topic Re: 0 observations after recoding or deleting values in SAS Programming

0 observations after recoding or deleting values

maciulei — Fri, 08 Nov 2019 01:52:20 GMT

Hi,

I am dealing with an error that produces 0 observations in my newly formed data set once I use "then delete" statement for missing data or if I want to delete unusual values. Why would that be? I pasted the code below.

Thanks!

data NHANES05.cycle_full;
set NHANES05.NHANES_0506_full_b;
/*Poverty Income Ratio*/
if INDFMPIR<=0.99 then poorcat1=1; else poorcat1=0;
if INDFMPIR>1 and INDFMPIR<=1.99 then poorcat2=1; else poorcat2=0;
if INDFMPIR>2 and INDFMPIR<=2.99 then poorcat3=1; else poorcat3=0;
if INDFMPIR>3 and INDFMPIR<=3.99 then poorcat4=1; else poorcat4=0;
if INDFMPIR>4 and INDFMPIR<=4.99 then poorcat5=1; else poorcat5=0;
if INDFMPIR>5 then poorcat6=1; else poorcat6=0;
if INDFMPIR=. then poorcat=.;

run;

**delete missing data;
data NHANES05.cycle_full_a;
set NHANES05.cycle_full;

/*Poverty Income Ratio*/
if INDFMPIR>=0 and INDFMPIR<=5 then poor1=1; else poor1=0;
if INDFMPIR=. then delete; *none based on this code but outside of the 0-5, N=72 coded as missing;

run;

Re: 0 observations after recoding or deleting values

unison — Fri, 08 Nov 2019 02:58:51 GMT

A few things, when you do "<=0.99" that is different than later when you specify ">=0". Also, I think you may have observations that are exactly equal to 1,2,3,4,5 but your bucketing does not capture them.

Given that I think you are trying to bucket the observations, I would use proc format instead. (if you don't want to include 1, 2, 3, etc. change 1-1.99 to 1<-1.99, 2-2.99 to 2<-2.99 etc.)

proc format;
	value poorcat low-0.99=1 1-1.99=2 2-2.99=3 3-3.99=4 4-4.99=5 5-5.99=6 other=0;
run;

data have;
	set sashelp.syr1001(keep=S0666);
	poor_cat=put(S0666, poorcat.);
run;

data want;
	set have;
	if poor_cat ne 0;
run;

-unison

Re: 0 observations after recoding or deleting values

Astounding — Fri, 08 Nov 2019 03:12:46 GMT

Your program uses INDFMPIR but doesn't change it in any way.

Therefore you need to go back to your original data set and look for missing values there.

Perhaps the incoming values for INDFMPIR are incorrect.

Re: 0 observations after recoding or deleting values

Kurt_Bremser — Fri, 08 Nov 2019 07:29:21 GMT

What does the log of both steps say? Please post it, using the {i} button.

Also inspect the contents of INDFMPIR in your initial dataset.

Re: 0 observations after recoding or deleting values

maciulei — Fri, 08 Nov 2019 15:59:15 GMT

I checked the log and one of the imported files gave the error that the file is sequential in red. I fixed the issue by checking the file location, which was wrong. Now everything works!

Re: 0 observations after recoding or deleting values

Kurt_Bremser — Fri, 08 Nov 2019 16:10:47 GMT

@maciulei wrote:
I checked the log and one of the imported files gave the error that the file is sequential in red. I fixed the issue by checking the file location, which was wrong. Now everything works!

That's why Maxim 2 is so high up on the list.