Using aes_string to change color of geom_point from dynamic column

Question

Here's my initial dataframe.

data.df

x y z label
2 3 4   1
1 2 3   2
2 4 3   3

To make ggplot, this works when there is only 1 column (label) :

    g <- ggplot(data.df) + 
       geom_point(data = data.df, aes(x= x, y= y, 
       color = ifelse(( label == 2), "a", "b")+
       scale_colour_manual(values= c("a" = "blue", "b" = "green"))

    return g

On clicking a button called "merge", new column gets added dynamically:

x y z label label2
2 3 4   1     1
1 2 3   2     2
2 4 3   3     2

Now in ggplot I need to access LAST column instead of label column (it could be label2, label3...) and update ggplot.

I tried two ways.

  g <- ggplot(data.df) + 
       geom_point(data = data.df, aes(x= x, y= y, 
       color = ifelse(( data.df[, ncol(data.df)] == 2, "a", "b")+
       scale_colour_manual(values= c("a" = "blue", "b" = "green"))


  return g

As shown while using data.df[, ncol(data.df)] , I'm getting the error:

Error: Aesthetics must be either length 1 or the same as the data (40): x, y, colour

I have a feeling aes_string can be used instead of aes:

 label <- paste("label", counter , sep="")

 g <- ggplot(data.df) + 
       geom_point(data = data.df, aes_string(x= "x", y= "y", 
       color = ifelse((label == 2), a, b))) +
       scale_colour_manual(values= c("a" = "blue", "b" = "green"))

I'm getting this error:

Error in ifelse((label == 2), a, b))),  : object a not found

How is this related to python? Also "none of these seem seem to work" is not a proper description of a programming issue. — Stop harming Monica, Dec 21 '16 at 13:58
You could do this by preprocessing the data to create your dynamic label then use `ggplot`. — Jake Kaupp, Dec 21 '16 at 14:05
Including reproducible data would make it substantially easier to help you http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example — Mark Peterson, Dec 21 '16 at 15:12
@JakeKaupp Thanks! Please check edited question and offer suggestion. — Panzer M, Dec 22 '16 at 08:05

score 0 · Accepted Answer · answered Dec 22 '16 at 11:48

My opinion is to do your standard evaluation that allows for the dynamic functionality before you go into ggplot2 functions.

The below takes advantage of Standard evaluation versions of dplyr functions. It creates a column in the data frame dynamically called formatCol and bases the colour scale on it.

data.df <- data.frame(x = c(2, 1, 2),
                      y = c(3, 2, 4),
                      z = c(4, 3, 3),
                      label = c(1, 2, 3),
                      label2 = c(1, 2, 2))

library(ggplot2)
library(dplyr)
library(lazyeval)

formatCol <- names(data.df)[ncol(data.df)]
formula <- interp(~ifelse((label == 2), "a", "b"), label = as.name(formatCol))

plot.df <- data.df %>% mutate_(formatCol = formula)

  g <- ggplot(plot.df, aes(x= x, y= y)) + 
      geom_point( aes(color = formatCol))+
        scale_colour_manual(values= c("a" = "blue", "b" = "green"))

g

I found my error was with scale_colour_manual when testing out this code. This helped me make my code cleaner. Thank you! — Panzer M, Dec 22 '16 at 12:57

Using aes_string to change color of geom_point from dynamic column

1 Answers1