topic Re: subsetting in SAS Data Management

subsetting

superbibi — Fri, 02 Dec 2016 02:34:02 GMT

I have a dataset as below. I want to keep the value based on the “studyeye”. For example, if it is OD in “studyeye", I keep only the value from “PrevIOP_OD_1". How can I do that in SAS studio?

Re: subsetting

RahulG — Fri, 02 Dec 2016 03:07:33 GMT

It is insterting problem, what I understand you want to control the column in output dataset based on the value in the table.

In below example, I tried to keep only column Type when Make='Audi'.

data cars ;
set sashelp.cars(obs=10);
run;

%macro control_column(cond=,keep_list=);
data cars_new;
set cars;
where &cond;
keep &keep_list;
run;
%mend;

%control_column(cond=Make='Audi', keep_list=Type);

Re: subsetting

superbibi — Fri, 02 Dec 2016 03:17:34 GMT

Thank you Rahul. Do you keep the whole column or just some values in the column? I want to keep only some value of the column. Could you upload part of the dataset?

Re: subsetting

RahulG — Fri, 02 Dec 2016 03:20:54 GMT

You can execute the above code in SAS studio and see if you are getting desired result.

Re: subsetting

Reeza — Fri, 02 Dec 2016 05:05:53 GMT

Based on your data what would you expect as your output?

Re: subsetting

superbibi — Fri, 02 Dec 2016 14:59:48 GMT

I want to have only the PrevIOPOD_1/PrevIOPOS_1 values based on the studyeye.

I come up with a maybe easier solution. I can split the dataset based on the sutdyeye into two subdatasets. Then I can keep one of the PrevIOPOS_1/PrevIOPOS_1. then I combine them again.

Re: subsetting

David_Luttrell — Fri, 02 Dec 2016 16:56:54 GMT

Write a PROC SQL and WHERE using a subquery on the dataset doing a count(*) as count on each scenario, where count >=10

Re: subsetting

mkeintz — Sun, 04 Dec 2016 12:37:01 GMT

You have shown us the source table. Now show us explicitly what the result table would look like.

Re: subsetting

superbibi — Mon, 05 Dec 2016 16:21:11 GMT

Hi, I hope the table output shown as the attached, keeping only values depending on "studyeye". Many thanks.