how to use array or macro to rewrite the code

[Bal23]

current working code:

data m2 m3;
set m1;
if substr(d_c_p1,1,3) = 'o9' or substr(d_c_p2,1,3) = 'o9' or substr(d_c_p3,1,3) = 'o9' 
or substr(d_c1,1,3) = 'E8' or substr(d_c2,1,3) = 'E8'
or substr(d_c3,1,3) = 'E8' or substr(d_c4,1,3) = 'E8' or substr(d_c4,1,3) = 'E8' or substr(d_c5,1,3) = 'E8' or substr(d_c6,1,3) = 'E8'or substr(d_c7,1,3) = 'E8'  then output m3;
else output m2;
run;

The one I modify but does not work so far.
array d_c_p{3}d_c_p1-d_c_p3;
do i=1 to 3;
if substr(d_c_p{i},1,3) = 'o9' then do output=m3; end;

array dx_code{7}d_c1-d_c7;
	do j=1 to 7;
		if substr(d_c{j}, 1,3)='E8' then do output=m3; end;
	end;end;
else output m2;

the error message is:

: No matching IF-THEN clause

for the else output m2

please give advice how to correct thanks

[Reeza]

1. Missing a semi colon after DO.

2. Not an error but it's easier if you declare all arrays at top of code

3. Reference for second array should be dx_code not d_c(j)

4. When trying to output it's OUTPUT <dataset names(s)>, no equal sign

Ways to avoid these errors:

• Program one piece at a time
• Read your log
• Format code with indends to see logic more clearly
• If you have only one action after an IF condition you don't need an DO/END. (if <condition met> then output <dataset name(s)>)

I didn't actually check if you have the correct do/end amounts or if the code makes sense, just the few things that a quick scan found. You may still have other errors, but you're on the correct path.

array d_c_p{3}d_c_p1-d_c_p3;
array dx_code{7}d_c1-d_c7;

	do i=1 to 3;

		if substr(d_c_p{i}, 1, 3)='o9' then
			do;
				output m3;
			end;
		

		do j=1 to 7;

			if substr(dx_code(j), 1, 3)='E8' then
				do;
					output m3;
				end;
		end;
	end;
	else
		output m2;

[Bal23]

Thanks. I use your code

still the error message showed

else marked as red

no matching if-then clause

[Reeza]

Count your do/end and make sure they match up. 

EDIT: Show your log and trace your logic from that statement. Find the matching If for your last else condition.

Spoiler

PS It doesn't exist.

[Bal23]

I have four do and four end

the error marked :else as saying no matching if-then clause

        output m3;
1330                  end;
1331          end;
1332  END;
1333      else
          ----
          160
ERROR 160-185: No matching IF-THEN clause.

1334          output M2;
1335

[ballardw · Accepted solution]

Your first data step will output one record into M3 if any of the values match.

Your array approach is going to output the record each time one of the values matches, is that what you want?

Reformatting the code with the error to show the relationship between your do loops:

data m2 m3;
   set m1;
   array d_c_p{3}d_c_p1-d_c_p3;
   array dx_code{7}d_c1-d_c7;

   do i=1 to 3;
      if substr(d_c_p{i},1,3) = 'o9' then do output=m3; end;

   	do j=1 to 7;
   		if substr(d_c{j}, 1,3)='E8' then do output=m3; end;
   	end;
   end; /* would end the i loop*/
   else output m2; /* this else is not associated with any IF.*/
run;

Provide an example of the input data, in the form of a data step. This links to instruction on a way to create a datastep from your M1 set: https://communities.sas.com/t5/SAS-Communities-Library/How-to-create-a-data-step-version-of-your-dat...

The provide the desired result.

