topic Re: can you detect errors in this code? i do not know what does not work in New SAS User

can you detect errors in this code? i do not know what does not work

ari2495 — Mon, 28 Nov 2022 18:06:04 GMT

data final_dataset;
set initial_dataset;
length variable $50.;
if cond not in (0) then do; 
	variable_1="S2";
	variable="xxx";
end;
else if cond_1 > 30 then do;
	variable_1="S2";
	variable="xxy";
	end;
else if cond_2 > soglia then do;
	variable_1 ="S2";
	variable="xyy";
	end;
else if cond_3="Stage 1" then do;
	variable="yyy";
	end;
run;

the problem is that, when i print the result, the new column i created named "variable" is empty. i cannot understand why

Re: can you detect errors in this code? i do not know what does not work

PeterClemmensen — Mon, 28 Nov 2022 18:09:13 GMT

Your code looks fine. The problem is likely in the initial_dataset data.

Re: can you detect errors in this code? i do not know what does not work

mkeintz — Mon, 28 Nov 2022 18:49:44 GMT

Your logical structure is equivalent to

if A then create results ;

else if B then create results;

else if C then create results;

else if D then create results;

Even though you are using "else if" syntax, the set of observations that would satisfy the above would be this intersection:

A ∩ B ∩ C ∩ D

If your log shows no errors messages, yet you have no observation with values for the result variables, then every obs in your dataset falls outside the above intersection. So, by De Morgan's Laws all of your observations fall in the complement, characterized by

(not A) ∪ (not B) ∪ (not C) ∪ (not D)

(where ∪ is the union set operator and is equivalent to and in any subsetting syntax in SAS.

Now, in your case

condition A <==> cond not in (0)

condition B <==> cond_1 > 30

condition C <==> cond_2>soglia

condition D <==> cond_3="Stage 1"

which means that all of your observations must have

cond in (0) and

cond_1<=30 and

cond_2<=soglia and

cond_3 ^="Stage 1"

Re: can you detect errors in this code? i do not know what does not work

SASKiwi — Mon, 28 Nov 2022 19:07:16 GMT

PROC FREQ will help you find what is wrong with your logic, given there are no syntax errors in your code:

proc freq data = initial_dataset;
  table cond * cond_1 * cond_2 * cond_3 / missing list;
run;

Examine the combination of values shown in the output.

Re: can you detect errors in this code? i do not know what does not work

Tom — Mon, 28 Nov 2022 19:45:23 GMT

Did SAS print any notes about uninitialized variables? Do you really have the variables you are referencing:

cond cond_1 cond_2 soglia cond_3

If there are such notes then fix issue first.

Otherwise look at the actual values of the variables you are testing.

For example try running this PROC FREQ and see what it tells you.

proc freq data= initial_dataset;
  tables cond*cond_1*cond_2*soglia*cond_3 / list missing;
run;

Are there any rows output that meet your conditions?

Re: can you detect errors in this code? i do not know what does not work

ari2495 — Tue, 29 Nov 2022 07:09:54 GMT

I really have to thank you all for all
Your answers; I detected that the problem was with the last “then do”; simply eliminating it the code worked well again. Many thanks!