Control the facecolor of histograms

Question

In the following histogram,

import numpy as np
import matplotlib.pyplot as plt
tinter = [0.8253909999999998, 0.8804020000000001, 0.14202099999999973, 0.7727520000000005, 1.1040220000000005, 0.10714699999999944, 0.5750040000000016, 0.7876929999999973, 0.8980040000000002, 1.1478490000000008, 0.44635099999999994, 1.052067000000001, 0.4327469999999991, 0.3272960000000005, 0.26918099999999967, 0.3854459999999982, 0.1827140000000007, 0.8867760000000011, 0.7774879999999982, 0.21082900000000038, 0.6758939999999996, 0.4335760000000022, 0.6791699999999992, 0.7758439999999993, 0.15755200000000258, 0.1414289999999987, 0.36975599999999886, 0.8993549999999999, 0.6577640000000002, 1.043728999999999, 0.19952800000000082, 0.12645200000000045, 0.3454990000000002, 0.9054520000000004, 0.7165229999999987, 0.9425269999999983, 0.7159280000000052, 0.7413279999999958, 0.12669900000000212, 0.2822880000000012, 0.3690029999999993, 0.7246340000000018, 0.4718329999999966, 0.7580859999999987, 0.744059, 0.19344999999999857, 0.12031900000000206, 0.47543600000000197, 0.437542999999998, 0.44232000000000227, 0.5250109999999992, 0.17673200000000122, 0.2440649999999991, 0.31524799999999686, 0.7674680000000009, 0.7837700000000041, 1.1910290000000003, 0.14404899999999543, 0.21560399999999902, 0.19931500000000568, 0.27113699999999596, 0.728234999999998, 0.5061920000000057, 0.6459329999999994, 0.7817450000000008, 0.8265129999999985, 0.17931199999999592, 0.30208600000000274, 0.32583699999999993, 0.41771599999999864]
n, bins, patches = plt.hist(x=tinter, bins='auto', color='#0504aa',
                            alpha=0.7, rwidth=0.9,)
plt.grid(axis='y', alpha=0.35)
plt.xlabel('time [s]')
plt.ylabel('Frequency')
plt.title('Hit interval')

I would like to set two different colors for the bars of the histogram. There is a precedent question (this link), but the solution sets the bars' colors in a range, e.g., bars 0 to 3 of one color.

This is a different problem since I would like to differentiate two zones in the histogram in general. I do not exactly how the particular histogram is and how many bars are. In the example above, the separation of colors should be at 0.6: left of time=0.6s of one color and right of time=0.6 of another. If it is not possible to make a bar of two colors (in the example, a range of time values is covered in the bar including 0.6), the separation of colors should go from the lowest bar in the example (bar of time=0.2) to the bar of time=0.6 included. And, in the example, the last four bars (above time=0.6s) of another color.

For example, in this example the values of n and bins that gives 'plt.hist()` are:

n = array([16., 11., 10.,  3., 16.,  8.,  3.,  3.])
bins=array([0.107147  , 0.24263225, 0.3781175 , 0.51360275, 0.649088  , 0.78457325, 0.9200585 , 1.05554375, 1.191029  ])

I would like bars with n 16, 11, 10 and 3 of one color different of the other group of bars with n 16, 8, 3,3, and make it for a general histogram plot.

Possible duplicate of [How can I specify the color for a partial histogram patch?](https://stackoverflow.com/q/62737627/7758804) — Trenton McKinney, Jun 15 '21 at 21:56

JohanC · Accepted Answer · 2021-06-15T18:35:09.790

You could loop through the bars and test whether it is completely to the right of the separation, completely to the left or crosses it. You change the bar's color correspondingly.

When a bar crosses the separator, the bar gets the color for the left area and its size is narrowed to touch the separator. A copy of the bar can be added, with the right-area color and its x-position moved.

