Matplotlib: plotting transparent histogram with non transparent edge

plt.hist accepts additional keyword arguments that are passed to the constructor for matplotlib.patches.Patch. In particular you can pass an fc= argument which lets you set the patch facecolor using an (R, G, B, A) tuple when you create the histograms. Changing the alpha value of the facecolor does not affect the transparency of the edges:

ax.hist(x, bins=np.arange(0, 1, 0.1), ls='dashed', lw=3, fc=(0, 0, 1, 0.5))ax.hist(y, bins=np.arange(0, 1, 0.1), ls='dotted', lw=3, fc=(1, 0, 0, 0.5))ax.hist(z, bins=np.arange(0, 1, 0.1), lw=3, fc=(0, 0, 0, 0.5))

I think perhaps the dumbest solution is the easiest and best? Just replot the histograms, without inner color and without alpha. That should replot just the histogram lines.

Additionally you might want to eliminate the chance of having the lines/dashes/dots not overlap by removing them completely from the first histogram plot

import matplotlib as mplimport matplotlib.pyplot as pltimport numpy as npx = np.random.random(20)y = np.random.random(20)z = np.random.random(20)fig = plt.figure()ax = fig.add_subplot(111)ax.hist(x, bins=np.arange(0, 1, 0.1), edgecolor='None', alpha = 0.5, color= 'b')ax.hist(y, bins=np.arange(0, 1, 0.1), edgecolor='None', alpha = 0.5, color= 'r')ax.hist(z, bins=np.arange(0, 1, 0.1), edgecolor="None", alpha = 0.5, color= 'k')ax.hist(x, bins=np.arange(0, 1, 0.1), ls='dashed', lw=3, facecolor="None")ax.hist(y, bins=np.arange(0, 1, 0.1), ls='dotted', lw=3, facecolor="None")ax.hist(z, bins=np.arange(0, 1, 0.1), lw=3, facecolor="None")plt.show()

without the second orders the graph should look like your figure without any borders. With the bottom 3 histogram commands borders are added to produce the image bellow. Works on Python3.4 Win7

[cut because apparently I don't have enough rep to post pictures]

I needed a solution which did not require me to set the color explicitly, i.e., I wanted to still use the default color cycle. The following solution builds on @ljetibo's idea to draw the histogram twice using @ali_m's idea to extract the state of the color cycle:

# increment and get the "props" cycle (and extract the color)color = next(ax._get_lines.prop_cycler)["color"]# 1. draw: inner area with alphaax.hist(data, color=color, alpha=0.3)# 2. draw: only edges with full colorax.hist(data, color=color, edgecolor=color, fc="None", lw=1)

Note that specifying color=color in addition to edgecolor seems to be necessary. Otherwise the hist call itself will increment the color cycle again.