How to create a loop

Question

Can you explain the structure/idea of loop in R-Code? I don't know where to start with this homework problem. I was able to create the exponential distribution I needed with

> rexp(n=200, rate=0.5) 

but now I need to create 2,000 samples of this rexp and calculate the mean of each sample. Then I need to calculate the mean and variance of the 2,000 sample means. I know how to find the mean and variance of a simple distribution (ex: mean(rexp) and var(rexp)), so my main issue is understanding this concept of a loop and putting it into action.

So I started with entering:

> rexp(n=200,rate=0.5)

Then I named this exponential distribution:

> exdi = rexp(n=200,rate=0.5)

Based off of an example dealing with sums, I entered

>y.exdi=vector(length=2000)
>for(i in 1:2000){y.exdi[ i ]=mean(exdi)}

The R Workspace gave no response to this, so I named the function and tried this:

>Twothou = for(i in 1:2000){y.exdi[ i ]=mean(exdi)}
>mean(Twothou)

But then I got this error message:

[1] NA
Warning message:
In mean.default(Twothou) : argument is not numeric or logical: returning NA

What should I do differently?

Answer 1

You've just created a really nice vector of 2000 elements of the same value - in the means of the mean of your sample brewed for exponential distribution only once. See by checking y.exdi in your console.

If you want to use a loop for this solution, you should create new and new samples in each iteration. You may do that easily with e.g. sapply (wrapper around the for loop) applied to 1:2000:

sapply(1:2000, function(x) mean(rexp(n = 200, rate = 0.5)))

Or rather directly calling e.g. replicate (that was intended to be used for such cases):

replicate(2000, mean(rexp(n = 200, rate = 0.5)))

Answer 2

I think you should put the random number generator rexp inside your for loop:

y.exdi=vector(length=2000)
for(i in 1:2000){
    y.exdi[ i ]=mean(rexp(n=200,rate=0.5))
}

Otherwise, for each i you just calculate the mean of a fixed vector defined outside the loop, instead of new ones generated each time the index i increases. Having that codes above, you can calculate the mean using mean(y.exdi), which gives you the average of the 2000 "averages".