Matplotlib error bar, capsize didn't work?

Question

I drew a custom error bar for my seaborn barplot because I need to use within-subject error bar. I used:

plt.errorbar([0, 1], [mean1, mean2], yerr = errBar_value,
             elinewidth = 3, ecolor = "k", capsize = 0.5)

Here is the result

How to increase the capsize? I tried different values of capsize but nothing is changed.

Many thanks

Jiajun

score 2 · Accepted Answer · answered Mar 03 '18 at 11:28

2

plt.errorbar([0, 1], [mean1, mean2], yerr = errBar_value,elinewidth = 3, ecolor = "k", capsize = 0.5,fmt=' ')

The fmt statement gets rid of the line. For me it works to put a larger value for capsize to increase its size. I am using python 3.6.1

answered Mar 03 '18 at 11:28

DZurico

637
4
11

thanks. strangely I tried capsize from 0.01 ~ 1000. But nothing change. I used 2.7. But the capsize on the seaborn.barplot totally worked. – J_yang Mar 03 '18 at 11:30
Just tried it with python 2.7. It works there as well (used capsize=10). My matplotlib is version 2.1.2 – DZurico Mar 03 '18 at 11:38

score 1 · Answer 2 · answered Mar 03 '18 at 11:35

At the end, I did this:

(_, caps, _) = plt.errorbar([0, 1], [ncv['VisualOnly'].mean(), ncme['Mode Explorer'].mean()], yerr = errBar_vo,
             elinewidth = 3, ecolor = "k", fmt = ' ',capsize = 5)
for cap in caps:
    cap.set_color('k')
    cap.set_markeredgewidth(3)

setting the caps manually seems to solve the issue.

Matplotlib error bar, capsize didn't work?

2 Answers2