For numeric variables, you can't do that.  A missing value for a numeric variable is displayed as a dot.  The only thing you can do differently is to change how it gets displayed:   options missing=" ";   If you happen to have character variables that contain a dot, you can change them to blanks:   data want; set have; array chars {*} _character_; do _n_= 1 to dim(chars); if chars{_n_} = "." then chars{_n_} = " "; end; run; ... View more

data have ; input product $1. month amount ; datalines; A 1601 1 A 1602 2 A 1603 3 A 1604 4 A 1605 5 A 1606 6 A 1607 7 A 1608 8 A 1609 9 A 1610 10 A 1611 11 A 1612 12 A 1701 13 A 1702 14 A 1703 15 A 1704 16 A 1705 17 A 1706 18 A 1707 19 A 1708 20 A 1709 21 A 1710 22 A 1711 23 A 1712 24 B 1601 1 B 1602 2 B 1603 3 B 1604 4 B 1605 5 B 1606 6 B 1607 7 B 1608 8 B 1609 90 B 1610 100 B 1611 110 B 1612 120 B 1701 130 B 1702 140 B 1703 150 B 1704 160 B 1705 170 B 1706 180 B 1707 190 B 1708 200 B 1709 210 B 1710 220 B 1711 230 B 1712 240 ; run; data want; if _n_=1 then do; if 0 then set have; dcl hash H (dataset:'have',ordered: "A") ; h.definekey ("product","month") ; h.definedata ("amount") ; h.definedone () ; dcl hiter ih('h'); end; set have(drop=amount); by product month; RC = IH.SETCUR() ; _p=product; do count = 1 to 13 while (RC = 0 and _p=product) ; sum=sum(sum,amount); RC = IH.next() ; end ; if count < 13 then sum=0; drop rc count product; rename _p=product; run; ... View more

thanks all for your precious help. unfortunatly, I did not succeed with your suggestions. But I did succeed with this code below   %macro mc_loop_repeat_missing(p_table=t_mois_trsp_A) ; %let i=1; %let an_loop = %scan(&liste_annee, &i, %str( )); %do %while (&an_loop ne ); %let an_loop = %scan(&liste_annee, &i, %str( ));   data &p_table ; set &p_table ; retain _&an_loop ; if not missing(&an_loop) then _&an_loop = &an_loop ; else &an_loop = _&an_loop ; drop _&an_loop ; run ;   %LET i = %eval(&i+1); %let an_loop = %scan(&liste_annee, &i, %str( )); %end ; %mend ;   ... View more

Something like (not tested as nothing to test on): data want; set have 1 (in=a) have2 (in=b); /* Note I do not know what variables are in each so I cannot put anything in the next row */ by ...; if a and not b; run; You could also use except in SQL, proc compare with dataset output.  Again, without test data in the form of a datastep I can't be anymore specific. ... View more

Yes, describe will give you the SQL to re-create a table/dataset.   ... View more