Mmm, was just coming to the same conclusion there, change picture to value.  No idea why it wouldn't work, maybe a storage issue in that picture formats do not have the required quotes or something? ... View more

Just to add, for proc report, adding the nowd option to the proc report will prevent the output window.  I find the output/results windows just to be annoying, wrap a simple ods statement in, then close the output window and results window, only need log then. ... View more

Hi, Well, I don't use EG or this query builder, however it should be straightforward in SQL, something (and you question was a bit vague) along the lines of: select     A.*,                B.TOTAL from       HAVE left join (select T1_SERIAL_NO,sum(HOUSEHOLD_INCOME) from HAVE group by T1_SERIAL_NO) B on          A.T1_SERIAL_NO=B.T1_SERIAL_NO ... View more

Well, with that example the following achieves the output (oh, and I shouldn't have used best. on the format of the array): data have;   attrib Var   D1   D2   D3   D4   D5   D6   D7   D8   D9   K2_1   K2_2   K2_3   K2_4   K2_5   K2_6   K2_7   K2_8   K2_9  maxi format=best.;   infile datalines;   input Var   D1   D2   D3   D4   D5   D6   D7   D8   D9   K2_1   K2_2   K2_3   K2_4   K2_5   K2_6   K2_7   K2_8   K2_9  maxi; datalines; 100   15    30   45    60   75   100 125  130  132    77        88      99      111       101    98        87      76        70      111 run; data want (keep=var d_result k2_result l1_var l2_var);   set have;   attrib d_result k2_result l1_var l2_var format=best.;   array K2_{9} 8.;   array D{9} 8.;   i=1;   do while (i<=9 and l1_var=.);     if K2_{i}=maxi then do;       d_result=D{i};       k2_result=K2_{i};       if var <= D{i} then l1_var=1; else l1_var=0;       if var > D{i} then l2_var=1; else l2_var=0;     end;     i=i+1;   end; run; ... View more

Could you provide some test input and desired output. ... View more

First thing I would change is: %macro def(var);      proc univariate .....     /*   to obtain the D1_   ... D9_  */          /* Change these to be Dx - where x=1 to 9 - no                                                                                                          underscores */           ...      proc freq data .....      /* to obtain K2_1    ...    K2_9  */           ...      data x;            set testing;        array D{9} best.;        array K2_{9} best.;        array I{9} best.;        do i=1 to 9;             I{I}= (K2_(i)=maxi and &var <= D(i));     /* This will fill the I array with 1 or 0 based on K and D array versus                                                                                      maxin and &var */             /* This logic does not appear in your original description             I2_&var=1-I2_&var;                        hence I remove it */        end;      run; %mend def; ... View more

Hi, Yes, you are mixing datastep and macro code there.  My example uses arrays in a normal datastep so you shouldn't need any macro code - though I am not sure and it doesn't say in your example where &var is defined.  Note also that you don't want the underscores as they then don't match the arrays : %let var=SOME_VARIABLE; data x;   set testing;   array D{9} best.;   array K2{9} best.;   array I{9} best.;   do i=1 to 9 by 1;   if (K2{i}=maxi and &var <= D{i}) then I{I}=1; else I{I}=0;   if (K2{i}=maxi and &var <= D{i}) then I{I}=0; else I{I}=1;   %end; run; The principal is that you have a set of variables all called the same thing, e.g. d1 d2 d3 - k2_1 k2_2 etc. these can be thought of as arrays.  You want a variable output for each one of these - another array.  Then its simply a matter of do looping over the 9 possibles and checking each one to your output array. ... View more

Hi, attrib newdate format=datetime15.; newdate=input(put(date,date7.)||put(time,hhmmss.),datetime.); ... View more