I am trying to set up a data step that expunges some 'incorrect' values from a dataset. For instance, I have a table:   data a; input (A1 A2 A3) ($); datalines; A  B  D B  B  D D  A  E ;   And I am trying to remake the table but where all values are within this column/dataset (only A, B and C):   data b; input (B1) ($); datalines; A   B   C  ;   Such that the final output is    data a_2; input (A1 A2 A3) ($); datalines; A  B  . B  B  . .   A  . ;   I know I can check if a variable is 'in' a column with PROC SQL, but it seems like the code to do so would be a bit long - my actual dataset involves far more than 3 variables and 3 allowed values so I was curious if there was a simpler way to do it. ... View more

I am trying to write a very simple procedure that just counts the distinct values of a given number of fields across a few variables, and tabluates them. PROC FREQ does it for one, but I'm trying to do more than one at once and have them all in a table - haven't had any luck with PROC TABLUATE as it seems to be not for this particular purpose.   For instance, given the below dataset:   data a; input (A1 A2 A3) ($); datalines; A  B  D B  B  D D  A  E ;   I would expect some output table like:           A1    A2    A3 A      1      1       0 B      1      2      0 D     1       0      2 E     0       0      1   Is there a way to do that with PROC FREQ or PROC TABLUATE?  ... View more

As an alternative to PROC SQL for speed, I want to use hash tables to left join multiple tables.  I have 4 tables - A, B, C, D. Tables B, C, and D each have their own 'key' fields that table A possesses as well. I am trying to left join B, C and D onto A on the basis of this. The below is my code, it is based on this article: https://www.lexjansen.com/nesug/nesug06/dm/da07.pdf   DATA COMBINED; IF 0 THEN SET A B C D; IF _N_ = 1 THEN DO;   DECLARE HASH B_(dataset: 'B');      B_.defineKey('B_key');      B_.defineData('B_data');      B_.defineDone(); DECLARE HASH C_(dataset: 'C');      C_.defineKey('C_key');      C_.defineData('C_data');      C_.defineDone(); DECLARE HASH C_(dataset: 'D');      D_.defineKey('D_key');      D_.defineData('D_data');      D_.defineDone(); END;   DO UNTIL (EOF);     SET A END = EOF     IF B_.FIND(KEY:'A_KEY')=0        THEN OUTPUT;     ELSE IF C_.FIND(KEY:'A_KEY')=0        THEN OUTPUT;     ELSE IF D_.FIND(KEY:'A_KEY')=0        THEN OUTPUT;     ELSE DO;          CALL MISSING(OF 'B_data', 'C_data', 'D_data');         OUTPUT;      END;     END;    STOP; RUN;   The trouble is at the end; the iteration through the table 'A' doesn't go the way I'd like, and it produces some strange results (like values being filled even though I wouldn't expect there to be a hash key match). Does anyone know of the correct way to iterate through a table and match multiple hash keys to simulate a SQL left join?   Much appreciated. ... View more