R ggplot: Apply label only to last N data points in plot

Question

I have created a line chart (plot) in R with labels on each data point. Due to the large number of data points, the plot becomes very fully with labels. I would like to apply the labels only for the last N (say 4) data points. I have tried subset and tail in the geom_label_repel function but was not able to figure them our or got an error message. My data set consist of 99 values, spread over 3 groups (KPI).

I have the following code in R:

library(ggplot)
library(ggrepel)

data.trend <- read.csv(file=....)

plot.line <- ggplot(data=data.trend, aes(x = Version, y = Value, group = KPI, color = KPI)) +

  geom_line(aes(group = KPI), size = 1) +
  geom_point(size = 2.5) +


  # Labels defined here
  geom_label_repel(
    aes(Version, Value, fill = factor(KPI), label = sprintf('%0.1f%%', Value)),
    box.padding = unit(0.35, "lines"),
    point.padding = unit(0.4, "lines"),
    segment.color = 'grey50',
    show.legend = FALSE
  )

);

I all fairness, I am quite new to R. Maybe I miss something basic.

Thanks in advance.

Please provide a reproducible example or at least show the resulting figure. — Roman Luštrik, Jan 07 '17 at 14:46

dww · Accepted Answer · 2017-01-07T15:55:49.823

The simplest approach is to set the data = parameter in geom_label_repel to only include the points you want labeled.

Here's a reproducible example:

set.seed(1235)
data.trend <- data.frame(Version = rnorm(25), Value = rnorm(25), 
                         group = sample(1:2,25,T), 
                         KPI = sample(1:2,25,T))

ggplot(data=data.trend, aes(x = Version, y = Value, group = KPI, color = KPI)) +
  geom_line(aes(group = KPI), size = 1) +
  geom_point(size = 2.5) +
  geom_label_repel(aes(Version, Value, fill = factor(KPI), label = sprintf('%0.1f%%', Value)),
    data = tail(data.trend, 4),                 
    box.padding = unit(0.35, "lines"),
    point.padding = unit(0.4, "lines"),
    segment.color = 'grey50',
    show.legend = FALSE)

Unfortunately, this messes slightly with the repel algorithm, making the label placement suboptimal with respect to the other points which are not labelled (you can see in the above figure that some points get covered by labels).

So, a better approach is to use color and fill to simply make the unwanted labels invisible (by setting both color and fill to NA for labels you want to hide):

ggplot(data=data.trend, aes(x = Version, y = Value, group = KPI, color = KPI)) +
  geom_line(aes(group = KPI), size = 1) +
  geom_point(size = 2.5) +
  geom_label_repel(aes(Version, Value, fill = factor(KPI), label = sprintf('%0.1f%%', Value)),
                   box.padding = unit(0.35, "lines"),
                   point.padding = unit(0.4, "lines"),
                   show.legend = FALSE,
                   color = c(rep(NA,21), rep('grey50',4)),
                   fill = c(rep(NA,21), rep('lightblue',4)))

Thanks, you put me in the right direction. However, I see that only the last data points of 1 line are shown. To overcome this I have added a list with the Versions that I want to plot and use a subset in the data field to only show them. This does mix up the repel function, but that is no problem in my case. Many thanks! `labels <- tail(data.trend$Version, 3) ..... geom_label_repel( data = subset(data.trend, data.trend$Version %in% labels), aes(Version, Value, fill = factor(KPI), label = sprintf('%0.1f%%', Value)), ... )` — user32556, Jan 07 '17 at 17:25

score 2 · Answer 2 · answered Aug 27 '21 at 15:33

If you want to show just the last label, using group_by and filter may work:

data = data.trend %>% group_by(KPI) %>% filter(Version == max(Version))

Full example:

suppressPackageStartupMessages(library(dplyr))
library(ggplot2)
library(ggrepel)

set.seed(1235)
data.trend <- data.frame(Version = rnorm(25), Value = rnorm(25), 
                         group = sample(1:2,25,T), 
                         KPI = sample(1:2,25,T))

ggplot(data = data.trend, aes(x = Version, y = Value, group = KPI, color = KPI)) +
  geom_line(aes(group = KPI), size = 1) +
  geom_point(size = 2.5) +

  # Labels defined here
  geom_label_repel(
    data = data.trend %>% group_by(KPI) %>% filter(Version == max(Version)), 
    aes(Version, Value, fill = factor(KPI), label = sprintf('%0.1f%%', Value)),
    color = "black",
    fill = "white")

Or if you want to show 4 random labels per KPI, data.trend %>% group_by(KPI) %>% sample_n(4):

suppressPackageStartupMessages(library(dplyr))
library(ggplot2)
library(ggrepel)

set.seed(1235)
data.trend <- data.frame(Version = rnorm(25), Value = rnorm(25), 
                         group = sample(1:2,25,T), 
                         KPI = as.factor(sample(1:2,25,T)))

ggplot(data = data.trend, aes(x = Version, y = Value, group = KPI, color = KPI)) +
  geom_line(aes(group = KPI), size = 1) +
  geom_point(size = 2.5) +
  
  # Labels defined here
  geom_label_repel(
    data = data.trend %>% group_by(KPI) %>% sample_n(4), 
    aes(Version, Value, fill = factor(KPI), label = sprintf('%0.1f%%', Value), fill = KPI),
    color = "black", show.legend = FALSE
    )
#> Warning: Duplicated aesthetics after name standardisation: fill

^{Created on 2021-08-27 by the reprex package (v2.0.1)}

R ggplot: Apply label only to last N data points in plot

2 Answers2