topic Re: Do Loop in SAS Programming

Do Loop

Tal — Sun, 21 Oct 2012 21:51:12 GMT

hi guys,

I have the code and some of the output below.If someone could explain the 5th observation from the output please?

I was expecting to see

10 20 30 999 50 but probably retain issue?

Thanks people

------------------------------------------

data old;
input a b c d e;
datalines;
10 20 30 999 50
50 40 999 20 10
999 1 1 1 999
;
run;

data new;

	array x(5) a b c d e;
	set old;
	do i=1 to 5;
	if x(i) eq 999 then x(i) = .;

output;

	end;
	drop i;

proc print;

run;

-----------------------------------------------------------------------------------------

and here is some of the output:

1	10	20	30	999	50
2	10	20	30	999	50
3	10	20	30	999	50
4	10	20	30	.	50
5	10	20	30	.	50

Re: Do Loop

Reeza — Sun, 21 Oct 2012 21:59:16 GMT

That makes sense, the values don't reset until the hit the end of the datastep, but since you're outputting within the loop they aren't reset, because it hasn't hit the end of the datastep.

Don't drop the i if that helps you to visualize it.

Re: Do Loop

Tal — Sun, 21 Oct 2012 22:04:12 GMT

sorry,i dont understand

Re: Do Loop

Tal — Sun, 21 Oct 2012 22:08:25 GMT

once the x(i) hits the missing value it never changes during each observation regardless of whether the if statement is true or not.Why is that so?

Re: Do Loop

art297 — Sun, 21 Oct 2012 22:15:12 GMT

Why would you expect to see "10 20 30 999 50" when your code is requiring "10 20 30 . 50"?

Do you really want 5 records output for each record? If not, remove the output; statement.

Re: Do Loop

Reeza — Sun, 21 Oct 2012 22:16:42 GMT

the do loop goes through the 5 numbers. When it hits the 4th number, the 999, it sets it to missing. Then the 5th number is checked, and its not 999 so it doesn't get set to missing. However the 4th was missing so it still is missing.

Basically your code loops through the 5 numbers and if any is 999 it sets it to missing for the rest of the loop. Not sure why you want to output it for every line, but that's what you're doing.

Re: Do Loop

Tal — Sun, 21 Oct 2012 22:27:58 GMT

i do get the right result without the "output" but i just wanted to play with it and understand it more.So i realize that once a value is set to missing it remains missing till the loop hits another observation but it just did not make sense to me .Is this because of the "output" statement?

Re: Do Loop

Tal — Sun, 21 Oct 2012 22:30:29 GMT

i mean if x(5) =50 and not 999 does not that mean x(5) is different then missing based on the if statement or does the missing values overrules the if statement?

Re: Do Loop

art297 — Sun, 21 Oct 2012 22:35:37 GMT

if x(5)=50 then the code won't change its value. The code will only change the value of x(whatever) if x(whatever)=999.

Re: Do Loop

Tal — Sun, 21 Oct 2012 22:37:27 GMT

oh wait ,when the loop checks the if statement does it check it against the original observation or each time it checks it against the last observation written out?

Re: Do Loop

Tal — Sun, 21 Oct 2012 22:38:30 GMT

right right,x(5) was not changed i just made a typo posting my message

Re: Do Loop

Tal — Sun, 21 Oct 2012 22:50:34 GMT

probably a stupid question but otherwise if the observation is 10 20 30 999 50 and if statement not true then i dont see why x(4) would be till missing.

Re: Do Loop

art297 — Sun, 21 Oct 2012 22:51:42 GMT

Run the following, then look at your log. I think it will answer your question:

data old;
input a b c d e;
datalines;
10 20 30 999 50
50 40 999 20 10
999 1 1 1 999
;
run;

data new;

array x(5) a b c d e;
set old;
do i=1 to 5;

if x(i) eq 999 then x(i) = .;

putlog _all_;

end;
drop i;
run;

Re: Do Loop

Tal — Sun, 21 Oct 2012 23:11:01 GMT

i changed the missing value to a "0" value and same scenario.So looks like for x(5) the loop checks the if statement against observation #4 not #1.Thank you guys

Re: Do Loop

art297 — Sun, 21 Oct 2012 23:15:09 GMT

I think you may not understand what is happening. The loop isn't checking across observations, it is check the five values, in each observation, or a, b, c, d and e.

