Scrambling a SAS Dataset

[alan0101]

Hi

I have a dataset with 10 variables and 600k records.  It is the end result of a number of preceding processes, and one of the effects of these processes is that the dataset is ordered by age.  Is there a simple sas procedure which can be written to get rid of the sorting ?  I don't want to do say 3 proc surveyselects and then put the datasets together again

[SASKiwi · Accepted solution]

This will order your data randomly:

data want;

  set have;

  scramble_var = ranuni(0);

run;

proc sort data = want;

  by scramble_var;

run;

[alan0101]

Great thanks SASKiwi !

[RW9]

Hi,

Just for laughs, try proc whisk?

For the above, you could do in one step:

proc sql;

  create table want as select * from sashelp.cars order by ranuni(0);

quit;

Also, I would ask why do you want to scramble the dataset.  Generally ordered datasets would be easier to use, maybe you just need to change the order or retain original ordering:

data original;

     set original;

     ord=_n_;

run;

Then later on sort by _n_ to get it to the original sort.

[alan0101]

Also great thanks RW 9 !  I want to scramble the datasets so that the customers are not called in any predefined sequence or bias which is built in via the preceding datasteps