Saving a loop of ggplot2 boxplots as a list [dupli

Question

This question already has an answer here:

• Trying to make a list of ggplot objects in a for loop; all items in list are written as last iteration from loop [duplicate] 2 answers

I want to create a series of boxplots using ggplot2. A toy data-frame, OR.df, consists of 14 variables I wish to create boxplots for over factor prop.bdlf, grouped by method_type. I want to save them in a list so that I can edify them later, or print them out.

set.seed(24542)
OR.df <- data.frame(
  matrix(rnorm(1400, 0, 1), ncol = 14, dimnames = list(NULL, paste0("Estimate.", 1:14))),
  method_type = paste0("Method", 1:5), prop.bdlf = as.factor(c(0, 3, 5, 10))
)

#Start plotting ...
my.plot <- vector(mode = "list", length = 14)
for(j in 1:14){
  title <-  gsub("Estimate.", "", colnames(OR.df)[j])
  cat("> Plotting...", paste0( "w = ", colnames(OR.df)[j]),
      "with title", title, "\n")
  p <- ggplot(  OR.df, aes(y= OR.df[ , j], x = method_type , fill = prop.bdlf) ) +
    ggtitle(paste(title, word[l], sep ="-")) +
    geom_boxplot() +
    labs( x = "Method Type", y  = "weight" ) +
    theme(legend.position = "right", legend.text=element_text(size = 11) ) +
    guides(fill=guide_legend(title="BDL Prop"))
  my.plot[[j]] <- p
} #end weight loop

multiplot( plotlist = my.plot[10:11]) 

The plots are the same! Why? The data is clearly different.

> summary(OR.df$Estimate.10)
    Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
-1.83717 -0.66358 -0.12748 -0.09981  0.36622  2.16782 
> summary(OR.df$Estimate.11)
    Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
-2.86220 -0.67540  0.07784  0.08359  0.73410  2.82225 

So, why is the same data being plotted in the loop? If I go inside the loop and set j = 11, the plot "p" matches exactly with the separate plot.

Thank you.

Answer 1

The reason is that in the y parameter passed is the column value instead of the column name, while the x and fill are correctly passed i.e.

OR.df[,1], OR.df[,2], ..., OR.df[, 14] would be replaced with the column names as a symbol and evaluate (!!) it

my.plot <- vector(mode = "list", length = 14)
for(j in 1:14){
  title <-  gsub("Estimate.", "", colnames(OR.df)[j])
  cat("> Plotting...", paste0( "w = ", colnames(OR.df)[j]),
      "with title", title, "\n")
  p <- ggplot(OR.df, aes(y= !! rlang::sym(names(OR.df)[j]),   # change here
                x = method_type , fill = prop.bdlf) ) +
    ggtitle(paste(title, j, sep ="-")) +
    geom_boxplot() +
    labs( x = "Method Type", y  = "weight" ) +
    theme(legend.position = "right", legend.text=element_text(size = 11) ) +
    guides(fill=guide_legend(title="BDL Prop"))
  my.plot[[j]] <- p
} 

-testing (multiplot from here)

multiplot(plotlist = my.plot[10:11]) 

-output