Date:         Tue, 9 Jun 2009 19:30:50 +0800
Reply-To:     Murphy Choy <goladin@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Murphy Choy <goladin@GMAIL.COM>
Subject:      Re: Sampling without replacement
Comments: To: karma <dorjetarap@googlemail.com>
In-Reply-To:  <c6b899fd0906090421o54b7ab3qaf43bf9e06fb4727@mail.gmail.com>
Content-Type: text/plain; charset=ISO-8859-1

Hi Karma,

The following will do as well for samples of equal size. The one I was
sending above was for just 3 sample without thinking about sample size.

*data a b c;
set temp;

if ranuni(54321) <= 0.33 then output a;

else do;

                 if ranuni(54321) <= 0.5 then output b;
                 else output c;

end;

run;*

As well as this,

*data a b c;
set temp;

if ranuni(12345) <= 0.33 then output a;
else if ranuni(12345) <= 0.5 then output b;
else output c;

run;*


Regards,
Murphy

On Tue, Jun 9, 2009 at 7:21 PM, Murphy Choy <goladin@gmail.com> wrote:

> Hi Karma,
>
> Agreed. I am merely suggesting a possibility.
>
> Regards,
> Murphy
>
>
> On Tue, Jun 9, 2009 at 6:20 PM, karma <dorjetarap@googlemail.com> wrote:
>
>> Hi Murphy,
>>
>> I think you should store the results of ranuni in a temp variable
>> before splitting the dataset as calling the ranuni function again will
>> return a new value. Otherwise dataset c will on average contain only
>> half the number of the records in dataset b.
>>
>>
>> data a b c;
>>   set Multi_Strat1_List;
>>    _temp = ranuni(12345) ;
>>   if _temp <= 0.33 then output a;
>>   else if _temp <= 0.66 then output b;
>>   else output c;
>>   run ;
>>
>>
>>
>> 2009/6/9 Murphy Choy <goladin@gmail.com>:
>> > Hi,
>> >
>> > Why not use data step?
>> >
>> > data a b c;
>> > set Multi_Strat1_List;
>> >
>> > if ranuni(12345) <= 0.33 then output a;
>> > else if ranuni(12345) <= 0.66 then output b;
>> > else output c;
>> >
>> > run;
>> >
>> > Regards,
>> > Murphy
>> >
>> > On Tue, Jun 9, 2009 at 5:39 PM, fabergoogle
>> > <fabrizio_marrama@bradycorp.com>wrote:
>> >
>> >> Hi all,
>> >>
>> >> I want to create 3 different samples whose union should give me back
>> >> the whole dataset.
>> >> Therefore an individual can only belong to one of those 3 samples. He
>> >> cannot appear in any of the other two samples.
>> >> I applied the proc surveyselect specifying rep=3. I noticed that an
>> >> individual is present in 2 samples. Why?
>> >>
>> >> How can I have 3 different samples?
>> >>
>> >> Thank you.
>> >>
>> >>
>> >> PROC SURVEYSELECT data = Multi_Strat1_List
>> >>        out = Multi_Strat1_Control_Sample
>> >>        method = srs
>> >>                rep=3
>> >>        seed = 735119000
>> >>        sampsize = 500;
>> >>        strata Deciles ContRev_Band;
>> >> RUN;
>> >>
>> >
>> >
>> >
>> > --
>> > Regards,
>> > Murphy Choy
>> >
>> > Certified Advanced Programmer for SAS V9
>> > Certified Basic Programmer for SAS V9
>> > DataShaping Certified SAS Professional
>> >
>>
>
>
>
> --
> Regards,
> Murphy Choy
>
> Certified Advanced Programmer for SAS V9
> Certified Basic Programmer for SAS V9
> DataShaping Certified SAS Professional
>



--
Regards,
Murphy Choy

Certified Advanced Programmer for SAS V9
Certified Basic Programmer for SAS V9
DataShaping Certified SAS Professional