Re: Do Loop

Tal — Sun, 21 Oct 2012 23:21:05 GMT

ok ,when the loop checks for x(5) or "e" he checks into 10 20 30 999 50 or into 10 20 30 0 50?

Re: Do Loop

Tal — Sun, 21 Oct 2012 23:23:43 GMT

i mean 10 20 30 999 50 or 10 20 30 . 50?

Re: Do Loop

art297 — Sun, 21 Oct 2012 23:42:01 GMT

When the loop checks for x(5) it is only checking for e or 50. Since 50 isn't equal to 999 it is left at 50.

Re: Do Loop

art297 — Sun, 21 Oct 2012 23:48:05 GMT

It might help if you did the same thing without either a loop or array. Your code is doing the same thing as below:

data old;
input a b c d e;
datalines;
10 20 30 999 50
50 40 999 20 10
999 1 1 1 999
;
run;

data new;

set old;

if a eq 999 then a=.;

if b eq 999 then b=.;

if c eq 999 then c=.;

if d eq 999 then d=.;

if e eq 999 then e=.;

run;

First, it is doing it for the first row (i.e., 10 20 30 99 and 50), then outputting the results, then doing it for the 2nd row and outputting the results, and then again for the third row.

Re: Do Loop

Tal — Mon, 22 Oct 2012 00:11:34 GMT

i could get the results using different methods but i wanted to understand how loop works with set and at the end i got stuck.

ok,a not 999 so sas writes the observation

same for b and c.Then d=999 so it gets set to missing and sas writes that observation too then when the loops checks for e,last element not 999 nothing changes and sas writes the 5th observation but with the missing value for d again.So it looked like the loop checked 10 20 30 . 50 for the value of d and not 10 20 30 999 50. Probably i am missing something huge about do loops and the output statement.

1	10	20	30	999	50
2	10	20	30	999	50
3	10	20	30	999	50
4	10	20	30	.	50
5	10	20	30	.	50
6	50	40	999	20	10
7	50	40	999	20	10
8	50	40	.	20	10
9	50	40	.	20	10
10	50	40	.	20	10
11	.	1	1	1	999
12	.	1	1	1	999
13	.	1	1	1	999
14	.	1	1	1	999
15	.	1	1	1	.

1	10	20	30	999	50
2	10	20	30	999	50
3	10	20	30	999	50
4	10	20	30	.	50
5	10	20	30	.	50
6	50	40	999	20	10
7	50	40	999	20	10
8	50	40	.	20	10
9	50	40	.	20	10
10	50	40	.	20	10
11	.	1	1	1	999
12	.	1	1	1	999
13	.	1	1	1	999
14	.	1	1	1	999
15	.	1	1	1	.

1	10	20	30	999	50
2	10	20	30	999	50
3	10	20	30	999	50
4	10	20	30	.	50
5	10	20	30	.	50
6	50	40	999	20	10
7	50	40	999	20	10
8	50	40	.	20	10
9	50	40	.	20	10
10	50	40	.	20	10
11	.	1	1	1	999
12	.	1	1	1	999
13	.	1	1	1	999
14	.	1	1	1	999
15	.	1	1	1	.

1	10	20	30	999	50
2	10	20	30	999	50
3	10	20	30	999	50
4	10	20	30	.	50
5	10	20	30	.	50
6	50	40	999	20	10
7	50	40	999	20	10
8	50	40	.	20	10
9	50	40	.	20	10
10	50	40	.	20	10
11	.	1	1	1	999
12	.	1	1	1	999
13	.	1	1	1	999
14	.	1	1	1	999
15	.	1	1	1	.