Error adding components to ggplot in for loop

Question

I am trying to use the code below to plot many ROC plots on one plot. x is a list of vectors (each containing x-coordinates) and y is a list vectors (each containing y-coordinates).

plotROC<- function(x, y, label=NA) {
  p <- ggplot() + ggtitle("ROC Plot") + xlab("False Positive Rate") + 
    ylab("True Positive Rate") + xlim(c(0,1)) + ylim(c(0, 1)) + 
    geom_segment(aes(x=0, y=0, xend=1, yend=1), colour="black")
  for (i in 1:length(x)) {
    auc <- AUC(x[[i]], y[[i]])
    lab <- ifelse(is.na(label[i]), "", paste(label[i], "\n", sep=""))
    p <- p + geom_line(
      aes(x[[i]], y[[i]], colour=paste(lab, "AUC: ", 
                             formatC(auc, format="f", 
                                     digits=2), sep="")))
  }
  p + scale_color_brewer(palette=palette) + 
    theme(legend.title=element_blank())
}

For some reason the the function only plots the final ROC plot. If anybody could help I would be very appreciative!

Thanks,

Jack

The function does not print any plot at all. When the function returns, it returns the final plot. That is then automatically printed by the REPL loop. Call `print` explicitly on your plot objects if you want to see them. — , Apr 04 '18 at 15:18
Possible duplicate of [Generate multiple graphics from within an R function](https://stackoverflow.com/questions/2547306/generate-multiple-graphics-from-within-an-r-function) — , Apr 04 '18 at 15:20
At the moment I set `p <- plotROC(x=list(x1, x2), y=list(y1, y2))` and then type `p` into the console. It only shows `(x2, y2)`. I am not trying to print them in the function. — , Apr 04 '18 at 15:24
Maybe the problem is ```colour=paste(lab, "AUC: ", formatC(auc, format="f", digits=2), sep="")```. This is a label not a colour. For conding style, use ```return``` in your function, do not hesitate to back to the line, and add empty lines for readibility. Moreover, ```palette``` seems to be an argument of your function (or you shloud hardcoded that). — Thomas, Apr 04 '18 at 16:01
Possible duplicate of ["for" loop only adds the final ggplot layer](https://stackoverflow.com/questions/26235825/for-loop-only-adds-the-final-ggplot-layer) — Calimo, Apr 04 '18 at 18:31

Thomas · Answer 1 · 2018-04-04T16:03:53.737

You should use data.frame to work with ggplot2, it will be far easier for you.

For example, you can easily convert your data into data.frame :

x = list(runif(4), runif(4))
y = list(runif(4), runif(4))

df = Reduce(rbind, mapply(function(x, y, z) {
  data.frame(x=x, y=y, z=z)
}, x, y, 1:length(x), SIMPLIFY = FALSE))
df$z = as.factor(df$z)

Note, that I add a z variable, that could correspond to whatever you want. I convert it into factor in order to ggplot consider it as factor and not continuous.

Then your plot can be define like this :

ggplot(df, aes(x=x, y=y, color=z)) + 
  geom_line() +
  ggtitle("ROC Plot") +
  xlab("False Positive Rate") + 
  ylab("True Positive Rate") + 
  xlim(c(0,1)) + ylim(c(0, 1)) + 
  geom_segment(aes(x=0, y=0, xend=1, yend=1), colour="black")

With the for loop (this is not a good idea to stay with a for loop), this code works for me :

library(ggplot2)

plotROC<- function(x, y, label=NA) {

  p <- ggplot() + 
    ggtitle("ROC Plot") +
    xlab("False Positive Rate") + 
    ylab("True Positive Rate") + 
    xlim(c(0,1)) + ylim(c(0, 1)) + 
    geom_segment(aes(x=0, y=0, xend=1, yend=1), colour="black")

  for (i in 1:length(x)) {
    p <- p + geom_line(
      aes(x[[i]], y[[i]]))
  }

  return(
    p + 
      scale_color_brewer(palette="YlOrRd") + 
      theme(legend.title=element_blank())
  )

}

x = list(runif(4), runif(4))
y = list(runif(4), runif(4))

plotROC(x, y)

Thank you for your reply. I will use dataframes in the future. Do you why my function won't work as I want it to though? As I wrote in earlier comment, I don't want to print the plots when I run the function. I will store the result of a function and then print that in the console. — , Apr 04 '18 at 15:43
@Jack the reason is ggplot's lazy evaluation: https://stackoverflow.com/q/26235825/333599 — Calimo, Apr 04 '18 at 18:33

Error adding components to ggplot in for loop

1 Answers1