Adding percentage labels on pie chart in R

Question

My data frame looks like

df
   Group   value
1 Positive    52
2 Negative   239
3 Neutral     9

I would like to make a pie chart of the data frame using ggplot.

pie <- ggplot(df, aes(x="", y=value, fill=Group)) +
       geom_bar(width = 1, stat = "identity") +
       coord_polar("y", start=0) 

This is my pie chart.

But when I try to add percentage labels on the chart

pie <- ggplot(df, aes(x="", y=value, fill=Group)) +
       geom_bar(width = 1, stat = "identity") +
       coord_polar("y", start=0) +
       geom_text(aes(y = value/2 + c(0, cumsum(value)[-length(value)]),
                 label = percent(value/300 )), size=5)

This is my result.

I have already seen many same question as mine,i.e R + ggplot2 => add labels on facet pie chart and the solutions are not helping.

Answer 1

I agree with @hrbrmstr a waffle chart would be better. But to answer the original question... your problem comes from the order in which the wedges are drawn, which will default to alphabetical. As you calculate where to place the labels based on the ordering in your data frame, this works out wrong.

As a general principle of readability, do all the fancy calculations of labels and positions they go before the actual code drawing the graphic.

library(dplyr)
library(ggplot2)
library(ggmap) # for theme_nothing
df <- data.frame(value = c(52, 239, 9),
                 Group = c("Positive", "Negative", "Neutral")) %>%
   # factor levels need to be the opposite order of the cumulative sum of the values
   mutate(Group = factor(Group, levels = c("Neutral", "Negative", "Positive")),
          cumulative = cumsum(value),
          midpoint = cumulative - value / 2,
          label = paste0(Group, " ", round(value / sum(value) * 100, 1), "%"))

ggplot(df, aes(x = 1, weight = value, fill = Group)) +
   geom_bar(width = 1, position = "stack") +
   coord_polar(theta = "y") +
   geom_text(aes(x = 1.3, y = midpoint, label = label)) +
   theme_nothing()               

Answer 2

How about:

vals <- c(239, 52, 9)
val_names <- sprintf("%s (%s)", c("Negative", "Positive", "Neutral"), scales::percent(round(vals/sum(vals), 2)))
names(vals) <- val_names

waffle::waffle(vals) +
  ggthemes::scale_fill_tableau(name=NULL)

instead?

It's "fresher" than a pie chart and you aren't really gaining anything with the level of precision you have/want on those pie labels now.

Answer 3

For example, I create a dataframe e3 with 400 vehicles:

e3 <- data.frame(400)
e3 <- rep( c("car", "truck", "other", "bike", "suv"), c(60, 120, 20, 50, 150))

Since pie charts are especially useful for proportions, let's have a look on the proportions of our vehicles, than we will report on the graph in this case:

paste(prop.table(table(e3))*100, "%", sep = "")
[1] "15%"   "5%"    "30%"   "12.5%" "37.5%"

Then you can draw your pie chart,

pie(table(e3), labels = paste(round(prop.table(table(e3))*100), "%", sep = ""), 
col = heat.colors(5), main = "Vehicles proportions - n: 400")

Answer 4

Here is an idea matching the order of groups in the pie chart and the order of labels. I sorted the data in descending order by value. I also calculated the percentage in advance. When I drew the ggplot figure, I specified the order of Group in the order in mydf (i.e., Negative, Positive, and Neutral) using fct_inorder(). When geom_label_repel() added labels to the pie, the order of label was identical to that of the pie.

library(dplyr)
library(ggplot2)
library(ggrepel)
library(forcats)
library(scales)

mydf %>%
arrange(desc(value)) %>%
mutate(prop = percent(value / sum(value))) -> mydf 

pie <- ggplot(mydf, aes(x = "", y = value, fill = fct_inorder(Group))) +
       geom_bar(width = 1, stat = "identity") +
       coord_polar("y", start = 0) +
       geom_label_repel(aes(label = prop), size=5, show.legend = F, nudge_x = 1) +
       guides(fill = guide_legend(title = "Group"))

DATA

mydf <- structure(list(Group = structure(c(3L, 1L, 2L), .Label = c("Negative", 
"Neutral", "Positive"), class = "factor"), value = c(52L, 239L, 
9L)), .Names = c("Group", "value"), class = "data.frame", row.names = c("1", 
"2", "3"))