1 0 1 Jan-18-2021 0 0 0 1 0 1 1 1 Feb-18-2021 1 1 1 0 1 0 1 0 Feb-18-2021 0 1 0 1 0 0 0 0 Feb-18-2021 1 0 1 0 0 1 0 1 Feb-18-2021 0 0 1 0 1 1 0 1 Feb-18-2021 1 1 0 0 1 0 1 1 Feb-18-2021 1 0 0 1 0 0 1 1 Feb-18-2021 0 1 1 1 0 1 0 0 Feb-18-2021 1 1 0 1 1 1 1 0 Mar-18-2021 0 1 0 1 1 0 0 0 Mar-18-2021 1 1 0 1 0 1 0 0 Mar-18-2021 1 1 1 0 0 1 0 1 Mar-18-2021 0 1 0 0 0 1 1 0 Mar-18-2021 1 1 1 0 1 1 0 1 Mar-18-2021 1 1 1 1 1 0 1 1 Mar-18-2021 1 0 0 1 0 1 0 0 Mar-18-2021 1 1 0 0 0 0 0 0 Mar-18-2021 1 1 1 1 1 1 0 1 Mar-18-2021 0 1 0 1 1 1 1 1 Mar-18-2021 0 1 1 0 0 0 0 0 Mar-18-2021 0 1 1 1 0 0 0 0 Mar-18-2021 0 1 1 1 1 1 1 0 Mar-18-2021 0 0 1 1 1 1 1 0 Mar-18-2021 1 0 0 1 0 1 1 1 ; RUN; DATA Want; DROP _: Cat1 - Cat7 Value; SET Have; BY Date NOTSORTED; ARRAY Cats {*} Cat1 - Cat7; ARRAY Sums {7} _TEMPORARY_; ARRAY Rates {7} Rate1 - Rate7; FORMAT Rate1 - Rate7 PERCENT10.2; DO _i = 1 TO DIM(Cats); IF Cats{_i} THEN Sums{_i} + Value; END; IF Last.Date THEN DO; DO _i = 1 TO DIM(Cats); Rates{_i} = Sums{_i} / 7; CALL MISSING(Sums{_i}); END; OUTPUT; END; ELSE DO; DELETE; END; RUN;   Results:   Jim ... View more

@Dodecaphonic wrote:  Are you by any chance aware of any documented analysis / benchmarks that would assess the overhead? I'm not sure how much benchmarks run on another system would tell you about the impact in your particular environment.  In practice, here, it hasn't made a material difference.   What you might do is run a SAS batch job multiple times, half of the time with -logparm "rollover=session write=immediate" And the other half with  -logparm "rollover=session write=buffered" After you complete maybe three runs of each type, compare the run times.  If it lengthens the run time by, say, 10%, then no way; it's not worth it.  But I think it's going to be less than 5%, maybe a couple of minutes extra in a 1 hour job.  Immaterial, really. Regarding directives such as #Y-#m-#d_#H-#M-#s and the like, I've found those effective from the command line as well as via RSUBMIT (and SYSTASK too if I recall correctly), but I'm not sure how they'd work if you're using the management console.  Can't hurt to try.   If you can use macro code in whatever part of the process defines the dataset and file names, then of course you can do something like &Year.-&Mon.-&Day._&Hour.-&Min.-&Sec or whatever you like.   Jim ... View more

First:  Are all those versions of the array found in cc_iv_sv?  I suspect only Dyn is.  If so, it's probably best to define the other arrays as _TEMPORARY_ and not give them individual field names.   Second:  I don't see anything wrong immediately.  I can't test it because of course I don't have access to cc_iv_sv.  The only thing I can recommend is that you use the Debug option of the Data step and follow the code as it executes.  If you have Enterprise Guide you would click on the little bug to enable debugging (if it is not already enabled).  You have to have at least EG 7.1.3 I think (maybe it's 7.1.5) in order to use the EG debugger.   If you're using Display Manager ("PC SAS"), then you would add /Debug as part of your Data statement like so: data want / debug; If you're using SAS Studio, it's going to be roughly similar to Enterprise Guide, but I don't have access to SAS Studio at the moment, so I can't give you any instruction there.   If you're not familiar with the debugger, you can just Google something like "SAS debugger commands" and you should be able to get started.   Sorry I can't be of more assistance here, but I can't access cc_iv_sv, so there's not much more I can do.   Jim ... View more

