R help - Apr 2014 - Generate Binary Matrix

I am trying to generate a binary matrix where every row in the matrix is
guaranteed to have at least one 1. Ideally, I would like most rowSums  to be
equal to 2 or 3 with some 1s and some 4s. But, rowSums cannot be equal to 0.

I can tinker with the vector of probability weights, but in doing so (in the way
I am doing it) this causes for more rowSums to be equal to 4 than I ideally
would like, but this never helps to guarantee a rowSum will not be equal to 0. I
could post-hoc tinker with any rows that are all 0, but seems like that may be
just inefficient.

Below is sample code, any ideas on how to best tackle this?

Harold



dimMat <- matrix(0, 1000, 4)
for(i in 1:1000){
                dimMat[i, ] <- sample(c(0,1), 4, replace = TRUE, prob = c(.3,
.7))
                }

table(rowSums(dimMat))

	[[alternative HTML version deleted]]

Doran, Harold <HDoran <at> air.org> writes:
> I am trying to generate a binary matrix where 
 row 
in the matrix is guaranteed to have at 
least one 1.
> Ideally, I would like most rowSums  to be equal 
 2 or 3 
with some 1s and some 4s. But, 
rowSums cannot be equal
> to 0.
> 
> I can tinker with the vector of probability weights, 
but in 
 so (in the way I am doing it) this 
causes for
> more rowSums to be equal to 4 than I ideally would 
, but this never helps to guarantee a 
rowSum will not be
> equal to 0. I could post-hoc tinker with any rows 
 are all 
0, but seems like that may be just 
inefficient.
> 
> Below is sample code, any ideas on how to best 
tackle this?
> 
> Harold
> 
> dimMat <- matrix(0, 1000, 4)
> for(i in 1:1000){
>     dimMat[i, ] <- sample(c(0,1), 4, replace = TRUE, prob = c(.3, .7))
>                 }
> 
> table(rowSums(dimMat))

Wht don't you sample from the distribution of row sums 
for each row and then distribute that many 1's randomly
among the columns.

Ken

You could randomly assign 1 to a single column in each row and
then use binomial draws on the remaining 0's:
> set.seed(42)
> dimMat <- matrix(0, 1000, 4)
> dimMat[cbind(1:1000, sample.int(4, 1000, replace=TRUE))] <- 1
> dimMat[dimMat<1] <- sample(0:1, 3000, replace=TRUE, prob=c(.6,
.4))
> table(rowSums(dimMat))
  1   2   3   4 
217 402 310  71 
> colSums(dimMat)
[1] 551 586 533 565
> sum(dimMat)
[1] 2235

-------------------------------------
David L Carlson
Department of Anthropology
Texas A&M University
College Station, TX 77840-4352

-----Original Message-----
From: r-help-bounces at r-project.org
[mailto:r-help-bounces at r-project.org] On Behalf Of Doran, Harold
Sent: Wednesday, April 9, 2014 3:56 PM
To: r-help at r-project.org
Subject: [R] Generate Binary Matrix

I am trying to generate a binary matrix where every row in the
matrix is guaranteed to have at least one 1. Ideally, I would
like most rowSums  to be equal to 2 or 3 with some 1s and some
4s. But, rowSums cannot be equal to 0.

I can tinker with the vector of probability weights, but in
doing so (in the way I am doing it) this causes for more rowSums
to be equal to 4 than I ideally would like, but this never helps
to guarantee a rowSum will not be equal to 0. I could post-hoc
tinker with any rows that are all 0, but seems like that may be
just inefficient.

Below is sample code, any ideas on how to best tackle this?

Harold



dimMat <- matrix(0, 1000, 4)
for(i in 1:1000){
                dimMat[i, ] <- sample(c(0,1), 4, replace = TRUE,
prob = c(.3, .7))
                }

table(rowSums(dimMat))

	[[alternative HTML version deleted]]

______________________________________________
R-help at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible
code.