topic generate data set for imputation in SAS Programming

generate data set for imputation

superbibi — Fri, 30 Nov 2018 16:25:34 GMT

Hi friend, I have a data set that some of the participants are missing some of the visit data. I want to make up those missing for later imputation. I can only come up with the method with generating a missing data set for those participants and then stack with the original data set. Could you show me a faster way?

Thanks.

data I have

Obs	randomization	visit	tac	base_tac
1	618001	2	8824	8824
2	618001	3	9088	8824
3	618001	4	9784	8824
4	618002	2	7962	7962
5	618002	3	8318	7962
6	618002	4	8819	7962
7	618003	2	10630	10630

data I want

Obs	randomization	visit	tac	base_tac
1	618001	2	8824	8824
2	618001	3	9088	8824
3	618001	4	9784	8824
4	618002	2	7962	7962
5	618002	3	8318	7962
6	618002	4	8819	7962
7	618003	2	10630	10630
7	618003	3		10630
7	618003	4		10630

Re: generate data set for imputation

PaigeMiller — Fri, 30 Nov 2018 16:27:46 GMT

Could you explain further ... why does 618003 get two extra rows, and the other IDs do not get additional rows?

Re: generate data set for imputation

superbibi — Fri, 30 Nov 2018 16:36:19 GMT

@PaigeMiller. Thank you for the interest. Each ID should have 3 visits. The visit3,4 is missing for ID 618003. Therefore, if I want to impute the missing values for ID 618003, I think I have to generate the missing cells for ID 618003. The rest of the ID have visit2,3,4, without missing. That is why only ID 618003 requires the additional two rows. Is it clear? Please let me know.

Thank you again.

Re: generate data set for imputation

r_behata — Fri, 30 Nov 2018 17:19:39 GMT

Is this what you are looking for :

data have;
input Obs randomization visit tac base_tac ;
cards;
1 618001 2 22 22
2 618001 3 33 22
3 618001 4 44 33
4 618002 2 44 33
5 618002 3 33 33
6 618002 4 33 33
7 618003 2 33 55
;
run;

Proc SQl;
Create table temp as 
	select distinct randomization 
	from have;
Quit;

data visit;
	do visit=2 to 4;
		output;
	end;
run;

data temp2;
set temp;
	do i=1 to nobs_;
		set visit point=i nobs=nobs_; output;
	end;
run;


proc sort data=have;
	by randomization visit;
run;

data want;
	merge have temp2;
	by randomization visit;
	
	retain base_tac_;

	if first.randomization then base_tac_=base_tac;
		if missing(base_tac) then base_tac=base_tac_;
	obs=_n_;
	drop base_tac_;
run;