Displaya count of observations horizontally

[Q1983]

data have;
item1 = 'central';
item2 = '';
item3 = 'Pacific';
run;

I want to apply a formula that gives me the total number observations by row and displays a field called row_cnt.  So in this case the total would be 2 since there are two observations.  The variables however are character in nature

[tarheel13]

Let me know if this is what you're looking for.

data have;
item1 = 'central';
item2 = '';
item3 = 'Pacific';
run;

proc sql;
	select count(name) into: total
		from dictionary.columns
		 where upcase(libname)='WORK' and upcase(memname)='HAVE';
quit;

%put &=total;

data have2;
	set have;
	totalvar=&total;
	num=totalvar-cmiss(of item:);
run;

[ballardw]

Please show what you want the output to look like.

Note: "Observation" is SAS for "row" or "record". So your example data only has 1 observation and 2 non-missing values.

One way for what I think you are asking for:

data have;
   item1 = 'central';
   item2 = '';
   item3 = 'Pacific';
   row_cnt = countw(catx('*',item1,item2,item3),'*');
   row_cnt2= countw(catx('*',of item:),'*');
run;

which creates a string of your values with * between the ones that are not missing and then uses the Countw function to count how many separations are there.

If you have more than 3 or 4 variables you might use an Array or a list such as the "of item:" to use all the variables whose names start with Item to build the string before counting.

[Kurt_Bremser]

---

@Q1983 wrote: So in this case the total would be 2 since there are two observations.

---

No. You have one observation containing three variables, of which two have values.

(deleted non-working code)