How to simulate data and visualize in a single R function - r

Question

I'm using replicate to simulate distributions in R and visualize how they change with different parameters (e.g., rbinom(100,1,0.5) vs. rbinom(100,1,0.01)).
I'd like to do all of this within a single function that 1. simulates replicates, 2. sets the plotting dimensions and parameters, and 3. loops through and draws density curves.
In separate pieces, this code works no problem:
n <- 100
d <- as.data.frame(
replicate(n,
expr = rbinom(n, 1, 0.5),
simplify = F)
)
colnames(d) <- 1:n
plot( NULL, xlim = c( min(d)-0.5, max(d)+0.5), ylim = c(0,2))
for(i in 1:n) lines( density( d[,i]) )
But inside a function, only a single density curve is returned:
plotcurves <- function(n, distr, ymax) {
d <- as.data.frame(
replicate(n,
expr = distr,
simplify = F)
)
colnames(d) <- 1:n
plot( NULL, xlim = c( min(d)-0.5, max(d)+0.5), ylim = c(0,ymax))
for(i in 1:n) lines( density( d[,i]) )
}
plotcurves(n = 100, distr = rbinom(100, 1, 0.5), ymax = 2)
The solution seems like it would be very simple but I cannot seem to find it.
What do I need to do to fix the code OR does a function like this already exist that I am unaware of?

Answer 1

The problem is that in your function, distr is evaluated before it reaches the call to replicate. You can see this if you make a variation of the function that just returns the data frame d instead of plotting it:
show_d <- function(n, distr, ymax)
{
d <- as.data.frame(
replicate(n,
expr = distr,
simplify = F)
)
return(d)
}
show_d(n = 3, distr = rbinom(5, 1, 0.5), ymax = 2)
#> c.1L..0L..1L..1L..1L. c.1L..0L..1L..1L..1L..1 c.1L..0L..1L..1L..1L..2
#> 1 1 1 1
#> 2 0 0 0
#> 3 1 1 1
#> 4 1 1 1
#> 5 1 1 1
You'll notice the columns are all the same. Effectively, the call to rbinom was evaluated then passed to replicate, which is the same as calling replicate(3, c(1, 0, 1, 1, 1)). So you are plotting all the lines - it's just that the lines are all the same.
What you need to do inside a function is to ensure that distr is passed as a call to replicate rather than being evaluated and sent as a vector. You can do this using match.call() and extracting the third element (which is the second parameter):
plotcurves <- function(n, distr, ymax) {
mc <- match.call()[[3]]
d <- as.data.frame(
replicate(n,
expr = mc,
simplify = F)
)
colnames(d) <- 1:n
plot( NULL, xlim = c( min(d)-0.5, max(d)+0.5), ylim = c(0,ymax))
for(i in 1:n) lines( density( d[,i]) )
}
plotcurves(n = 100, distr = rbinom(100, 1, 0.5), ymax = 2)