R help - Nov 2013 - Randomize two categories testing a specific condition R version 3.0.2 windows 32 bit

Hello everyone I am new to R. At this time I am trying to create a vector
so i can randomize two categories a,b under the following conditions :
Take a random x from random uniform(0,1) if x<0.5 then the vector takes
value a else takes b . This continues until the number of a's-the number of
b's > 2 at this point i want to change the possibility that a's and
b's are
assigned to the vector to 0.2 (if x<0.2 then a). This probability i want to
change again if the number of b's - the number of a's > 2 to 0.8(if
x<0.8
then a). However because i have random numbers i may have the above
situation vice versa. (first b's-a's>2 then a's-b's>2)
.finally sth like
vector<- a,b,a,a,a,b,b,a . I thought to use a for loop and this is my code ;

set.seed(10)
x<- rep(NA, 20)
tmp<- rep(FALSE, 20)for(i in 1:20) {
x[i]<- runif(1)
tmp<- x[i]<0.5if(2 * sum(tmp) - i + 1 > 2) {
tmp[i]<- x[i]<0.2} else {
tmp[i]<- x[i]<0.8}}

Although the code runs without errors it doesnt return what i want . How

am I wrong? Should i have to use next or break statement here or sth else?

please help me.Thanks in advance

	[[alternative HTML version deleted]]

> -----Original Message-----
> From: r-help-bounces at r-project.org [mailto:r-help-bounces at
r-project.org]
> On Behalf Of Manos Spanakis
> Sent: Saturday, November 23, 2013 1:48 AM
> To: R-help at r-project.org
> Subject: [R] Randomize two categories testing a specific condition R
> version 3.0.2 windows 32 bit
> 
> Hello everyone I am new to R. At this time I am trying to create a vector
> so i can randomize two categories a,b under the following conditions :
> Take a random x from random uniform(0,1) if x<0.5 then the vector takes
> value a else takes b . This continues until the number of a's-the
number
> of
> b's > 2 at this point i want to change the possibility that a's
and b's
> are
> assigned to the vector to 0.2 (if x<0.2 then a). This probability i want
> to
> change again if the number of b's - the number of a's > 2 to
0.8(if x<0.8
> then a). However because i have random numbers i may have the above
> situation vice versa. (first b's-a's>2 then
a's-b's>2) .finally sth like
> vector<- a,b,a,a,a,b,b,a . I thought to use a for loop and this is my
code
> ;
> 
> set.seed(10)
> x<- rep(NA, 20)
> tmp<- rep(FALSE, 20)for(i in 1:20) {
> x[i]<- runif(1)
> tmp<- x[i]<0.5if(2 * sum(tmp) - i + 1 > 2) {
> tmp[i]<- x[i]<0.2} else {
> tmp[i]<- x[i]<0.8}}
> 
> Although the code runs without errors it doesnt return what i want . How
> 
> am I wrong? Should i have to use next or break statement here or sth else?
> 
> please help me.Thanks in advance
> 
It is not clear to me what your exact boundary conditions are for changing
probabilities, or even what your real goal is, but maybe something like the
following will get you started.  Stage=1 is before there are greater than 2 
'a's and 2 'b's.

set.seed(3724)
x <- character(0)
p <- .5
stage <- 1
for(i in 1:20){
  x <- c(x, sample(c('a','b'),1,prob=c(p,1-p)))
  suma <- sum(x=='a')
  sumb <- sum(x=='b') 
  #check if first condition is met
  if(suma > 2 & sumb > 2) {
    #select a starting value for changed p value and do this only once
    if(stage == 1) {stage <- 2; if(suma >= sumb) p <-.8 else p <-
.2}
    #check for excess of 'a's or 'b's    if(suma - sumb > 2)
p <- .2
    if(sumb - suma > 2) p <- .8
    }
}      

If this doesn't do what you want, then write back to R-help describing what
your ultimate goal is (i.e., what this method of randomization is supposed to
accomplish), and someone may be able to give you better advice.

Hope this is helpful,

Dan 

Daniel Nordlund
Bothell, WA USA