Multiple seaborn pointplots not showing the right color on the legend

Question

I have three point plot i'm trying to chart and show a legend. The colors do not match the colors called out in the plots. I tried using the solution from this post, but that did not work.

Here is the code I'm using:

fig, ax = plt.subplots()
a = sns.pointplot(x=l[1:], y = np.exp(model_m.params[1:]), label = 'factor',
              ax = ax, color = 'green')

b = sns.pointplot(x=l[1:], y = np.exp(model_m.conf_int()[1:][:,1]), 
              ax = ax, label = 'conf_int+', color = 'red')

c = sns.pointplot(x=l[1:], y = np.exp(model_m.conf_int()[1:][:,0]), 
              ax = ax, label = 'conf_int-', color = 'blue')
plt.title('Model M Discrete')
ax.legend(labels = ['factor', 'conf_inf+', 'conf_inf-'],
           title = 'legend')

Here is what it produces:

This might be a dumb comment but are you missing the list of handles in the call to `ax.legend()`? That is, does `ax.legend(handles = [a, b, c], labels = ['factor', 'conf_inf+', 'conf_inf-'], title = 'legend')` do anything? — havingaball, Nov 11 '21 at 20:06

JohanC · Accepted Answer · 2021-11-11T21:59:49.487

The easiest solution would be to use sns.lineplot instead of sns.pointplot:

import matplotlib.pyplot as plt
import seaborn as sns
import numpy as np

fig, ax = plt.subplots()
x = np.arange(10)
sns.lineplot(x=x, y=1 + np.random.rand(10).cumsum(),
             ax=ax, label='factor', color='green', marker='o')
sns.lineplot(x=x, y=2 + np.random.rand(10).cumsum(),
             ax=ax, label='conf_int+', color='red', marker='o')
sns.lineplot(x=x, y=3 + np.random.rand(10).cumsum(),
             ax=ax, label='conf_int-', color='blue', marker='o')
ax.set_title('Model M Discrete')
ax.legend(title='legend')
plt.tight_layout()
plt.show()

Another option would be to iterate through the generated "pathCollections" and assign a label (for some reason label= doesn't work in sns.pointplot).

fig, ax = plt.subplots()
sns.pointplot(x=x, y=1 + np.random.rand(10).cumsum(),
              ax=ax, color='green')
sns.pointplot(x=x, y=2 + np.random.rand(10).cumsum(),
              ax=ax, color='red')
sns.pointplot(x=x, y=3 + np.random.rand(10).cumsum(),
              ax=ax, label='conf_int-', color='blue')
for curve, label in zip(ax.collections, ['factor', 'conf_int+', 'conf_int-']):
    curve.set_label(label)
ax.set_title('Model M Discrete')
ax.legend(title='legend')

Still another way is to mimic a long form dataframe with hue which automatically creates a legend:

fig, ax = plt.subplots()
x = np.arange(10)
y1 = 1 + np.random.rand(10).cumsum()
y2 = 2 + np.random.rand(10).cumsum()
y3 = 3 + np.random.rand(10).cumsum()
sns.pointplot(x=np.tile(x, 3),
              y=np.concatenate([y1, y2, y3]),
              hue=np.repeat(['factor', 'conf_int+', 'conf_int-'], len(x)),
              ax=ax, palette=['green', 'red', 'blue'])

Note that in both cases only a dot is shown in the legend, not a line.

Multiple seaborn pointplots not showing the right color on the legend

1 Answers1