Using for loop to create multiple ggplots

Question

p[[2]]<-ggplot(data = icadata, aes(x = icadata[[2]])) + geom_histogram()    
p[[3]]<-ggplot(data = icadata, aes(x = icadata[[3]])) + geom_histogram()    
p[[4]]<-ggplot(data = icadata, aes(x = icadata[[4]])) + geom_histogram()    
p[[5]]<-ggplot(data = icadata, aes(x = icadata[[5]])) + geom_histogram()    
p[[6]]<-ggplot(data = icadata, aes(x = icadata[[6]])) + geom_histogram()    
p[[7]]<-ggplot(data = icadata, aes(x = icadata[[7]])) + geom_histogram()    
p[[8]]<-ggplot(data = icadata, aes(x = icadata[[8]])) + geom_histogram()    
p[[9]]<-ggplot(data = icadata, aes(x = icadata[[9]])) + geom_histogram()    
p[[10]]<-ggplot(data = icadata, aes(x = icadata[[10]])) + geom_histogram()    
p[[11]]<-ggplot(data = icadata, aes(x = icadata[[11]])) + geom_histogram()    
p[[12]]<-ggplot(data = icadata, aes(x = icadata[[12]])) + geom_histogram()    
p[[13]]<-ggplot(data = icadata, aes(x = icadata[[13]])) + geom_histogram()    
p[[16]]<-ggplot(data = icadata, aes(x = icadata[[16]])) + geom_histogram()    

p <- list()    
for(i in c(2:13,16) ) {      
           p[[i]]<-ggplot(data = icadata, aes(x = icadata[[i]])) + geom_histogram()    
}

What's the difference between the codes above?
Why do I get always the same picture after runnning the for loop code?

Welcome to SO! Zhi li, this makes very little sense without some context, please realize that nobody else knows what you do about `icadata`. Please make this question *reproducible*. This includes sample code (including listing non-base R packages), sample *unambiguous* data (e.g., `dput(head(x))` or `data.frame(x=...,y=...)`), and expected output. Refs: https://stackoverflow.com/questions/5963269, https://stackoverflow.com/help/mcve, and https://stackoverflow.com/tags/r/info. — r2evans, Dec 23 '19 at 04:25
Every time `ggplot` is called, it overwrites the previous plots and draw new plots on a fresh canvas, then displays that last plot drawn. This is why you get only the last plot every time you run the code, because the loop makes a new plot every iteration. What are you trying to achieve? — Nuclear03020704, Dec 23 '19 at 04:30
I want to get all the histgrams of numerical variables in my icadata.It's inconvenientto plot one-by-one,so I try to use this for-loop to plot them and save them in a list.Do you know how to correct it?Thanks. — zhi li, Dec 23 '19 at 04:58
You want to get all the histograms into one plot? That would be unreadable. Using facets instead would be more helpful. — Phil, Dec 23 '19 at 05:17

score 1 · Answer 1 · answered Dec 23 '19 at 07:16

if you look at the plot you get, it is the last plot that appears. ggplot2 only evaluates the function when you call it. And by entering the x value as icadata[[i]], ggplot2 only plots the icadata[[16]].

So if you do need to use the for loop, it is much better to point it directly to the column inside icadata:

icadata = data.frame(sapply(1:16,function(i)rnorm(100,i,1)))

for(i in c(2:13,16) ) {
COLUMN=colnames(icadata)[i]
p[[i]]<-ggplot(data = icadata, aes_(x = as.name(COLUMN))) + geom_histogram()  
# or use 
# p[[i]]<-ggplot(data = icadata, aes_string(x = COLUMN)) + geom_histogram()    
}  

grid.arrange(grobs=p[c(2:13,16)])

score 0 · Answer 2 · answered Dec 23 '19 at 06:04

library(reshape2)
d <- melt(icadata[,-c(1,14,15)])
ggplot(d,aes(x = value)) + 
  facet_wrap(~variable,scales = "free_x") + 
  geom_histogram()

I tried facets instead.It worked.But I am still confused why I can't get a plot list using that for-loop code unless I plot them one-by-one like the very beginning of my code.

Uwe · Answer 3 · 2019-12-23T08:44:22.073

I have been able to reproduce the observed behaviour using the mtcars dataset. To me, it seems that the way how aes(x = icadata[[2]]) is specified might be the issue.

The code below does work for me:

icadata <- as.data.frame(mtcars) # create reproducible data
library(ggplot2)

p <- list()    
for(i in c(2:3, 6)) {
  p[[i]] <- ggplot(data = icadata, aes(x = !!sym(names(icadata)[i]))) + geom_histogram()    
}

Output is a list as requested by the OP

class(p)

[1] "list"

with the list elements either empty or a ggplot object

lapply(p, class)

[[1]]
[1] "NULL"

[[2]]
[1] "gg"     "ggplot"

[[3]]
[1] "gg"     "ggplot"

[[4]]
[1] "NULL"

[[5]]
[1] "NULL"

[[6]]
[1] "gg"     "ggplot"

I am able to plot the whole list by

lapply(p, print)

or a single graph with, e.g.,

p[[3]]

Please, note that the x-axis is labeled by the column name, e.g., disp instead of icadata[[3]].

If desired, the for loop can be replaced by a call to lapply() which returns a list object without empty elements. In addition, the list elements can be named appropriately.

cols <- names(icadata)[c(2:3, 6)]
p2 <- lapply(
  cols, 
  function(col) ggplot(data = icadata, aes(x = !!sym(col))) + geom_histogram()
  )
names(p2) <- paste0(cols, "_histogram")

lapply(p2, class)

$cyl_histogram
[1] "gg"     "ggplot"

$disp_histogram
[1] "gg"     "ggplot"

$wt_histogram
[1] "gg"     "ggplot"

Sorry to bother you again.Can you explain !!sym(col) to me?Is it same with as.name()? — zhi li, Dec 23 '19 at 09:49

Using for loop to create multiple ggplots

3 Answers3