CODEWITHSUNDEEP
rstatisticsdistribution
anova
anova

Reputation: 125

Error in R: x must be numeric

i try to write a program in r "generate a random sample from any distribution using function". but it shows "Error in hist.default(xbars) : 'x' must be numeric" my program is here

sim.clt <- function(n, ran.func,..., simsize,...)
{
  xbars<-vector()
  for(i in 1:simsize=simsize)
  {
    x<-function(ran.func)

      xbars[i]<-mean(x)
 }
par(mfrow=c(2,1))
hist(xbars)
qqnorm(xbars)
return(xbars)
}
sim.out<-sim.clt(n=20,ran.func="rexp",simsize=5000)
shapiro.test(sim.out)

#

i am new in r programming, so can't figure it out, how to solve the problem. thanks...

Upvotes: 0

Views: 1411

Answers (1)

Gavin Simpson
Gavin Simpson

Reputation: 174898

There are lots of things wrong here.

for(i in 1:simsize=simsize)

should be throwing an error:

> for(i in 1:simsize=simsize) { print(i)}
Error: unexpected '=' in "for(i in 1:simsize="

Better is

for(i in seq_len(simsize))

Then

x <- function(ran.func)

is not doing what you thought it was; it is returning a function with xbars[i]<-mean(x) as its body, as in:

> x <- function(ran.func)
+ 
+       xbars[i]<-mean(x)
> x
function(ran.func)

      xbars[i]<-mean(x)
> is.function(x)
[1] TRUE

I think you wanted to call ran.func so you may need

FUN <- match.fun(ran.func)
x <- FUN()

But that will fail because you don't seem to be passing any argument for ran.func to work, even just n in the example using rexp.

The error message stems from this last point. You defined xbars to be the empty vector(), which by default created an empty logical vector:

> xbars <- vector()
> xbars
logical(0)
> is.numeric(xbars)
[1] FALSE

Now, this wouldn't have been a problem if you hadn't made the error in defining x (recall xbars[i]<-mean(x) is now in the body of the function x and has never been explicitly called), which means xbars remains a empty logical vector. As that is not numeric, hist throws the error you are seeing.

Another error is that you can't use ... in the function definition twice. Are you trying to have the first contain arguments to pass to ran.func and the second ... to be for something else. You just can't do that in R.

Is this what you wanted?

sim.clt <- function(n, ran.func, simsize, ...) {
  ## ... passed to ran.func for other parameters of distribution
  xbars <- numeric(simsize)
  for(i in seq_len(simsize)) {
    FUN <- match.fun(ran.func)
    x <- FUN(n = n, ...)
    xbars[i] <- mean(x)
  }
  ## plot
  op <- par(mfrow=c(2,1))
  on.exit(op)
  hist(xbars)
  qqnorm(xbars)
  xbars
}

> sim.out<-sim.clt(n=20,ran.func="rexp",simsize=5000)
> shapiro.test(sim.out)

    Shapiro-Wilk normality test

data:  sim.out
W = 0.9867, p-value < 2.2e-16

Upvotes: 6

Related Questions