topic Re: Subsampling within Group in SAS Programming

Subsampling within Group

abak — Mon, 12 Mar 2018 23:30:20 GMT

I need to subsample a dataset within groups. Here is an example- two columns.

Group Data

G1 v1

G1 v2

G1 v3

G1 v4

G2 v5

G2 v6

G3 v7

G3 v8

G3 v9

G3 v10

G3 v11

I actually have a lot more data points for some groups. For the groups with more than 2 (for example), I want to only sample 60% of the values from that group in a subset.

How do I accomplish that?

Re: Subsampling within Group

PGStats — Tue, 13 Mar 2018 04:36:17 GMT

You could use proc surveyselect and specify your sample sizes in a secondary dataset

proc sql;
create table sampleSize as
select 
    group,
    max(2, 0.6*count(*)) as _nsize_
from myData
group by group;
quit;

proc surveyselect data=myData selectall out=mySample sampsize=sampleSize;
strata group;
run;

(untested)

Re: Subsampling within Group

abak — Tue, 13 Mar 2018 18:23:34 GMT

Thank you! The only thing missing was round. Easily added!

round(max(2, 0.6*count(*)))