Add error bars to show standard deviation on a plot in R

Question

For each X-value I calculated the average Y-value and the standard deviation (sd) of each Y-value

x  = 1:5
y  = c(1.1, 1.5, 2.9, 3.8, 5.2)
sd = c(0.1, 0.3, 0.2, 0.2, 0.4)

plot (x, y)

How can I use the standard deviation to add error bars to each datapoint of my plot?

also see `plotrix::plotCI` – Ben Bolker Feb 25 '13 at 15:13 — Ben Bolker, Feb 25 '13 at 15:13

score 32 · Answer 1 · answered Feb 25 '13 at 09:06

32

A solution with ggplot2 :

qplot(x,y)+geom_errorbar(aes(x=x, ymin=y-sd, ymax=y+sd), width=0.25)

enter image description here

answered Feb 25 '13 at 09:06

juba

47,631
14
113
118

setting geom_errorbar(width = 0) removes caps from both error bar ends. Any idea of how this could be done only on one side of the error bar? Thanks – AJMA May 27 '17 at 05:33

score 22 · Answer 2 · answered Feb 25 '13 at 10:39

22

In addition to @csgillespie's answer, segments is also vectorised to help with this sort of thing:

plot (x, y, ylim=c(0,6))
segments(x,y-sd,x,y+sd)
epsilon <- 0.02
segments(x-epsilon,y-sd,x+epsilon,y-sd)
segments(x-epsilon,y+sd,x+epsilon,y+sd)

enter image description here

answered Feb 25 '13 at 10:39

thelatemail

91,185
12
128
188

score 21 · Answer 3 · edited Feb 10 '16 at 00:01

21

You can use arrows:

arrows(x,y-sd,x,y+sd, code=3, length=0.02, angle = 90)

edited Feb 10 '16 at 00:01

HubertL

19,246
3
32
51

answered Mar 04 '15 at 07:34

SmartCH

231
2
6

2

This answer is better than the answers involving `segments`; simpler, and works even when the x axis is on a log scale (which would cause the solutions using `segments` to draw different lengths to the left versus the right). – bhaller Feb 25 '16 at 13:25
@SmartCH, how do I compute the interquartiles by row of a matrix and then plot the interquartiles and mean as above, please? – Joke O. Mar 23 '16 at 16:23

csgillespie · Answer 4 · 2013-02-25T12:35:33.523

You can use segments to add the bars in base graphics. Here epsilon controls the line across the top and bottom of the line.

plot (x, y, ylim=c(0, 6))
epsilon = 0.02
for(i in 1:5) {
    up = y[i] + sd[i]
    low = y[i] - sd[i]
    segments(x[i],low , x[i], up)
    segments(x[i]-epsilon, up , x[i]+epsilon, up)
    segments(x[i]-epsilon, low , x[i]+epsilon, low)
}

As @thelatemail points out, I should really have used vectorised function calls:

segments(x, y-sd,x, y+sd)
epsilon = 0.02
segments(x-epsilon,y-sd,x+epsilon,y-sd)
segments(x-epsilon,y+sd,x+epsilon,y+sd)

enter image description here

score 18 · Accepted Answer · answered Sep 06 '13 at 14:21

A Problem with csgillespie solution appears, when You have an logarithmic X axis. The you will have a different length of the small bars on the right an the left side (the epsilon follows the x-values).

You should better use the errbar function from the Hmisc package:

d = data.frame(
  x  = c(1:5)
  , y  = c(1.1, 1.5, 2.9, 3.8, 5.2)
  , sd = c(0.2, 0.3, 0.2, 0.0, 0.4)
)

##install.packages("Hmisc", dependencies=T)
library("Hmisc")

# add error bars (without adjusting yrange)
plot(d$x, d$y, type="n")
with (
  data = d
  , expr = errbar(x, y, y+sd, y-sd, add=T, pch=1, cap=.1)
)

# new plot (adjusts Yrange automatically)
with (
  data = d
  , expr = errbar(x, y, y+sd, y-sd, add=F, pch=1, cap=.015, log="x")
)

Add error bars to show standard deviation on a plot in R

5 Answers5

Linked

Related