Pandas boxplot: set color and properties for box, median, mean

Question

I have a DataFrame with a MultiIndex:

# -*- coding: utf-8 -*-
import numpy as np
import pandas as pd

# dataframe with dates
dates = pd.DataFrame()
dates['2016'] = pd.date_range(start='2016', periods=4, freq='60Min')
dates['2017'] = pd.date_range(start='2017', periods=4, freq='60Min')
dates['2018'] = pd.date_range(start='2018', periods=4, freq='60Min')
dates.reset_index()
dates = dates.unstack()

# multi-indexed dataframe
df = pd.DataFrame(np.random.randn(36, 3))
df['concept'] = np.repeat(np.repeat(['A', 'B', 'C'], 3), 4)
df['datetime'] = pd.concat([dates, dates, dates], ignore_index=True)
df.set_index(['concept', 'datetime'], inplace=True)
df.sort_index(inplace=True)
df.columns = ['V1', 'V2', 'V3']
df.info()

returning:

                                   V1        V2        V3
concept datetime                                         
A       2016-01-01 00:00:00 -0.303428  0.088180 -0.547776
        2016-01-01 01:00:00 -0.893835 -2.226923 -0.181370
        2016-01-01 02:00:00  2.934575  1.515822  0.343609
        2016-01-01 03:00:00 -1.341694  1.681015  0.099759
        2017-01-01 00:00:00  1.515894  0.519595  0.102635
        2017-01-01 01:00:00 -0.266949 -0.035901  0.539084
        2017-01-01 02:00:00  1.336603  0.286928 -0.352078
        2017-01-01 03:00:00  0.480137  0.185785  0.595706
        2018-01-01 00:00:00 -0.385640  1.813604 -0.839973
        2018-01-01 01:00:00  0.568706  1.165257 -1.352020
        2018-01-01 02:00:00  0.498388  0.382034 -1.190599
        2018-01-01 03:00:00  1.897356 -0.293143  0.177787
B       2016-01-01 00:00:00 -1.111196 -1.644588  0.333936
        2016-01-01 01:00:00  0.232206 -0.202987 -0.334564
        2016-01-01 02:00:00  1.264637 -1.472229  0.888451
        2016-01-01 03:00:00  1.033163  0.504090  1.325476
        2017-01-01 00:00:00 -0.199445  0.088792 -0.797965
        2017-01-01 01:00:00 -1.116359  0.574789 -1.055830
        2017-01-01 02:00:00  1.267970  0.287501  0.001420
        2017-01-01 03:00:00  1.554647  2.865833  0.089875
        2018-01-01 00:00:00  0.030871 -1.783524 -1.457190
        2018-01-01 01:00:00  0.073978 -0.735599 -0.420115
        2018-01-01 02:00:00  0.931073 -2.543869 -0.649976
        2018-01-01 03:00:00  0.325443  1.134799  0.445788
C       2016-01-01 00:00:00 -0.489454 -0.646136 -0.111308
        2016-01-01 01:00:00 -0.501965 -0.197183  0.025899
        2016-01-01 02:00:00 -0.714251 -1.846856  0.197658
        2016-01-01 03:00:00  0.609357  0.456263 -0.041581
        2017-01-01 00:00:00 -1.004726 -0.956688 -0.068980
        2017-01-01 01:00:00 -0.036204 -1.236450 -0.895681
        2017-01-01 02:00:00 -0.840374  0.561443  1.401854
        2017-01-01 03:00:00  0.325433  1.406280 -1.033267
        2018-01-01 00:00:00 -0.029315 -1.591510 -0.739032
        2018-01-01 01:00:00 -0.761522 -0.896236  0.537450
        2018-01-01 02:00:00  1.081961  0.126248 -0.911462
        2018-01-01 03:00:00  0.070915 -1.036460  1.187859

and want to plot one grouped column in a boxplot:

# demonstrate how to customize the display different elements:
boxprops = dict(linestyle='-', linewidth=4, color='k')
medianprops = dict(linestyle='-', linewidth=4, color='k')

ax = df.boxplot(column=['V1'],
                by=df.index.get_level_values('datetime').year,
                showfliers=False, showmeans=True,
                boxprops=boxprops,
                medianprops=medianprops)
# get rid of the automatic title
plt.suptitle("")
ax.set_xlabel("")
ax.set_title("Boxplot of V1")

returning:

Obviously, some styling options for the boxplot are working and some are not.

So here's my question:

How can I set the color of the box/median/mean?

Thanks in advance!

############################ EDIT 1 ############################

I have found this answer and adapted my plot:

bp = data.boxplot(column=['eex_da_price_mean'],
                  by=data.index.get_level_values('date').year,
                  showfliers=False, showmeans=True,
                  return_type='dict')

[[item.set_linewidth(4) for item in bp[key]['boxes']] for key in bp.keys()]
[[item.set_linewidth(4) for item in bp[key]['fliers']] for key in bp.keys()]
[[item.set_linewidth(4) for item in bp[key]['medians']] for key in bp.keys()]
[[item.set_linewidth(4) for item in bp[key]['means']] for key in bp.keys()]
[[item.set_linewidth(4) for item in bp[key]['whiskers']] for key in bp.keys()]
[[item.set_linewidth(4) for item in bp[key]['caps']] for key in bp.keys()]

bp.set_xlabel("")
bp.set_title("Some plot", fontsize=60)
bp.tick_params(axis='y', labelsize=60)
bp.tick_params(axis='x', labelsize=60)
plt.suptitle("")

returns:

But now the axis formatting does not work anymore and I get errors like this:

bp.set_xlabel("")
AttributeError: 'OrderedDict' object has no attribute 'set_xlabel'

Any hints?

