topic Re: Assign value to missing value in SAS Programming

Assign value to missing value

Q1983 — Sun, 18 Aug 2019 17:23:29 GMT

data temp;

infile DATALINES dsd missover;

input a $ b $ c $ d $;

CARDS;

, 2, 3, 4

, 3, , 5

, 3

;

run;

This code will list a with no records. Actually I am using a proc sql however using the above discussion. In my proc sql, the A column does not show up if there were no records. My question is, in the event of no records how can I still show the column and assign a 0 to show that there were no records for (in this case column A) The reason for this need is that I need to do a proc transpose and need to show all columns even if there are no records for a column

Re: Assign value to missing value

novinosrin — Sun, 18 Aug 2019 17:32:53 GMT

Hi @Q1983 Do you mean , should a column be all missings, you want to replace with zero?

Re: Assign value to missing value

Q1983 — Sun, 18 Aug 2019 18:04:52 GMT

For example in this datastep column A would show up with no records. So for A I still want to show the column A and assign 0 in the rows.

Re: Assign value to missing value

novinosrin — Sun, 18 Aug 2019 21:33:53 GMT

Hi @Q1983 Well FWIW, I'm offering an approach based on my understanding of your objective. Of course you are better off waiting for somebody like Tom to chime in as they can provide more guided and pointed approaches/solutions


data temp;
infile DATALINES dsd missover;
input a $ b $ c $ d $;
CARDS;
, 2, 3, 4
, 3, , 5
, 3
;
/*Get a subset of columns that has all missing values*/
ods output nlevels =nlevels(where=(nlevels=nmisslevels));
proc freq data=temp nlevels ;
tables _all_ /noprint ;
run;

/*Assign zero for the identified columns in the previous step*/
data _null_;
set nlevels(keep=tablevar) end=z;
if _n_=1 then call execute('data want;set temp;');
call execute(tablevar||'='||'"0";');
if z then call execute('run;');
run;

Re: Assign value to missing value

Patrick — Sun, 18 Aug 2019 21:52:21 GMT

@Q1983

Before trying to provide a solution we need to understand the problem. What you describe doesn't add up for me because Proc Transpose will treat a missing like any other value and won't drop observations even if all variables are missing. Using your sample below code illustrates this behavior. So... What's the problem?

data temp;
  infile DATALINES dsd missover;
  input a $ b $ c $ d $;
  CARDS;
 
, 2, 3, 4
, 3, , 5
, 3
;
run;

proc transpose data=temp out=want prefix=wantvar_;
  var a b;
run;

proc print data=want;
run;