Another way to pass variable as factor to fill argument in ggplot2 function

Question

I have a function that I use to build bar charts in R-markdown. I wonder if there is a better way to do the aes-part of my ggplot2 function. At the moment I use "eval(parse(text=" but that is kinda hard to read/understand.

I would like something that is more direct/readable. I tried aes_string but I think the "fill = factor()" part didn't work since while I did get a chart, I got an erroneous one (I got one big, fat bar). Were did I go wrong? Or is this the best way I can do my chart.

Reproducible example:

library("data.table")
library("ggplot2")

dt <- data.table(year.var = c(2014, 2014, 2014, 2015, 2015, 2015),
                 grp.var  = c("Grp1", "Grp2", "Grp3", "Grp1", "Grp2", "Grp3"),
                 val.var  = c(100, 200, 230, 105, 260, 23))

test <- function (dts,
                  x.var,
                  y.var,
                  fill.var,
                  order.var)  {

setorderv(dts, order.var)

ggplot(dts, 
       aes(x = eval(parse(text = x.var)), 
           y = eval(parse(text = y.var)), 
           fill = eval(parse(text = fill.var))
           )
       ) +
       geom_bar(stat="identity", position="dodge")+
       scale_fill_manual(values =  c("#9badcd", "#5a6c9c"),
                         name = "Year")+
       scale_y_continuous(labels = function(val.var) { 
                                     format(val.var, big.mark = " ", 
                                     scientific = FALSE)
                                   }
                          )+
       xlab("x.label") + 
       ylab("y.label")+
       ggtitle("str.title")
}

test(dt, "grp.var", "val.var", "factor(year.var)", c("year.var"))

I would use aes_string, first manipulate the dts so it has a column called fill.var which is a factor. (would give code but don't know how to do this with a data.table) — Richard Telford
– Richard Telford, Commented Aug 12, 2016 at 12:14
@RichardTelford Thanks!! I think I got to caught up in my old tracks, that did the trick. If you want to post it as an answer you do it like this: dt <- dt[, f.var := as.factor(year.var)] — ErrantBard
– ErrantBard, Commented Aug 12, 2016 at 12:22

Richard Telford · Accepted Answer · 2016-08-12 14:33:18Z

2

I think the solution is to try to keep the mapping in aes simple, and to do the data manipulation directly in the data.table. Here I have made a new column in the data.table which is a factor of the column wanted as the fill. Then aes_string can be used

test <- function (dts,
                  x.var,
                  y.var,
                  fill.var,
                  order.var)  {

  dts <- dts[, f.var := as.factor(dts[, get(fill.var)])] 
   #NB this will modify your original table
  setorderv(dts, order.var)

  ggplot(dts, 
         aes_string(x = x.var, y = y.var, fill = "f.var")) +
    geom_bar(stat="identity", position="dodge")+
    scale_fill_manual(values =  c("#9badcd", "#5a6c9c"), name = "Year")+
    scale_y_continuous(labels = function(val.var) { 
      format(val.var, big.mark = " ", scientific = FALSE)
    })+
    labs(x = "x.label", y = "y.label", title = "str.title")
}

test(dt, "grp.var", "val.var", "year.var", "year.var")

answered Aug 12, 2016 at 14:33

Richard Telford

9,9836 gold badges43 silver badges53 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

Richard Telford Over a year ago

My handling of the data.table might not be elegant. Not a package I use.

ErrantBard Over a year ago

Thanks, and a good point about data manipulation outside the aes. I think your handling is fine, you could write that statement a bit shorter though, dts <- dts[, f.var := as.factor(get(fill.var))]

Collectives™ on Stack Overflow

Another way to pass variable as factor to fill argument in ggplot2 function

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related