import matplotlib.pyplot as plt
import numpy as np
from copy import copy

tinter = [0.82539, 0.8804, 0.14202, 0.77275, 1.10402, 0.10715, 0.575, 0.78769, 0.898, 1.14785, 0.44635, 1.05207, 0.43275, 0.3273, 0.26918, 0.38545, 0.18271, 0.88678, 0.77749, 0.21083, 0.67589, 0.43358, 0.67917, 0.77584, 0.15755, 0.14143, 0.36976, 0.89935, 0.65776, 1.04373, 0.19953, 0.12645, 0.3455, 0.90545, 0.71652, 0.94253, 0.71593, 0.74133, 0.1267, 0.28229, 0.369, 0.72463, 0.47183, 0.75809, 0.74406, 0.19345, 0.12032, 0.47544, 0.43754, 0.44232, 0.52501, 0.17673, 0.24406, 0.31525, 0.76747, 0.78377, 1.19103, 0.14405, 0.2156, 0.19932, 0.27114, 0.72823, 0.50619, 0.64593, 0.78175, 0.82651, 0.17931, 0.30209, 0.32584, 0.41772]
n, bins, patches = plt.hist(x=tinter, bins='auto', color='#0504aa', alpha=0.7, rwidth=0.9)
plt.grid(axis='y', alpha=0.35)
plt.xlabel('time [s]')
plt.ylabel('Frequency')
plt.title('Hit interval')
separator = 0.6
left_color = 'turquoise'
right_color = 'tomato'
for bar in patches:
    x, y = bar.get_xy()
    w = bar.get_width()
    if x > separator:
        bar.set_color(right_color)
    else:
        bar.set_color(left_color)
        if x + w > separator:
            bar.set_width(separator-x)
            new_bar = copy(bar)
            new_bar.set_x(separator)
            new_bar.set_width(x+w-separator)
            new_bar.set_color(right_color)
            plt.gca().add_patch(new_bar)
plt.show()

If you want the complete bar that contains the separator value to be one color, you can leave out the block below if x + w > separator:.

n, bins, patches = plt.hist(x=tinter, bins='auto', color='#0504aa', alpha=0.7, rwidth=0.9)
plt.grid(axis='y', alpha=0.35)
plt.xlabel('time [s]')
plt.ylabel('Frequency')
plt.title('Hit interval')
separator = 0.6
left_color = 'turquoise'
right_color = 'tomato'
for bar in patches:
    x, y = bar.get_xy()
    w = bar.get_width()
    if x > separator:
        bar.set_color(right_color)
    else:
        bar.set_color(left_color)
plt.show()

It is the solution. Thank you. To make a more general solution, Finding the separation value or plot the full bar with the time=0.6 just with the color turquoise. It is because it seems difficult to find a separation value unless setting it in the script. — user1993416, Jun 15 '21 at 18:20
It is possible to plot the bar with the time 0.6 with just the turquoise color?. It looks difficult to compute the separation value. — user1993416, Jun 15 '21 at 18:25
it would be to set all the bar with the time 0.6 to the color turqouise. — user1993416, Jun 15 '21 at 18:30
It is possible to find observing the values of `n` and ` bins` a way to compute an approximate separation value? Is because I was looking for a general solution to plot different histograms separating two zones. — user1993416, Jun 15 '21 at 18:37
Maybe `separation = np.median(tinter)`? Or `np.mean(tinter)`? — JohanC, Jun 15 '21 at 19:00

score 1 · Answer 2 · answered Oct 08 '22 at 19:18

Alternatively, without using transparent colors (alpha), one can do the following:

left_color = 'turquoise'
right_color = 'tomato'
separator = 0.6

tinter = [0.82539, 0.8804, 0.14202, 0.77275, 1.10402, 0.10715, 0.575, 0.78769, 0.898, 1.14785, 0.44635, 1.05207, 0.43275, 0.3273, 0.26918, 0.38545, 0.18271, 0.88678, 0.77749, 0.21083, 0.67589, 0.43358, 0.67917, 0.77584, 0.15755, 0.14143, 0.36976, 0.89935, 0.65776, 1.04373, 0.19953, 0.12645, 0.3455, 0.90545, 0.71652, 0.94253, 0.71593, 0.74133, 0.1267, 0.28229, 0.369, 0.72463, 0.47183, 0.75809, 0.74406, 0.19345, 0.12032, 0.47544, 0.43754, 0.44232, 0.52501, 0.17673, 0.24406, 0.31525, 0.76747, 0.78377, 1.19103, 0.14405, 0.2156, 0.19932, 0.27114, 0.72823, 0.50619, 0.64593, 0.78175, 0.82651, 0.17931, 0.30209, 0.32584, 0.41772]

# Set the color of histogram bars with the left_color from the beginning
n, bins, patches = plt.hist(x=tinter, bins='auto', color=left_color, rwidth=0.9)
plt.grid(axis='y', alpha=0.35)
plt.xlabel('time [s]')
plt.ylabel('Frequency')
plt.title('Hit interval')

for i, bar in enumerate(patches):
    x, y = bar.get_xy()
    w = bar.get_width()
    
    if x > separator:
        bar.set_color(right_color)

    # Get the dimension for the partial bar 
    else:
        width = x + w - separator
        height = n[i]

# Create the partial bar        
plt.bar(x=separator, height=height, width=width, color=right_color, align='edge')

plt.show()

Control the facecolor of histograms

2 Answers2