simulating binary variables

Dear list

As a relative newbie to R I am after some basic help.

I am wanting to simulate a data set consisting of a Y variable and several X variables, all either binary or discrete. I am wondering how to go about doing this and have failed to find anything about this in the R -help.

Thanks in advance 
Laura

Laura,

have a look at help(rbinom) and help(rpois).

Uli Flenker

	Institute of Biochemistry
	German Sport University Cologne
	Carl-Diem-Weg 6
	50933 Cologne

	Phone 0049/221/4982-506

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

I am wanting to simulate a data set consisting of a Y variable

and several X variables, all either binary or discrete. I am
 wondering how to go about doing this and have failed to find
 anything about this in the R -help.

Try runif() in the base package.  It generates random numbers
from a uniform distribution.  So, for example, if you want a
binary variable with an expectation of .75, and 1000
observations, say:

runif(1000)<=.75

Or, if you want to see the numbers right away:

(runif(1000)<=.75)+0

To generate a factor with several levels, you can apply cut() to
runif().  That may be sufficient, but note that factors are
"categorical variables."

Of course, you assign these to variables, e.g.,

x1 <- runif(1000)<=.75

and then use these in your model.

Jon

Jonathan Baron, Professor of Psychology, University of Pennsylvania
Home page:            http://www.sas.upenn.edu/~baron
Questionnaires:       http://www.psych.upenn.edu/~baron/qs.html
Psychology webmaster: http://www.psych.upenn.edu/
R page:               http://finzi.psych.upenn.edu/
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

As the previous mail, rbinom and rpois are your best bets for binomial and
discrete.
If you use runif or rnorm you will get continuos variables which you can
convert to discrete by using round(a*runif(100) +b)  mod m [but you run the
risk of getting a cycle of numbers if you're not careful].

If you got a non-standard distribution you could use sample. For eg
sample(0:1, 5000, replace=T) produces the similiar result as binomial with
probability 0.5. More interestingly, say if you want to simulate n obs from
a distributon that place a an equal mass on the first 100 prime numbers, the

x <- c( 2, 3,5,7,11, 13, 17,   19 ,  ... .........  ,521,  523, 541 )
y <- sample( x, n, replace=T) and you can turn the replace =F is you want
sampling without replacement

On 08/09/02 13:28, laura at bayesian-bay.freeserve.co.uk wrote:

I am wanting to simulate a data set consisting of a Y variable

 and several X variables, all either binary or discrete. I am
 wondering how to go about doing this and have failed to find
 anything about this in the R -help.

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._