Search code examples
matplotliblegenduncertainty

Line plus shaded region for error band in matplotlib's legend

I would like to have a legend similar to graph with uncertainty band

but with the dashed line and the yellow region merged like this:

enter image description here

Solution

  • To do what you want, you would need to call legend combining in a single item the two lines/patches you want.

    To see how to do this in practice, here is a simple working example:

    # Import libraries
import numpy as np
import matplotlib.pyplot as plt

# Create some fake data
xvalue = np.linspace(1,100,100)
pop_mean = xvalue
walker_pos = pop_mean + 10*np.random.randn(100)

# Do the plot
fig, ax = plt.subplots()

# Save the output of 'plot', as we need it later
lwalker, = ax.plot(xvalue, walker_pos, 'b-')

# Save output of 'fill_between' (note there's no comma here)
lsigma = ax.fill_between(xvalue, pop_mean+10, pop_mean-10, color='yellow', alpha=0.5)

# Save the output of 'plot', as we need it later
lmean, = ax.plot(xvalue, pop_mean, 'k--')

# Create the legend, combining the yellow rectangle for the 
# uncertainty and the 'mean line'  as a single item
ax.legend([lwalker, (lsigma, lmean)], ["Walker position", "Mean + 1sigma range"], loc=2)

fig.savefig("legend_example.png")
plt.show()

    This code results in this figure: legend_example.png

    You can take a look at the Legend guide to follow what's going on, and to adapt the legend to what you want exactly.