What I very often do is exactly what I demonstrated. By allocating the file using the "temp" directive, it gets allocated in the Work directory (slightly different on z/OS, but that's another story). And the file will automatically be deleted when freed. ... View more

Here is an SQL solution: proc sql; create table want as select distinct Patid, advevent,date from have group by Patid, date having count(distinct advevent)>1; quit; ... View more

Most of the time, I use a datastep which writes code, and then a %INCLUDE statement to execute the code.   I prefer it to CALL EXECUTE because when you are developing, you can take a look at the code you have generated before submitting it, and when you have many steps in the generated code, you can submit them one at a time, to see if the results are as expected before going on.   E.g.: filename tempsas temp; data _null_; file tempsas; /* here are the statements to write the code */ run; %include tempsas; ... View more

The short answer is: there is not. If you want the probabilities for each result, try something like this: data prob; array prob(2:12) prob2-prob12; do d1=1 to 6; do d2=1 to 6; prob(d1+d2)+1/36; end; end; drop d1 d2; run; But if you just want the probability for each throw, it is 1/(6**2)=1/36. I do not think there is a special function for that. ... View more

I assumed that you could just look for non-whitespace, but of course your SQL expression may look like e.g. "a.col1=dsn.col4&a.col3...", without whitespace. So you may want to use "\w" (word character) instead of "\S" (non whitespace). You could also check for the first character being non-numeric, but is that really necessary, if your input is valid SQL code? ... View more

The macro variable max_dt will contain a string like 12MAR2019. To use that as a date constant in SAS, use "&max_dt"d, so you get a date constant, like "12MAR2019"d.   ... View more

Pearl Regular Expressions are probably the easiest, e.g.: data test; text="AB.name = dsn.pname and AB.ID = dsn.varID and AB.custid=dsn.customerid"; prxid=prxparse('/(?<=\bdsn\.)\S*/i'); start=1; pos=0; len=0; do until(0); call prxnext(prxid,start,-1,text,pos,len); if len=0 then leave; outvar=substr(text,pos,len); output; end; run; An explanation of the search string in PRXPARSE: "(?<=\bdsn\.)" is a look-behind assertion: it has to be there but is not included in the output. The meaning is a word boundary (\b) plus the text "dsn." ("." has to be escaped with a "\"). After that comes "\S*", meaning zero or more non whitespace characters, which is the string you are looking for. the final "i" just means that the search is case independent, the search will find "DSN." as well as "dsn.". ... View more

You forgot a colon after SVY_SENT_DT, that's all! ... View more

I could not find the routine for calculating geohashes. But you can get the result from geohash.org, e.g.: data have; /* just som random test data */ input latitude longitude; cards; 45.63 114.67 18.876 11.44 ;run; filename geohash url; data want; set have; fname=cats('http://geohash.org/?q=',latitude,',',longitude,'&format=url&redirect=0'); infile geohash url filevar=fname truncover; input url $40.; /* the result is like http://geohash.org/c216ne */ geohash=scan(url,-1,'/'); run; ... View more

Another possibility is to use SORT and MERGE: proc sort data=lib1.table1; by id; run; proc sort data=lib2.table1; by id; run; data want; merge lib2.table1 lib1.table1; by id; run;  Note that I put lib2.table1 before lib1.table1 on the MERGE statement; that way, the data from LIB1 overwrites the data from LIB2 if both are present.   If you want to do the same with more tables, just remember to put the tables with highest priority last in the MERGE statement. ... View more