Date:         Fri, 6 Nov 2009 12:48:24 -0500
Reply-To:     Gerhard Hellriegel <gerhard.hellriegel@T-ONLINE.DE>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Gerhard Hellriegel <gerhard.hellriegel@T-ONLINE.DE>
Subject:      Re: how to split data set
Content-Type: text/plain; charset=ISO-8859-1

you cannot split the dataset according to a variable, only to the CONTENTS
of that variable. For that it is necessary to find out what that contents
are and how many datasets you want to have as result.
Example:
you have a dataset with 1 million obs and the split variable is a primary
key (no duplicates). If you want to use that variable as split variable,
you cannot use simply the content 1:1, because then you will get 1 million
datasets! If there are let's say 10 different contents in that variable,
next question is, how they are distributed. If you have 9 different values
in the first 9 obs and the rest all have the same content, that is not
what you want. In that case you would get 10 datasets, 9 with 1 obs and
one with 1 million.
If the 10 obs are evenly distributed and have many values, e.g a counter,
there might be a way to get the smalles, the biggest and divide that range
in 10 parts. Then use that as delimiter like

  if split <1000 then output out1;
  if 1000 <= split < 2000 then output out2;
  if ....

Hope that helps.

Gerhard





On Fri, 6 Nov 2009 09:24:32 -0800, nickyli <nickyli0753@GMAIL.COM> wrote:

>On 11月6日, 下午10时58分, gerhard.hellrie...@T-ONLINE.DE (Gerhard
>Hellriegel) wrote:
>> don't know for sure what you mean. It might be the question, how you can
>> split a dataset by the VALUES (contents) of a variable.
>> If you know the contents, you could use something like:
>>
>> data a b;
>> �  set big;
>> �  if x="a" then output a;
>> �  else output b;
>> run;
>>
>> if not, you could also use a macro:
>>
>> %macro split(ds=,var=);
>> �  %let no=0;
>> �  proc sort data=&ds(keep=&var) out=temp nodupkey;
>> �  �  by &var;
>> �  run;
>> �  data _null_;
>> �  �  set temp;
>> �  �  call symput("v"!!put(_n_,8. -L),&var);
>> �  �  call symputx("no",_n_);
>> �  run;
>>
>> �  data
>> �  �  %do i=1 %to &no;
>> �  �  �  out&i
>> �  �  %end;
>> �  �  ;
>> �  �  set &ds;
>> �  �  %do i=1 %to &no;
>> �  �  �  if &var = "&&v&i" then output out&i;
>> �  �  �  �  �  � /* if it is numeric, without quotes, or add another
>> �  �  �  �  �  �  �  parameter for the type and add a bit of macro
logic */
>> �  �  %end;
>> �  run;
>> %mend;
>> %split(ds=big,var=split_var);
>>
>> (don't know if that works...)
>>
>> Gerhard
>>
>>
>>
>>
>>
>> On Fri, 6 Nov 2009 14:27:51 +0000, Murphy Choy <gola...@GMAIL.COM>
wrote:
>> >Hi nicky,
>>
>> >For this type of sampling, please refer to the proc surveyselect
>> documentations.
>>
>> >------Original Message------
>> >From: nickyli
>> >Sender: SAS(r) Discussion
>> >To: SA...@LISTSERV.UGA.EDU
>> >ReplyTo: nickyli
>> >Subject: how to split data set
>> >Sent: Nov 6, 2009 10:21 PM
>>
>> >hi,
>> > �  how to get several data set according to split a big data set by a
>> >variable?can it realize by using macro?
>> > �  thanks a lot
>>
>> >Sent from my BlackBerry Wireless Handheld
>>
>> >--
>> >Regards,
>> >Murphy Choy
>>
>> >Certified Advanced Programmer for SAS V9
>> >Certified Basic Programmer for SAS V9- 隐藏被引用文字 -
>>
>> - 显示引用的文字 -
>
>as like i have a big data set qttndist,and it had a variable stkcd.the
>question is that split the qttndist to sevaral small data sets
>according to stkcd.
>thanks