topic Re: weighted random selection in SAS Programming

weighted random selection

PSIOT2 — Mon, 05 Dec 2022 08:32:07 GMT

Hello,

I would like to extract a sample with these fields: patient_id, Size of nodule (3 classes) and diagnostic (cancer or benign) from population of 1731 patients.

The sample will be done without replacement and i need to have this distribution in final sample for :

size of nodule:

01:[4-10mm] => 37.5%

02:[10-20mm] => 37.5%

03:[20-30mm] => 25%

Diagnostic:

Cancer => 33.33%

Benign => 66.67%

The final sample will include 200 patients.

Thanks a lot for your help.

Re: weighted random selection

sbxkoenk — Tue, 06 Dec 2022 16:15:45 GMT

Hello,

No time to write an example (on simulated data) for you right now.

But you need :

PROC SURVEYSELECT
with STRATA statement
with ALLOC=SAS-data-set option.

See here :
https://go.documentation.sas.com/doc/en/pgmsascdc/9.4_3.5/statug/statug_surveyselect_details20.htm

Good luck,

Koen

Re: weighted random selection

webart999ARM — Wed, 07 Dec 2022 00:47:32 GMT

To extract a sample in SAS, you can use the SAMPLE statement in a PROC SURVEYSELECT procedure. Here is an example of how you might use it to extract your sample:

PROC SURVEYSELECT 
    SAMPLE=200
    SEED=12345 
    OUT=sample_data
    METHOD=SRS
    NOPRINT;
    STRATA SizeOfNodule
        (01: 0.375, 02: 0.375, 03: 0.25)
        DIAGNOSTIC
        (Cancer: 0.3333, Benign: 0.6667);
RUN;
This will create a new dataset called sample_data that contains a sample of 200 patients, with the proportions of nodule sizes and diagnostic outcomes that you specified.

The SEED option allows you to specify a seed value for the random number generator, so that you can obtain the same sample each time you run the code. You can change the seed value to any integer that you like.

The METHOD=SRS option specifies that the sample will be selected using simple random sampling without replacement. This means that each patient has an equal probability of being selected, and that each patient can only be selected once.

The STRATA statement allows you to specify the proportions of the sample that should be allocated to each combination of nodule size and diagnostic outcome. In this case, the sample will be stratified by nodule size and diagnostic outcome, with the proportions that you specified for each stratum.

I hope this helps! Let me know if you have any other questions.

Re: weighted random selection

PSIOT2 — Thu, 08 Dec 2022 11:35:02 GMT

Thank you for your answers,

I tried this one but some errors appear:

PROC SURVEYSELECT
N=200
SEED=12345
OUT=sample_data
METHOD=SRS
NOPRINT;
STRATA strateDiam3
(Min: 0.375, Mid: 0.375, Max: 0.25)

ERROR 22-322: Syntax error, expecting one of the following: a name, ;, -, /, :, DESCENDING, NOTSORTED, _ALL_, _CHARACTER_, _CHAR_,
_NUMERIC_.
ERROR 76-322: Syntax error, statement will be ignored.

FinalDiagnosis

(Cancer: 0.3333, Benign: 0.6667);
RUN;