Set no title for pandas boxplot (groupby)

Question

When drawing a pandas boxplot, grouped by another column, pandas automatically adds a title to the plot, saying 'Boxplot grouped by....'. Is there a way to remove that? I tried using

suptitle('')

as per Pandas: boxplot of one column based on another column

but this does not seem to work. I am using latest pandas (0.13.1) version.

Transgress

Answer 1

Make sure your calling suptitle('') on the right figure.

In [23]: axes = df.boxplot(by='g')

In [24]: fig = axes[0][0].get_figure()

In [25]: fig.suptitle('')
Out[25]: <matplotlib.text.Text at 0x109496090>

Topliffe

I'm getting 'AxesSubplot' object is not subscriptable when I try this. Has something changed since this answer in 2014? – Everick

I'm having the same problem. The answer from nick (Jan 20, 2019) worked for me. – Zulmazulu

Try: axes.iloc[0].get_figure() If you look at the axes variable, you'll likely see it's a pandas Series. And you need to get any of the several Axes in this Series and then call get_figure on it. By doing iloc[0] you'll get the first element in the Series – Busybody

It should be axes.get_figure().suptitle("") – Looselimbed

Answer 2

I had the same problem. Ended up using this solution

import matplotlib.pyplot as plt    
# df is your dataframe
df.boxplot(column='value', by='category')
title_boxplot = 'awesome title'
plt.title( title_boxplot )
plt.suptitle('') # that's what you're after
plt.show()

Eada

Hi nick. I notice that you're still suggesting edits that just add the dataframe tag to questions. Please don't. I realise that the pandas tag excerpt used to explicitly say that this should be done, but it's been recently updated since that guidance didn't really make sense - see meta.https://mcmap.net/q/374256/-getting-python-to-work-internal-server-error/1709587 for the discussion about this. Not your fault since you were acting in good faith on guidance in the excerpt, but those edits are pointless at best and require reviewer time. Feel free to flag this comment as "no longer needed" once you've read it. – Makassar

Answer 3

In Pandas 1.1.4, the following worked for me:

ax = pft[[col, "abuse_marked"]].boxplot(by="abuse_marked", vert=False)
ax.get_figure().suptitle("")

Rathbun

Answer 4

After trying all the suggestions, only this modification worked for me, which also lets you modify other parameters:

ax = df.boxplot(by ='value', column =['category'], grid = False);
plt.title('')
plt.suptitle('')
ax.set_title('');
ax.set_xlabel("x_label");
ax.set_ylabel("y_label");
ax = plt.show()

Flank

Answer 5

None of the above solutions worked for me, but this one did:

axes = df.boxplot(column=values, by=index, ax=ax, rot=90)
axes.set_title('')

Dagan

Answer 6

I as having problems with this and generally never liked the canned title that the pandas was adding as it was dependent on the column names which are typically never publishing ready.

You can edit the source code in ~\pandas\plotting\_core.py

On line 2698 you will find:

fig.suptitle('Boxplot grouped by {byline}'.format(byline=byline))

Simple comment this line out and pandas will no longer add the title to the top of the boxplot by default. You will have to redo this change as you upgrade pandas versions.

Primary

Never change the source library locally. Your results will never be replicable. – Vidavidal

Answer 7

This worked for me:

ax = df.boxplot(column = col1, by = col2)

ax.get_figure().suptitle()

Ricercar

Answer 8

Although the previous answers correctly remove the text in the suptitle, they still leave space for it. This might be an issue in certain cases, such as when using fig.tight_layout(). Therefore, to completely remove the title, this can be done:

axes = df.boxplot(by='g')
fig = axes[0][0].get_figure()
fig._suptitle.set_visible(False)
fig._suptitle.set_in_layout(False)

Isis