[R] sample() question

[Chen,Shaofei]

I believe you mean
c(sample(1:3,3), sample(1:3, 2, replace = TRUE)
is that right?

However, we cannot see the combination like 33312. This is a problem.


-----Original Message-----
From: Erik Iverson [mailto:eriki at ccbr.umn.edu] 
Sent: Wednesday, August 18, 2010 7:56 PM
To: Chen,Shaofei
Cc: r-help at r-project.org
Subject: Re: [R] sample() question

On 08/18/2010 07:51 PM, Chen,Shaofei wrote:
> Consider an example:
> There are 3 stores (1,2,3), and 5 customers. Each store must have at least
> one customer. Thus a possible combination is 11213. On the other hand, a
> combination 22333 is not what I want
>
> I have considered your solution earlier, but in this case, first three
> customers have to go to store 123 in sequence, and the randomness has been
> reduced. You are right, the sampling procedure is not totally random, but
I
> want to be as random as possible.

So just use my previous solution and permute the result using sample.

>
> Thanks!
>
> -----Original Message-----
> From: Erik Iverson [mailto:eriki at ccbr.umn.edu]
> Sent: Wednesday, August 18, 2010 7:11 PM
> To: Chen,Shaofei
> Cc: r-help at r-project.org
> Subject: Re: [R] sample() question
>
> On 08/18/2010 06:58 PM, Chen,Shaofei wrote:
>> Hello all,
>>
>>
>>
>> I have a question regarding sample() in R. For example, I have a set:
>>
>> set<- c(2,3,5)
>>
>> and I want to draw 5 samples from this set, so replacement is true:
>>
>> m<- sample(set, 5, replace=TRUE)
>>
>> However here comes a problem, for example, I will have (2,3,3,2,5), but I
>> will also  get (3,3,5,5,3) in some cases. This means element 2 has not
> been
>> sampled in this case.
>>
>> The way I want to do is to random sample with replacement, but all
> elements
>> have to be sampled.
>>
>> Any solutions?
>
> Well, if all elements have to be sampled once, then those 3 elements
> will not be random, since they must be present.
>
> So you're really only sampling 2 elements from the set, so just do that.
>
> c(1:3, sample(set, 2, replace = TRUE)
>
>
> although I really wonder what you're trying to do here?
>