Superb code . I was stuck since two days due to this issue. Should have come to this answer earlier .   Thanks a lot . Made my day.   Good day wishes. ... View more

Note that SAS does not have a "BEST" informat. If you do specify BEST32. then SAS will just use the normal 32. informat which can handle all of the various formats that the BEST format can generated.  (It will also handle scienfitic notation done using D instead of E to separated the value form the magnitude).  If you want to read values that can contain commas, dollar signs or parenthesis then you can use the COMMA informat (you can also call this the DOLLAR informat).     So in many ways the COMMA informat is the "BEST" informat since it can read anything that the default informat can read plus it handles values with commas and dollar signs.   Note that if your values are using commas for decimal places and periods for thousands separators then you can use the COMMAX informat to reverse the meaning of those two characters. ... View more

thanks @Patrick! I'm not sure I understand the proc sql part of your answer, but yes that is essentially what I ended up doing! I did a data step merge to output the records from conc_dados that did not merge with conc_tx_base and then I did a proc sql left join to merge the two datasets.   thank you! ... View more

If you don't use a BY statement then the data step doesn't have any reason to care if the data is sorted or not.   One other imapct that a BY statement can have is that if the input data set has an index for the BY variables then SAS will use the index to read the data in the order specified even if it is not physically sorted in that order.  Similarly if you are reading from an external database, like ORACLE or Teradata, then SAS can ask the database to sort the records before passing them to your data step.   You can mimic the calculation of the FIRST. variable of the BY statement in the NOTSORTED mode by using the LAG() function.  Try this little program. data x ; set sashelp.class ; by sex notsorted ; f1= first.sex ; f2= sex ne lag(sex) ; run; proc print; var name sex f1 f2 ; run; ... View more

You could alse try proc tabulate (assuming variable values are stored as sas date)   data test; input dt date9.; datalines; 01jul2009 01may2007 . 01dec2010 ; proc tabulate data=test format=date9.; var dt; table dt=''*(min='min date' max='max date') ; run;   min date max date 01MAY2007 01DEC2010 ... View more

The license your school offers will be a better solution. It will likely have SAS/GRAPH, ETS and a different interface.  For example, one area where SAS UE is weak is on the SAS/ETS module. Not all the time series procs are included. SAS Studio is what's used in UE and your University will likely offer the course in SAS Base or EG. If it's Base it'll be fine but EG is significantly different.     The ideal situation would be to pay the $50 and get the SAS version. You'll also have to pay for a Windows License for the Virtual Machine to run SAS. Hopefully your University should have good offers from Microsoft because the $249 for a Windows license is high, SAS requires Windows Pro the last time I checked.  VMWare would then be about $40 which means you're looking at 3-400. At that point you may as well buy a Windows computer.  ... View more

@Rick_SAS thank you for those resources. That is very helpful!  ... View more

Thought you might (at cost of a little complexity) have reduced the code to Calc_date_accident = date_accident - not ( weekday( date_accident ) in( 1, 7 ) ) ; Since the results of SAS logical operations are either 0 or 1, the composite above achieves what appears to be wanted.... I'm just interested in what activity requires this adjustment for Monday to Friday but not for Saturday and Sunday....? ... View more

@confused_saser wrote: @data_null__ what do you mean do the test for missing? I'm not sure I understand what that means In your code you have a test for missing if tsi_days < 0 then tsi_exclude=1; else if tsi_days=. then tsi_exclude=0; else tsi_exclude=0; I'm saying do that first and then do the test for TSI_DAYS < 0. ... View more

If you don't have a big table, you also could try SQL. data key; input k $20.; cards; Butrans Codeine Dalmacol Demerol ; run; data have; input id $20.; cards; BBButrans Ceine Dcol Demerolll ; run; proc sql; select *,case when exists(select * from have,key where id=a.id and upcase(id) contains strip(upcase(k))) then 1 else 0 end as opioid from have as a; quit; ... View more

Your code works regardless of the log warning and as @ballardw states the log message is a warning only and the option allows you to switch it off.   ... View more