score 19 · Answer 1 · answered Jan 03 '19 at 13:50

19

I just found another solution to plot with much less code directly from pandas (without having to manipulate the matplotlib-object afterwards):

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt


df = pd.DataFrame(np.random.rand(10, 5), columns=['A', 'B', 'C', 'D', 'E'])
ax = df.plot(kind='box',
             color=dict(boxes='r', whiskers='r', medians='r', caps='r'),
             boxprops=dict(linestyle='-', linewidth=1.5),
             flierprops=dict(linestyle='-', linewidth=1.5),
             medianprops=dict(linestyle='-', linewidth=1.5),
             whiskerprops=dict(linestyle='-', linewidth=1.5),
             capprops=dict(linestyle='-', linewidth=1.5),
             showfliers=False, grid=True, rot=0)
ax.set_xlabel('Foo')
ax.set_ylabel('Bar in X')
plt.show()

yields:

The only thing I haven't figured out is how to adjust the color of the means when showmeans=True. But in most cases this should be fine..

Hope it helps!

answered Jan 03 '19 at 13:50

Cord Kaldemeyer

6,405
8
51
81

Any idea how to apply separate properties to each box AND have them plotted on one Axes object, as in your example? If I plot them individually, passing the argument `ax=ax` so they all use the same Axes, they of course just then get overlaid on one another, right in the middle. – n1k31t4 May 08 '19 at 21:57
And to fill the boxes with color, use this: `patch_artist=True`. https://stackoverflow.com/questions/39297093/change-the-facecolor-of-boxplot-in-pandas – Nav Jan 03 '21 at 16:34
If people also try to colour the outliers, you will need to add `flierprops` to your `.plot()` command. E.g. `flierprops = dict(marker='o', markerfacecolor='r', markersize=12, linestyle='none', markeredgecolor='g')` – Felix Seifert Feb 28 '22 at 14:42

score 13 · Accepted Answer · answered Feb 04 '16 at 09:33

Screenpavers answer worked well.

Here's a complete example:

# -*- coding: utf-8 -*-
import numpy as np
import pandas as pd

# dataframe with dates
dates = pd.DataFrame()
dates['2016'] = pd.date_range(start='2016', periods=4, freq='60Min')
dates['2017'] = pd.date_range(start='2017', periods=4, freq='60Min')
dates['2018'] = pd.date_range(start='2018', periods=4, freq='60Min')
dates.reset_index()
dates = dates.unstack()

# multi-indexed dataframe
df = pd.DataFrame(np.random.randn(36, 3))
df['concept'] = np.repeat(np.repeat(['A', 'B', 'C'], 3), 4)
df['datetime'] = pd.concat([dates, dates, dates], ignore_index=True)
df.set_index(['concept', 'datetime'], inplace=True)
df.sort_index(inplace=True)
df.columns = ['V1', 'V2', 'V3']
df.info()


# demonstrate how to customize the display different elements:
boxprops = dict(linestyle='-', linewidth=4, color='k')
medianprops = dict(linestyle='-', linewidth=4, color='k')

bp = df.boxplot(column=['V1'],
                by=df.index.get_level_values('datetime').year,
                showfliers=False, showmeans=True,
                boxprops=boxprops, medianprops=medianprops,
                return_type='dict')

# boxplot style adjustments
[[item.set_linewidth(4) for item in bp[key]['boxes']] for key in bp.keys()]
[[item.set_linewidth(4) for item in bp[key]['fliers']] for key in bp.keys()]
[[item.set_linewidth(4) for item in bp[key]['medians']] for key in bp.keys()]
[[item.set_linewidth(4) for item in bp[key]['means']] for key in bp.keys()]
[[item.set_linewidth(4) for item in bp[key]['whiskers']] for key in bp.keys()]
[[item.set_linewidth(4) for item in bp[key]['caps']] for key in bp.keys()]

[[item.set_color('g') for item in bp[key]['boxes']] for key in bp.keys()]
# seems to have no effect
[[item.set_color('b') for item in bp[key]['fliers']] for key in bp.keys()]
[[item.set_color('m') for item in bp[key]['medians']] for key in bp.keys()]
[[item.set_markerfacecolor('k') for item in bp[key]['means']] for key in bp.keys()]
[[item.set_color('c') for item in bp[key]['whiskers']] for key in bp.keys()]
[[item.set_color('y') for item in bp[key]['caps']] for key in bp.keys()]

# get rid of "boxplot grouped by" title
plt.suptitle("")

# label adjustment
p = plt.gca()
p.set_xlabel("")
p.set_title("Some plot", fontsize=30)
p.tick_params(axis='y', labelsize=30)
p.tick_params(axis='x', labelsize=30)

returns:

As of Pandas version 0.23.4 the working code is as follows: `[item.set_color('crimson') for item in bp['boxes']]` — Yakzan, Sep 21 '18 at 13:19

score 3 · Answer 3 · answered Feb 03 '16 at 19:08

3

Before your bp.set_xlabel("") statement, try this instead:

p = plt.gca()
p.set_xlabel("")
p.set_title("Some plot", fontsize=60)
p.tick_params(axis='y', labelsize=60)
p.tick_params(axis='x', labelsize=60)

answered Feb 03 '16 at 19:08

screenpaver

1,120
8
14

score 0 · Answer 4 · answered May 24 '20 at 12:58

0

Try seaborn

# Box Plot
import seaborn as sns
%matplotlib inline
sns.boxplot(data=data['fixed acidity'])
plt.show()

answered May 24 '20 at 12:58

Ankit Kumar Rajpoot

5,188
2
38
32

2

I dunno man...seaborn's boxplots look like a baobab tree. Not something I'd ever like to use. – Nav Jan 03 '21 at 15:47

Pandas boxplot: set color and properties for box, median, mean

4 Answers4

Linked