The eof marker on the sashelp.class dataset is being reached before the second observation in the have dataset can be processed, causing the data step to terminate prematurely. You could try the following code instead, but I don’t see how this would be any more efficient than the SQL Cartesian product. [pre] data have; input x y; cards; 1 2 3 4 ; data want; set have; do i=1 to n; set sashelp.class nobs=n point=i; output; end; run; [/pre] ... View more

Maybe I’ve misunderstood your objective, but wouldn’t it be a lot easier to just use the TRANWRD function? ... View more

Here is a way to do it with a data step. [pre] data _null_; if 0 then set dataset1 nobs=n; array char{*} _character_; array num{*} _numeric_; vars = dim(char) + dim(num); put "Note: The dataset has " n "observations and " vars "variables."; stop; run; [/pre] ... View more

As KSharp pointed out, the biggest problem with SQL is maintaining the sort order. If this is important, and SET2 is small enough to load into memory, consider using a multidata hash object solution. [pre] data set1; input Hit Z zn BSi$ PRP$; datalines; 5 100 23 L Out 5 100 23 L Hit 5 100 23 L Out 5 100 23 L Hit 5 100 23 L Out 14 87.5 23 R Out 14 87.5 23 R Hit 14 87.5 23 R Hit 14 87.5 23 R Out run; data set2; input zn BSi$ PRP$ fbp Percent; datalines; 23 L Hit 3 .8 23 L Hit 4 .1 23 L Hit 9 .1 23 R Hit 3 .6 23 R Hit 4 .3 23 R Hit 9 .1 run; data set3(drop=rc next); set set1; if _n_=1 then do; if 0 then set set2; declare hash set2(dataset:'set2', multidata:'YES'); set2.definekey('zn', 'BSi', 'PRP'); set2.definedata('fbp', 'Percent'); set2.definedone(); end; rc=set2.find(); if rc ne 0 then do; call missing(fbp, Percent); output; end; else do; output; set2.has_next(result:next); do while (next ne 0 ); set2.find_next(); output; set2.has_next(result:next); end; end; run; [/pre] ... View more

You could put the PROC OPTLOAD in your autoexec file so that your desired options are loaded whenever you start SAS. ... View more

I’m somewhat perplexed why the solutions which were already proposed did not work. One additional alternative is to set all your options the way you want them, use PROC OPTSAVE to save them, and then use PROC OPTLOAD any time you want to reset your options. [pre] proc optsave out=sasuser.myoptions; run; proc optload data=sasuser.myoptions; run; [/pre] ... View more

DBailey: No, if you use MODIFY the data step approach does not necessarily require sorted files. That is not to say that using modify would be a more efficient solution than using a hash as a lookup. That probably depends on the size of the master and transaction datasets, whether or not the datasets are already sorted or indexed, and many other factors. Here is some information from the documentation on using the MODIFY statement with a BY statement. See http://support.sas.com/documentation/cdl/en/lrdict/64316/HTML/default/viewer.htm#a000173361.htm • If you use a BY statement with a MODIFY statement, MODIFY works much like the UPDATE statement, except that o neither the master data set nor the transaction data set needs to be sorted or indexed. (The BY statement that is used with MODIFY triggers dynamic WHERE processing.) Note: Dynamic WHERE processing can be costly if the MODIFY statement modifies a SAS data set that is not in sorted order or has not been indexed. Having the master data set in sorted order or indexed and having the transaction data set in sorted order reduces processing overhead, especially for large files. o both the master data set and the transaction data set can have observations with duplicate values of the BY variables. MODIFY treats the duplicates as described in Duplicate BY Values. o MODIFY cannot make any changes to the descriptor information of the data set as UPDATE can. Thus, it cannot add or delete variables, change variable labels, and so on. ... View more

As an alternative, you could use a modify statement to update the master data set with the transaction data set. data master; modify master transaction; by key_variable; run; If you use modify, neither dataset requires sorting. ... View more

Proc sql has an undocumented NOWARN option that will get rid of this warning message. PROC SQL NOWARN; CREATE TABLE TEST AS SELECT U1.*, U2.* FROM DAILY AS U1 LEFT JOIN TEMP AS U2 ON U1.YEAR=U2.YEAR AND U1.SEASON=U2.SEASON ; QUIT; ... View more

Actually SPR’s program works fine. While a title statement is a global statement, the statement is never passed to the input stack to be executed unless the %if statement is true. You can’t conditionally execute global statements in a data step but you can conditionally execute them in a macro. ... View more

Have you tried using an index to combine the tables? [pre] proc sql; create unique index ID on table2 (ID); quit; data table1; set table1; set table2(keep=ID FirstName LastName) key=ID; if _iorc_ ne 0 then do; _error_=0; call missing(FirstName, LastName); end; run;[/pre] A hash object might not be a viable alternative given that the master table contains a million records which would need to be loaded into memory. Message was edited by: polingjw ... View more

Will this work? [pre] data sales; input year qtr rep sales:dollar4.; datalines; 2010 1 100 $1000 2010 2 100 $2000 2011 1 100 $3000 2011 2 100 $4000 ; run; data sales; if _n_=1 then do while(not eof); set sales end=eof; array sales_array[2010:2011, 1:2] _temporary_; sales_array[year, qtr]=sales; end; set sales; retain sales_array; if year > 2010 then previousyrqtrsales = sales_array[year-1, qtr]; else previousyrqtrsales = 0; run; [/pre] Alternatively, you could sort by qtr and year and then use the lag function. ... View more

Of course, this solution assumes that you have no summary lines in the report. When you have summary lines in the report, splitting the dataset into two parts does not work. I have also run into this problem and I never found an acceptable alternative. See http://support.sas.com/forums/thread.jspa?messageID=40487鸧. ... View more

I’m not sure that I fully understand your problem. If you need to display the ≥ symbol in a report generated by SAS, here is one way that it can be done. [pre] ods html file='C:\temp.html'; ods listing close; data temp; x='greater than or equal to'; label x='Symbol is ^{unicode 2265}'; run; ods escapechar='^'; proc print data=temp label noobs; run; ods html close; ods listing; [/pre] ... View more

Cynthia: I appreciate your advice. I’ll research these options for automatic printing to PDF more when I get some free time on my hands. Ksharp: That’s a pretty clever workaround. I still run into a problem when I use by-groups with the STARTPAGE=NO option. In this case, the PAGE option on the BREAK statement appears to have no affect on the report. I’m sure I could get your code to work for my report if I were to stop using by-groups and use LINE statements to separate the groups instead, although the report would not look exactly the same. I plan to try this method in the near future. Many thanks to both of you for the help. ... View more