@binhle50,   Would it be possible for you to use the SAS code icon when you post your program?  It is very hard to read otherwise.   Jim ... View more

@ChrisNZ,   Interesting technique using a "dummy" datalines (cards) to generate a test file.  Kind of handy.  I had not seen that before. Also an interesting technique using the INPUT with @'literal'.  I didn't even think of that although I have (rarely) seen it used before.   I noticed that you used a comma as a separator.  I believe that if there were additional data beyond the name that you would pull in that data.  For example:     I modified it to a space, and it worked as coded: data T; infile cards truncover; input @; do until(LASTOBS); set SASHELP.CLASS end=LASTOBS; LENGTH NM $32; _INFILE_=catt('age:', AGE, ' name:', upcase(NAME), ' sex:', UPCASE(SEX)); PUTLOG "NOTE: " _INFILE_; input @1 @'name:' NM $ @; output; end; keep NAME NM; LENGTH NM $32; cards; dummy run; Alternatively, one could specify a comma delimiter. INFILE Cards DSD DLM=','; Yes, sorry.  Fussing over small details, I know, but I like to know how things work.  I was curious about the comma and wondered if I understood the syntax of input @1 @'name:' NM $ @; so I just had to play with it.  🙂   Jim     ... View more

You've got the right idea using the COMPRESS function with 'kd' as the third parameter.  Then you would want to use the INPUT function (function not statement) to cast the data into numeric form.   Like this: data have; input a $20.; datalines; CYCLE 15 DAY 22 CYCLE 33 DAY 22 ; data want; set have; b = INPUT(TRANWRD(strip(COMPBL(compress(a,,'kd'))),' ', '.'), 8.2); run; Which yields:   Are you sure that's what you want?  It almost seems like cycle and day should be two separate fields instead of a single numeric fields with two values, in essence, separated by a decimal point.   Jim ... View more

@binhle50 wrote: I want variable_want=1 for all lines 4,12,19. OK, so like this, yes? The code I posted should already be doing this.  Is it not working for you?  If it is not working, did you change the program?  If so, I would need to see the actual code that you are running in order to help you.  I would also want to see the log.   Is the data you are running different?  If so, can you post the data here that is giving you the incorrect values?   Jim ... View more

Thank you.   What is the difference between Line 4 and Line 12?  In Line 4 we want "1", but in Line 12, we want -9999.  Is it because for X4T39JHGK there is additional data whereas for APJV74TMX there is not?   In other words, is this what you want?     Jim ... View more

@Reeza wrote:  Mainframes aren't really used anymore Interesting statement.  🙂  In a lot of large corporations, mainframes are still used for OLTP (think ATM networks, airline reservations, etc.).  A large university might have one even if only to run legacy systems on.     In analytical environments, I think mainframe users are few and far between unless perhaps the mainframe is there for some other reason.   Jim ... View more

Probably your laptop will work. I would certainly try it before going to another platform.  5 million isn't that big, but of course size is dependent on how many columns and the size of the columns.   Make sure you have OPTIONS COMPRESS=YES in effect if your columns are mainly character or COMPRESS=BINARY if your columns are mostly numeric -- at least that's the standard advice.  In actual practice I've found that sometimes binary compression is more space efficient than character compression even if there are a lot of character columns present.  To really know, one has to experiment, but for now, just making sure that you are using compression should go a long way towards efficient space allocation.   Jim ... View more

@binhle50 wrote: And Jim, only switch -9999 to 1 for all following year, not previous years. I got everything worked except this. Thanks so much for your help! Isn't that the behavior of the program now?  I must not be understanding.   In the below, in which line numbers would the value of Variable_Want change?  And what value would Variable_Want change to?       Jim       ... View more

I don't understand. What ate you asking? If you are getting errors, please post your log. Jim ... View more