Adding and Labeling a Reference Line
With a lot of graph types, you may want to add a reference line so that the data can be compared to it. For example, perhaps you have a graph that shows growth over time, and want to have a reference line for “no growth” so you can easily see how far things have come. Or perhaps an event happens at a particular time and you want to mark when the event is. Or maybe you just want it to be easy to compare different categories to a mean.
Keep in Mind
- Make sure that it’s clear what your reference line is. A reader might not guess that it represents a mean, or a particular event, or something else. In some cases, the line extending to a particular value on the x- or y-axis does the job. Other times you might want a direct label.
Also Consider
- This page will show how to place a line but not how to style it. You may want your line to be dashed, or bold, or a different color. In most cases the stylistic controls for your reference line will be the exact same as those for a regular line graph. See styling line graphs
Implementations
These implementations will add a line indicating the mean area to the bar graphs found in line graph. They will also show how to place a vertical line, this time at a particular value between the bars, showing how reference lines can be placed on discrete axes as well.
Python
# Load packages
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
df = pd.read_csv("https://vincentarelbundock.github.io/Rdatasets/csv/DAAG/Manitoba.lakes.csv", index_col=0)
# Calculate the value where we want the reference line to be, the mean
# Note we could pick any other value here that we wanted
mean_area = np.mean(df['area'])
# This uses pandas' built-in bar plot function, but this uses
# matplotlib under the hood; any other matplotlib bar graph works the same
plt.style.use('seaborn')
ax = df.plot.bar(y='area', legend=False, ylabel='Area', rot=15)
ax.set_title('Area of lakes in Manitoba', loc='left')
# Place the line
plt.axhline(mean_area)
# Look at our result and figure out the appropriate x/y location
# We can figure out visually that it should be a bit above 5000, where
# the line is. But how about x? We can set x by trial and error, or
# note that there are 9 bars and get the x-coordinate of the last one
# using ax.patches.get_x(), and adjust from there
plt.annotate('Mean Area', xy = (ax.patches[8].get_x() - .5, 5500))
# Similarly if we want to position a vertical line, we can use plt.axvline.
# How to position it on a discrete non-numeric x axis?
# place it after the third bar using get_x to find the third bar
# and get_width to move over to the right side of the bar, then a bit more to adjust
plt.axvline(ax.patches[2].get_x() + ax.patches[2].get_width() + .25)
R
In this example, we will place the line’s label using the ggplot2 function annotate(), which will require us to figure out the annotation’s coordinates ourselves. However, if you prefer, you can use the point-and-click annotation tool ggannotate.
library(tidyverse)
df = read_csv("https://vincentarelbundock.github.io/Rdatasets/csv/DAAG/Manitoba.lakes.csv") %>%
rename(Location = ...1)
# Get the reference value we want to mark
mean_area = df %>% pull(area) %>% mean()
# Make the bar plot, and order the bars by height, why not
p <- ggplot(df, aes(x = reorder(Location, -area), y = area)) +
geom_col() +
labs(x = 'Location', y = 'Area') +
# add a horizontal line using geom_hline, specifying its y intercept
geom_hline(yintercept = mean_area)
# Look at our result so far and figure out the appropriate x/y location
# for our annotation.
p
# We can figure out visually that it should be a bit above 5000, where
# the line is. But how about x? We want it over the 9th bar, so we start with 9
# and can adjust from there by changing x or the horizontal justification (hjust)
p <- p + annotate(geom = 'text', label = 'Mean Area', x = 9, y = 5500) +
# Now we can add a vertical reference line with geom_vline
# If we want it between bars 3 and 4, that puts the line at x = 3.5
geom_vline(xintercept = 3.5)
p
Stata
import delimited "https://vincentarelbundock.github.io/Rdatasets/csv/DAAG/Manitoba.lakes.csv", clear
rename v1 Location
* Get the reference value we want to mark
summ area
local mean_area = r(mean)
* Make the bar plot, and order the bars by height, why not
* And add a horizontal reference line with yline
graph bar area, over(Location, sort(1)) yti(Area) ///
yline(`mean_area')
* From here, you need to use the Graph Editor
* Right-click the graph, do "Start Graph Editor", and add the annotation
* And you can also add a line object to place a vertical reference line
* If we want to do things ourselves, we need to switch to twoway
* (which we should do here anyway for a method that works with non-bar graphs)
* Unfortunately, twoway bar doesn't like categorical x axes so we need to do some work there
* Put the bars in order and create a labeled numeric variable
* that's in the order we want the bars
gsort -area
g Location_n = _n
* Make sure labutil is installed with ssc install labutil
labmask Location_n, values(Location)
* Get the reference value we want to mark
summ area
local mean_area = r(mean)
* NOW A DILEMMA:
* we can easily add vertical and horizontal lines in twoway with yline and xline (or tline for time series graphs)
* BUT these go BEHIND the graph, not in the foreground
* that's okay for our vertical line between bars 3 and 4 (at 3.5) so let's do that
* but for our horizontal line we'll need to draw it ourselves with function
* which is annoying since we'll have to specify its range by hand
* Start with our basic graph that mimics the graph bar we started with
twoway (bar area Location_n, xti("Location") yti("Area") xlab(1/9, valuelabel) legend(off) ///
xline(3.5)) /// Our vertical line goes between bars 3 and 4, i.e. at 3.5
(function y = `mean_area', range(.5 9.5) /// Now our horizontal line at the mean
)
* Now we can look at our result and see where we think the annotation
* should go. For our x-axis value we have 9 bars so should aim somewhere around 9
twoway (bar area Location_n, xti("Location") yti("Area") xlab(1/9, valuelabel) legend(off) ///
xline(3.5) text(5500 9 "Mean Area")) /// Our vertical line goes between bars 3 and 4, i.e. at 3.5
(function y = `mean_area', range(.5 9.5))
Tableau
To add a reference line in Tableau is easy when it’s a continuous variable. Start by making your graph.
Then, go to the analytics pane and drag a reference line to your graph.
Select the calculation (or value) you want
And the result will be automatically labeled.
For a reference line on a discrete variable, the process is much more involved. See this guide.