CODEWITHSUNDEEP
rsample
banan3'14
banan3'14

Reputation: 4996

take a sample that has a specific mean

Let's say I have a population like {1,2,3, ..., 23} and I want to generate a sample so that the sample's mean equals 6.

I tried to use the sample function, using a custom probability vector, but it didn't work:

population <- c(1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23)
mean(population)
minimum <- min(population)
maximum <- max(population)
amplitude <- maximum - minimum 
expected <- 6
n <- length(population)
prob.vector = rep(expected, each=n)
for(i in seq(1, n)) {
  if(expected > population[i]) {
    prob.vector[i] <- (i - minimum) / (expected - minimum)
  } else {
    prob.vector[i] <- (maximum - i) / (maximum - expected)
  }
}
sample.size <- 5
sample <- sample(population, sample.size, prob = prob.vector)
mean(sample)

The mean of the sample is about the mean of the population (oscillates around 12), and I wanted it to be around 6.

A good sample would be:

The problem is different from sample integer values in R with specific mean because I have a specific population and I can't just generate arbitrary real numbers, they must be inside the population.

The plot of the probability vector: plot

Upvotes: 3

Views: 759

Answers (1)

Onyambu
Onyambu

Reputation: 79188

You can try this:

m = local({b=combn(1:23,5);
           d = colMeans(b);
           e = b[,d>5.5 &d<6.5];
           function()sample(e[,sample(ncol(e),1)])})
m()
[1] 8 5 6 9 3
m()
[1]  6  4  5  3 13

breakdown:

b=combn(1:23,5) # combine the numbers into 5
d = colMeans(b) # find all the means
e = b[,d>5.5 &d<6.5] # select only the means that are within a 0.5 range of 6
sample(e[,sample(ncol(e),1)]) # sample the values the you need

Upvotes: 2

Related Questions