seaborn histplot and displot output doesn't match

Question

• Histograms generated by seaborn.histplot and seaborn.displot do not match.
  • Default plot for sns.displot is kind='hist'
• Tested with python3.8.11, seaborn 0.11.2, and matplotlib 3.4.2
• Why do the outputs not match, and how can this be resolved?
• The expectation is, given bins, the density of the corresponding plots should match.
  • bins is passed to numpy.histogram_bin_edges
• Information contained in Visualizing distributions of data doesn't resolve the question.

import seaborn as sns
import matplotlib.pyplot as plt

# sample data: wide
dfw = sns.load_dataset("penguins", cache=False)[['bill_length_mm', 'bill_depth_mm']].dropna()

# sample data: long
dfl = dfw.melt(var_name='bill_size', value_name='vals')

seaborn.displot

1. Ignores 'sharex': False, though 'sharey' works
2. Ignores bins

fg = sns.displot(data=dfl, x='vals', col='bill_size', kde=True, stat='density', bins=12, height=4, facet_kws={'sharey': False, 'sharex': False})
plt.show()

3. Setting xlim doesn't make a difference

fg = sns.displot(data=dfl, x='vals', col='bill_size', kde=True, stat='density', bins=12, height=4, facet_kws={'sharey': False, 'sharex': False})
axes = fg.axes.ravel()
axes[0].set_xlim(25, 65)
axes[1].set_xlim(13, 26)
plt.show()

seaborn.histplot

fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(8, 4))

sns.histplot(data=dfw.bill_length_mm, kde=True, stat='density', bins=12, ax=ax1)
sns.histplot(data=dfw.bill_depth_mm, kde=True, stat='density', bins=12, ax=ax2)
fig.tight_layout()
plt.show()

Update

• As suggested by mwaskom, common_bins=False gets the histograms into the same shape, resolving the issues of ignoring bins and sharex. However, the density seems to be affected by the number of plots in the displot.
  • If there are 3 plots in the displot, then the density is 1/3 that shown in the histplot; for 2 plots, the density is 1/2.

Oriya

Answer 1

• As suggested by mwaskom in a comment, common_bins=False gets the histograms into the same shape, resolving the issues of ignoring bins and sharex, and the density in a faceted plot is scaled by the number of data points in each facet, not the number of facets.
• The issue with density being split by the number of plots in displot is resolved by using common_norm=False

Plot Code

# displot
fg = sns.displot(data=dfl, x='vals', col='bill_size', kde=True, stat='density', bins=12, height=4,
                 facet_kws={'sharey': False, 'sharex': False}, common_bins=False, common_norm=False)

fg.fig.subplots_adjust(top=0.85)
fg.fig.suptitle('Displot with common_bins & common_norm as False')
plt.show()

# histplot
fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(8, 4))

sns.histplot(data=dfw.bill_length_mm, kde=True, stat='density', bins=12, ax=ax1)
sns.histplot(data=dfw.bill_depth_mm, kde=True, stat='density', bins=12, ax=ax2)

fig.subplots_adjust(top=0.85)
fig.suptitle('Histplot')

fig.tight_layout()
plt.show()

Oriya