topic Re: "Randomizer" for variables :) in SAS Data Management

"Randomizer" for variables :)

MajaFerencakovic — Sun, 07 Dec 2014 13:13:50 GMT

Hi guys!

I believe this question is already answered somewhere but I have spent last 2 hours finding elegant solution for this problem.

Imagine data set with for example 200 individuals and every individual has 40000 variables (genotypes).

ID	Var_1	Var_2	Var_3	Var_...	Var_40000
1	A	A	B	...	C
2	A	C	A	...	A
3	B	A	A	...	B
...	...	...	...	...	...
200	B	B	C	...	B

I want to keep all individuals but would like to randomly select 1000 variables (genotypes).

Ok, I can do proc transpose, then randomly select and transpose back, but sometimes I have 1000000 variables and transposing is really lost of time.

Selection of variables by list would be awesome, as well direct selection based on some random function.

I believe first way is more likable and I would be very happy if someone could give a hand with this one!

Thanks!

Re: "Randomizer" for variables :)

data_null__ — Sun, 07 Dec 2014 13:59:57 GMT

proc contents data=sashelp.heart noprint out=names(keep=name);
   run; 
proc surveyselect sampsize=3 data=names out=sample;
   run; 
%let keepvars=;
proc sql noprint; 
   select name into :keepvars separated by ' ' from sample;
   quit; 
   run; 
%put NOTE: &=keepvars;
   run; 
NOTE: KEEPVARS=AgeCHDdiag Sex Smoking

Re: "Randomizer" for variables :)

MajaFerencakovic — Sun, 07 Dec 2014 14:09:08 GMT

OK, that works but can you give me a hint about last step which would be having heart data only with this three variables?

Thanks!

Re: "Randomizer" for variables :)

MajaFerencakovic — Sun, 07 Dec 2014 14:11:50 GMT

sorry! I got it! I just copy paste note to data step!

Sunday afternoon...

Thanks!

Re: "Randomizer" for variables :)

data_null__ — Sun, 07 Dec 2014 14:21:56 GMT

data randomvars;

set sashelp.heart(keep=&keepvars);

run;