You might be looking for something similar to this, untested

data m2 m3;
   set m1;
   array var {10} d_c_p1-d_c_p3 d_c1-d_c7;
   array val {10} $ 2 _temporary_ ('o9','o9','o9','E8','E8','E8','E8','E8','E8','E8') ;

   do i=1 to dim(var);
      if substr(var{i},1,3) = val{i} then do;
         output=m3; 
         Found=1;
         leave ;
      end;
   end;

   if missing(found) then output m2; 
   drop found;
run;

[Bal23]

in reality, , J=30 ( I need to repeat 30 times), i=3, so to use your code, seems to be a little bit time consuming

any other advice?

[Reeza]

Why does it matter how many times it occurs? 

[ballardw]

---

@Bal23 wrote:

in reality, , J=30 ( I need to repeat 30 times), i=3, so to use your code, seems to be a little bit time consuming

any other advice?

---

If you are referring to my code with the temporary array being "time consuming" I think you have another issue. Adding another 20 variables and 20 values to the temporary array (in the correct order) as long as you are looking at the first 3 characters of each variable is a trivial amount of time compared to writing 30 if then else type statements and relatively easy to maintain.

[Tom]

It really helps if you explain what the code is trying to do.

I notice that the two blocks of code are doing different things.  Is that on purpose?

Your first example outputs one copy of the record to either of two datasets.

Your second would (if it ran) output multiple copies of the record when there are multiple variables that match the condition.

[KachiM]

EDIT:

To show an improved version of the earlier program. Here, after taking the 2-byte portions, put all together into a STRING- just taking not more than 50 bytes, using INDEX function we can search for the presence of '09' OR 'E8'. No need for 2 do-loops.

data m2 m3;
   set m1;
   length dcp1 dcp2 dcp3 dc1 dc2 dc3 dc4 dc5 dc6 dc7 $2;
   length STR $50;
   dcp1 = d_c_p1;
   dcp2 = d_c_p2;
   dcp3 = d_c_p3;
   
   dc1 = d_c1;
   dc2 = d_c2;
   dc3 = d_c3;
   dc4 = d_c4;
   dc5 = d_c5;
   dc6 = d_c6;
   dc7 = d_c7;

   call catx(' ',STR, dcp1,dcp2,dcp3,dc1,dc2,dc3,dc4,dc5,dc6,dc7);

   match = (index(STR, '09') | index(STR, 'E8'));
 
   if match then output m3;
   else output m2;
run;

1. You are taking 3 characters and comparing with 2 characters. I presume that you want 2 characters.
2. Assigning to a character variable of length 2 with your string will be faster than the use of SUBSTR() function.

With these changes I placing the program below. This is not tested as your data set M1 is not known. Please test it and let the Community know about it.

data m2 m3;
set m1;
length dcp1 dcp2 dcp3 dc1 dc2 dc3 dc4 dc5 dc6 dc7 $3;
dcp1 = d_c_p1;
dcp2 = d_c_p2;
dcp3 = d_c_p3;

dc1 = d_c1;
dc2 = d_c2;
dc3 = d_c3;
dc4 = d_c4;
dc5 = d_c5;
dc6 = d_c6;
dc7 = d_c7;
do i = dcp1, dcp2, dcp3;
if i = '09' then do;
Match1 = 1;
leave; * Early match ;
end;
end;
do i = dc1, dc2, dc3, dc4, dc5, dc6, dc7;
if i = 'E8' then do;
Match2 = 1;
leave; *Early Match;
end;
end;
if Match1 or Match2 then output m3;
else output m2;
run;

[Bal23]

Thanks for the effort. In fact, j=30, i=3, as I mentioned abover, so with your code, it generates a lot more variables, which take much time for later analysis. Any other tips.

[KachiM]

Only i is used in the program and j is not.
Where do you want j = 30? Place your code and the LOG message for helping you further.

[KachiM]

I found i and j are used by ballardw. 

I have a Question.

Comparing a substring of 3-characters with 2-character value('09', 'E8') will give wrong results. Did you check this?

[Bal23]

Thanks. You are right. The value should be three but I typed wrong. I have used your code which works.

The fact is i have dcp1 to 3, and dc 1-30 instead of 7 that I previously posted.  when i use you code, I have to type many times, adding more variables

that is why I am asking whether you have a better suggestion on this