Adding random numbers to the end of a vector

Question

I am trying to repeatedly compute the median of a random vector, constructing a vector of all the medians. Why do I get a NULL result from my code below, in which I repeatedly compute the median and add it to the end of the vector m?

medainfunc<-function(n,mu,sigma,iterate){
  m<-c()
  for (i in itreate){
    x<-rnorm(n,mu,sigma)
    y<-median(x)
    m<-c(m,y)
  }
  m
}
medianfunc(10,3,15,10000)
NULL

Check [here](http://stackoverflow.com/questions/32366694/adding-values-to-the-vector-inside-for-loop-in-r). Also try with `for(i in seq(iterate))` Check the typo in `iterate` and `itreate` and `medainfunc` and `medianfunc` as well. — akrun, Sep 03 '15 at 05:36

akrun · Accepted Answer · 2015-09-03T06:05:52.450

3

We have multiple typos in the OP's code, i.e. iterate vs. itreate and calling medainfunc while the original function is medianfunc. In addition, we are providing only a single input value for 'iterate', so seq(iterate may be we want inside the function. Otherwise, we get a single value output.

medianfunc<-function(n,mu,sigma,iterate){
  m<-c()
  for (i in seq(iterate)){
    x<-rnorm(n,mu,sigma)
    y<-median(x)
    m<-c(m,y)
  }
  m
}

set.seed(25)
medianfunc(10,3,15, 5)
#[1]  0.9770646 -6.4852741  4.6768291 -6.4167869  5.3176253

This could be vectorized by getting the rnorm of 'n*iterate' values. Convert this to a matrix and use colMedians from library(matrixStats).

medianfunc1 <- function(n, mu, sigma, iterate){
  m1 <- matrix(rnorm(n*iterate, mu, sigma), ncol=iterate)
  library(matrixStats)
  colMedians(m1)
}

set.seed(25)
medianfunc1(10,3,15, 5)
#[1]  0.9770646 -6.4852741  4.6768291 -6.4167869  5.3176253

edited Sep 03 '15 at 06:05

answered Sep 03 '15 at 05:41

akrun

874,273
37
540
662

should he also add `return` to the customized function? – Diego Aguado Sep 03 '15 at 05:43
2

@DiegoAgher Wrapping `return` around 'm' is not needed. For clarity, it can be used. – akrun Sep 03 '15 at 05:44
@akrun I corrected those typos. But still I get `null` even when I used `seq(iterate)` – clarkson Sep 03 '15 at 05:57
@clarkson Using the function in my solution returns the values as I showed. You can check whether you are still having some typos. – akrun Sep 03 '15 at 05:58
@akrun The problem was wrote as `m}` in single line. Why can't I do it? – clarkson Sep 03 '15 at 06:05
@clarkson Sorry, didn't get your comment. If you are writing in a single line, then use `;` to separate each if I understand your comment. – akrun Sep 03 '15 at 06:10

josliber · Answer 2 · 2015-09-03T06:08:24.937

Building a vector one-by-one is both inefficient (see the second circle of the R inferno) and code-intensive. Instead, you can repeat a randomized operation a certain number of times in a single line of code with the replicate function:

medianfunc <- function(n, mu, sigma, iterate) {
  replicate(iterate, median(rnorm(n, mu, sigma)))
}
set.seed(25)
medianfunc(10,3,15, 5)
# [1]  0.9770646 -6.4852741  4.6768291 -6.4167869  5.3176253

score 0 · Answer 3 · answered Sep 03 '15 at 05:40

I dont know if this is just the code you wrote on the post but the function is called medainfunc and then on your code you are calling the medianfunc. That might be just one error, the other thing I noticed is that you should add return at the end of your function like so:

medainfunc<-function(n,mu,sigma,iterate){
   m<-c()
   for (i in itreate){
     x<-rnorm(n,mu,sigma)
     y<-median(x)
     m<-c(m,y)
     }
   return(m)
   }

Still a typo, as highlighted by akrun. – Sep 03 '15 at 05:40 — , Sep 03 '15 at 05:40

Adding random numbers to the end of a vector

3 Answers3