How to modify a legend with hatches

Question

I have created a barplot with hatches with the code below:

import matplotlib.pyplot as plt
import seaborn as sns

tips = sns.load_dataset("tips")

hatches = ['\\\\', '//']

fig, ax = plt.subplots(figsize=(6,3))
sns.barplot(data=tips, x="day", y="total_bill", hue="time")

# loop through days
for hues, hatch in zip(ax.containers, hatches):
    # set a different hatch for each time
    for hue in hues:
        hue.set_hatch(hatch)

# add legend with texture
plt.legend()

plt.show()

And it outputs the following plot:

If I try to modify the legend to (a) add a title and (b) change the hue labels, the code breaks down, and I lose the color and hatches from the legend:

import matplotlib.pyplot as plt
import seaborn as sns

tips = sns.load_dataset("tips")

hatches = ['\\\\', '//']

fig, ax = plt.subplots(figsize=(6,3))
sns.barplot(data=tips, x="day", y="total_bill", hue="time")

# loop through days
for hues, hatch in zip(ax.containers, hatches):
    # set a different hatch for each time
    for hue in hues:
        hue.set_hatch(hatch)

# add legend with texture
plt.legend(title='My Title', labels=['Lunch_1','Dinner_2'], loc=1)

plt.show()

So my questions is: How do I add a title and modify the labels without losing the color and hatches from the legend?

score 3 · Accepted Answer · answered May 20 '23 at 06:23

You need to call legend with the signature legend(handles, labels) after retrieving the original ones with get_legend_handles_labels :

handles, _ = ax.get_legend_handles_labels()

plt.legend(handles, ["Lunch_1", "Dinner_2"], title="My Title", loc=1)

Output :

Trenton McKinney · Answer 2 · 2023-05-20T19:02:09.023

It's a best practice to clean the data in a DataFrame prior to plotting, and analysis. As such, the easiest way to deal with legend labels/titles, and axis labels, is to clean associated columns.
- Update the column name used for hue (the legend title), with .rename or .str methods, and the categorical column values (the legend labels) with .map.
Move seaborn plot legend to a different position shows how to move seaborn axes level legends with plt.legend (implicit), ax.legend (explicit), and the preferred sns.move_legend.
- However, in this case, plt.legend or ax.legend must be used for the hatches to show up on the legend handles, which requires resetting the legend title.
- Now the legend labels don't need to be updated, so it's not necessary to pass extracted handles.
Tested in python 3.11.2, pandas 2.0.1, matplotlib 3.7.1, seaborn 0.12.2

import seaborn as sns
import matplotlib.pyplot as plt

tips = sns.load_dataset('tips')

# change column names for better presentation of axis labels and titles
tips.columns = tips.columns.str.replace('_', ' ').str.title()

# change a specific column name
tips = tips.rename({'Time': 'Hobbit Meals', 'Day': 'Shire Days'}, axis=1)

# change values in the column to have a better value
tips['Hobbit Meals'] = tips['Hobbit Meals'].map({'Dinner': 'Supper', 'Lunch': 'Elevensies'})
tips['Shire Days'] = tips['Shire Days'].map({'Thur': 'Trewsday', 'Fri': 'Hevensday', 'Sat': 'Mersday', 'Sun': 'Highday'})

fig, ax = plt.subplots(figsize=(6, 3))
sns.barplot(data=tips, x='Shire Days', y='Total Bill', hue='Hobbit Meals', ax=ax)

# loop through axes bar containers
for c, hatch in zip(ax.containers, ['\\\\', '//']):
    # iterate through each patch in the container
    for patch in c:
        patch.set_hatch(hatch)

# in order to have the hatches appear on the legend hanles, the lengend method must me called
ax.legend(title=tips.columns[5], bbox_to_anchor=(1, 0.5), loc='center left', frameon=False)

# remove the spines for some additionaly cosmetic changes
ax.spines[['right', 'top']].set_visible(False)

How to modify a legend with hatches

2 Answers2