[Tutor] Plotting asymmetric error bars for a single point in matplotlib

[Peter Otten]

Colin Ross wrote:

> Hi all,
> 
> Goal: To plot asymmetric x error bars for a single point using errorbar. I
> am interested in displaying the inter quartile range (IQR) for a data set.
> 
> Code:
> 
> import numpy as np
> import matplotlib.pyplot as plt
> 
> y = 1.0
> data = np.random.rand(100)
> 
> median = np.median(data)
> upper_quartile = np.percentile(data, 75)
> lower_quartile = np.percentile(data, 25)
> IQR = upper_quartile - lower_quartile
> 
> plt.errorbar(median, y, xerr=[lower_quartile ,upper_quartile], fmt='k--')
> 
> plt.savefig('IQR.eps')
> plt.show()
> 
> Error:
> 
> Traceback (most recent call last):
>   File "IQR.py", line 15, in <module>
>     plt.errorbar(median, y, xerr=[0.5,0.75], fmt='k--')
>   File "/usr/lib/pymodules/python2.7/matplotlib/pyplot.py", line 2251, in
> errorbar
>     ret = ax.errorbar(x, y, yerr, xerr, fmt, ecolor, elinewidth, capsize,
> barsabove, lolims, uplims, xlolims, xuplims, **kwargs)
>   File "/usr/lib/pymodules/python2.7/matplotlib/axes.py", line 5327, in
> errorbar
>     in cbook.safezip(x,xerr)]
>   File "/usr/lib/pymodules/python2.7/matplotlib/cbook.py", line 1294, in
> safezip
>     raise ValueError(_safezip_msg % (Nx, i+1, len(arg)))
> ValueError: In safezip, len(args[0])=1 but len(args[1])=2
> 
> My understanding is that safezip zips together the x array with the
> specified upper and lower x limits?

If you have not solved this yet: my guess is that 

xerr=[lower_quartile, upper_quartile]

is interpreted as two error values that would correspond to x and y vectors 
of length 2 with symmetrical error bars. Try passing a list of two length-
one lists two avoid the ambiguity:

plt.errorbar(
    median, y, 
    xerr=[[lower_quartile], [upper_quartile]], fmt='k--')