topic Re: Dividing into number of dataset thatcontais equal number of records in SAS Programming

Dividing into number of dataset thatcontais equal number of records

MSK4 — Mon, 14 Sep 2020 11:56:03 GMT

hi team,

Here im having 3lakhs 30 k records in dataset i want to divide them equally into n number of datasets.

Please looking for code many thanks

Re: Dividing into number of dataset thatcontais equal number of records

Tom — Mon, 14 Sep 2020 13:22:15 GMT

@MSK4 wrote:

hi team,

Here im having 3lakhs 30 k records in dataset i want to divide them equally into n number of datasets.

Please looking for code many thanks

Not sure what a "lakhs" is, but assuming it is just some unit of counting this is a common question.

Here is link to a previous question about splitting datasets by the number of observations that has an answer:

https://communities.sas.com/t5/Statistical-Procedures/Splitting-dataset-based-on-total-observations/td-p/104460

Re: Dividing into number of dataset thatcontais equal number of records

jimbarbour — Mon, 14 Sep 2020 14:45:31 GMT

1 lakh = 100,000

30 k = 30,000

So @MSK4 is talking about 330,000 records.

Jim

Re: Dividing into number of dataset thatcontais equal number of records

mkeintz — Mon, 14 Sep 2020 15:56:55 GMT

@MSK4 wrote:

hi team,

Here im having 3lakhs 30 k records in dataset i want to divide them equally into n number of datasets.

Please looking for code many thanks

You haven't said anything about the data except its size. If the data are already in random order and there is no "group" identifier then this is the simplest way, assuming you want 6 datasets:

data data1 data2 data3 data4 data5 data6;
  set have;
  select (mod(_n_-1,6)+1);
    when (1) output data1;
    when (2) output data2;
    when (3) output data3;
    when (4) output data4;
    when (5) output data5;
    when (6) output data6;
  end;
run;

If you want to make the number of datasets a parameter, then you could:

%let ndatasets=6;
filename tmp temp ;
data _null_;
  file tmp;
  length statement $200;
  statement='DATA ';
  do i=1 to &ndatasets;
    statement=catx(' ',statement,cats('data',i));
  end;
  put statement ';';
  put 'set have;';
  put "select (mod(_n_-1,&ndatasets)+1);";
  do i=1 to &ndatasets;
    put "when(" i ") output data" i ";" ;
  end;
  put "end;" / "run;" ;
run;
%include tmp / source2;

The program writes sas code to a temporary file (sas will delete it at the end fo the session). Then the %include statement calls it for execution. The "/source2" option tells sas to print the included statements in the sas log.

Re: Dividing into number of dataset thatcontais equal number of records

MSK4 — Tue, 15 Sep 2020 03:52:28 GMT

yes 3,30,000

Re: Dividing into number of dataset thatcontais equal number of records

andreas_lds — Tue, 15 Sep 2020 08:21:09 GMT

Why do you want to divide a rather small dataset at all?