topic Trying to split a dataset . in SAS Programming

Trying to split a dataset .

dennis_oz — Fri, 18 Oct 2019 04:52:16 GMT

Hi All,

I am trying to just extract 10 observations for a dataset and chose to do it in the below method.

But it does not seem to work . Can you help me with my logic below

Also before using the do while i tried using do until but the code would not process . i put put statements and looked in the log

but do until would not even go past 'point 1' ..

thanks .

data test;
set sashelp.cars;
counter+1;
run;

data test_out;
set test;
put 'point1';

do while (counter =<10);
output;
put 'point2';
counter +1;
end;
run;

Re: Trying to split a dataset .

PeterClemmensen — Fri, 18 Oct 2019 04:54:42 GMT

You simply want to extract the first 10 obs from a data set, correct?

Re: Trying to split a dataset .

dennis_oz — Fri, 18 Oct 2019 05:21:14 GMT

yes , if possible using do while

Re: Trying to split a dataset .

PeterClemmensen — Fri, 18 Oct 2019 05:33:49 GMT

You can use Do While like this

data want;
    set sashelp.cars;
    do while (c < 10);
        output;
        c+1;
    end;
run;

However, problems like this are easier solved with the automatic _N_ variable

data want;
    set sashelp.cars;
    if _N_ <= 10;
run;

Re: Trying to split a dataset .

Shmuel — Fri, 18 Oct 2019 05:40:04 GMT

You don't need a counter, sas has its own implicit counter _N_:

You simply do:

data want;
  set have(obs=10);
run;

or if you insist use do while then

data want;
  set have;
       do while _N_ le 10;
            output;
      end;
run;

data want;
   set have;
        counter +1;
        do while counter le 10;
             output;
       end;
run;

in your code you probably got 10 times same first observation.

Re: Trying to split a dataset .

Amir — Fri, 18 Oct 2019 09:01:21 GMT

Note that some responses output the first observation 10 times and some require brackets around conditions.

If you really want the first 10 observations output, as opposed to the first observation output 10 times, using a do-while or a do-until loop then the set statement should be in the loop so that a new observation is read from the input data set in each iteration of the loop.

As you might already be aware, the SAS data step automatically iterates over a data set in a set statement without the need for explicit looping code. In which case my preferred solution is the obs=10 data set option demonstrated by @Shmuel.

Otherwise, if you are learning about loops and how they can be used to read data sets then you could try something like:

/* do-until loop */
data want(drop = counter);
   do until(counter ge 10);
      set sashelp.class;
      output;
      counter + 1;
   end;

   /* prevent any more data being read */
   stop;
run;


/* do-while loop */
data want(drop = counter);
   counter + 1;

   do while(counter le 10);
      set sashelp.class;
      output;
      counter + 1;
   end;

   /* prevent infinite looping */
   stop;
run;

/* do-iterative loop */
data want(drop = counter);
   do counter = 1 to 10;
      set sashelp.class;
      output;
   end;

   /* prevent any more data being read */
   stop;
run;

Amir.

Re: Trying to split a dataset .

ballardw — Fri, 18 Oct 2019 21:06:06 GMT

If you want to use a specific set of sequential observations in almost any procedure there is no need to create an additional data set at all. Data set options FIRSTOBS and OBS will do it.

proc print data=sashelp.cars(firstobs=1 obs=10);
run;

proc print data=sashelp.cars(firstobs=5 obs=15);
run;

Re: Trying to split a dataset .

hashman — Sat, 19 Oct 2019 03:53:37 GMT

@dennis_oz:

You can do both WHILE or UNTIL, and the auto-dropped counter _ N_ (or _IORC_ if you wish) can be incorporated into the loop specifications:

data want ;                                                                                                                             
  do _n_ = 1 by 1 while (_n_ <= 10) ;                                                                                                   
* do _n_ = 1 by 1 until (_n_  = 10) ;                                                                                                   
    set sashelp.class ;                                                                                                                 
    output ;                                                                                                                            
  end ;                                                                                                                                 
  stop ;                                                                                                                                
run ;

However, your requirement of using WHILE or UNTIL doesn't make much sense because if you use FROM-TO specification, WHILE is applied implicitly:

data want ;                                                                                                                             
  do _n_ = 1 to 10 ;                                                                                                                    
    set sashelp.class ;                                                                                                                 
    output ;                                                                                                                            
  end ;                                                                                                                                 
  stop ;                                                                                                                                
run ;

I agree with other responders that for a thing like that, you only need the OBS=10 data set option. Or, if you need to select N consecutive records from anywhere in the file, you can combine FIRSTOBS= with OBS=. For example, to select 10 records starting from record 5:

data want ;                                                                                                                             
  set sashelp.class (firstobs=5 obs=14) ;                                                                                               
run ;

Kind regards

Paul D.