I don't think so. Cartesian Product of Data step can't compete SQL's at all. Yours is very very slowly . data a; infile cards truncover ; informat   id $20.           date mmddyy10.            code $50.; input id date code; format date mmddyy10.; cards; 19075F106  4/29/2014  a10,axau10,a1010,axau1010 6219640    4/29/2014  n10,v10,x10,w10,n1010,v1010,x1010,w1010 6366007    4/29/2014  s10,v10,s1010,v1010 B63FY34    4/29/2014  a10,axau10,a1010,axau1010 25179M103  4/29/2014  a10,axau10,a1010,axau1010 ; data B; infile cards truncover      ; informat date mmddyy10.    code $8.; input date code; format date mmddyy10.; cards; 4/29/2014  a10 4/29/2014  a10 4/29/2014  a10 4/29/2014  x1010 4/29/2014  w10 ; proc sort data=b out=bb nodupkey;by date code;run; proc sql; create table temp as select a.*,bb.code as new_code   from a,bb    where a.date=bb.date and a.code contains strip(bb.code)     order by id,date,code; quit; data want; set temp; by id date code; length _code $ 40; retain _code; _code=catx(',',_code,new_code); if last.code then do;output;call missing(_code);end; drop new_code; run; Xia Keshan ... View more

You are confused about what a macro is.  SAS macro language is a tool to generate code.  In general you cannot treat a macro as if it was a subroutine function. Here is an example of how you might modify your macro to allow the user to specify the macro variable that you want it use to return the counts. %macro CompareUniqueCols(data1, data2, var, mvar=x); %local noCompareObs; ... %let &mvar=&noCompareObs; %mend; %let x=; %CompareUniqueCols(data1=today,date2=yesterday,var=CustomerID, mvar=X); %put Result = &x ; ... View more

1) I dont see all the available tick values for the columns; Fitpolicy option on the rowaxis may help. The default is THIN which removes values unless there is lots of space. Fitpolicy=Rotate will angle text and may allow more text to display or Fitpolicy=Stagger which will shift some values away from the axis to provide more space for text. Last option may be to try, or modify a style, that uses smaller fonts for the axis labels. also the y axis does not show 0% but ** for some reason The format you assigned wants more space than available is most likely, I'd try a percent3.0 2) How do I increase the size of the x axis and y axis values and the title Bigger text conflicts with your desire to show more values, at least on the y axis. Default size is controlled by style elements. 3) How do I show my sgpanel chart sorted according to SSEC for the rows and the REG for the columns as given above Rowaxis or colaxis statement with a discreteorder option. Choices are basically Data, the order they appear in the dataset, formatted or unformatted. The default is unformatted and character values appear character sort order. If you can sort your data into the order you want then the option data should work. 4) This would be good to have if I can color the titles of each panel "AU", "CA" by the REG specified 5) Any other suggestions to improve the visual.. If you need more control, which appears likely, then you probably need to investigate GTL and proc sgrender. ... View more

Proc summary or univariate with a By or Class statement. proc univariate data=sashelp.cars noprint; class make;   var mpg_city;   output pctlpre=P_ pctlpts= 50, 75 to 100 by 5; run; ... View more

Effective macro coding always starts with the same place: A base SAS section of code that works. Provide an example of ONE of the proc exports you are attempting that works and then indicate which pieces change. And you almost never need to include quotes in the parameters as they will be treated by the macro processor as text. ... View more

I recommend using normal SAS space delimited variable lists when passing them into macros. So to do what you want you could use this simple macro: %macro mytest(out,in1,in2,keys,vars); data &out;   merge &in1(in=in1) &in2(keep &keys &vars);   by &keys;   if in1; run; %mend mytest; Then the call to add both AGE and SEX would be: %mytest(out=dataset,in1=dataset,in2=dataset1,keys=myfldon,vars=age sex); Now if you really want to use SQL, don't force the users to type commas and then have to add macro quoting to prevent the commas from confusing the macro call. Just convert the spaces to commas inside the macro. %macro mytest2(out,in1,in2,keys,vars); %local varlist join i ; %let varlist=%sysfunc(compbl(&vars)); %let varlist=b.%sysfunc(tranwrd(&varlist,%str( ),%str(, b.))); %do i=1 %to %sysfunc(countw(&keys));   %if &i > 1 %then %let join=&join and ;   %let join=&join a.%scan(&keys,&i)=b.%scan(&keys,&i) ; %end; proc sql noprint ;   create table &out as     select a.*,&varlist     from &in1 a left join &in2 b     on &join   ; quit; %mend mytest2; ... View more

Hi HDG.. We can do this using proc rank very easily ..here is the code for you..this may help for your requirement.. data cluster;    input date:mmddyy. cluster_id code:$1.;    format date mmddyy10.;    cards; 05/31/2000 1 A 05/31/2000 1 B 05/31/2000 3 C 05/31/2000 3 E 05/31/2000 5 F 05/31/2001 1 A 05/31/2001 2 B 05/31/2001 5 C 05/31/2001 7 D 05/31/2001 7 E 05/31/2001 7 F ;;;;    run;    proc print;    run; proc rank data = cluster ties = dense  out = want; by date; var cluster_id; ranks clstr_wnt; run; proc print data = want; run; Thanks, Yash... ... View more

Thanks Reeza. yes the number of clusters is known before hand in the cluster data set ... View more

Consider this. proc corr data=sashelp.heart out=cov(where=(_type_ eq 'COV')) cov noprint;    run; data cov;    set cov;    _type_ = 'SCORE';    run; proc print;    run; data coeff; *random weights;    set cov(keep=_type_ _name_);    x = ranuni(4);    run; proc transpose data=coeff out=coeff;    var x;    by _type_;    run; proc print;    run; proc score data=coeff score=cov out=step1;    id _name_;    run; proc score data=step1 score=coeff out=step2;    id _name_;    run; data step3;    set step2;    s = sqrt(x);    run; proc print;    run; ... View more

So the input data may have multiple months for a given ID, but not all months may be present (up to including the "last" desired month)? ... View more