Python: Sharing scale between matplotlib and seaborn

Question

I'd like to create a seaborn heatmap which has also scatter plot color points. I'd like the final result to use the grid of the scatter plot, with the squares of the heatmap being "centered" on the scatter points.

Unfortunately, I don't find how to share scales between the two layers, as shown in the example below.

What can I do?

Thanks a lot for your help.

%matplotlib inline
import seaborn as sns
import matplotlib.pyplot as plt
import pandas as pd
import numpy as np

npoints = 3
x = np.tile(np.arange(npoints), npoints)
df = pd.DataFrame({'x': np.tile(np.arange(npoints), npoints), 'y': np.repeat(np.arange(npoints), npoints)})

df['z'] = 0
df.loc[df['x'] == df['y'], 'z'] = df.loc[df['x'] == df['y'], 'x']
df['c'] = np.random.choice(np.arange(3) + 1, df.shape[0])
df.loc[df['x'] != df['y'], 'c'] = 0

sns.heatmap(df[['x', 'y', 'z']].set_index(['x', 'y'])['z'].unstack())
plt.gca().set_title('Heatmap only')

df.plot(x='x', y='y', color=df['c'], kind='scatter')
plt.gca().set_title('Scatter points only')

fig, ax = plt.subplots()
sns.heatmap(df[['x', 'y', 'z']].set_index(['x', 'y'])['z'].unstack(), ax=ax)
df.plot(x='x', y='y', ax=ax, color=df['c'], kind='scatter')
ax.set_title('Heatmap and scatter points - scales problem')

This question is a little misphrased because there isn't really a seaborn "layer" in the plot, seaborn only ever adds matplotlib objects. Your issue isn't about sharing scales, it is in realizing that the cells in the heatmap span integral values (look at, e.g.` ax.get_xticks()`). — mwaskom
– mwaskom, Commented Nov 4, 2016 at 18:34

Kennet Celeste · Accepted Answer · 2016-11-04 18:19:55Z

1

A workaround would be to shift your scatter data by 0.5:

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

npoints = 3
x = np.tile(np.arange(npoints), npoints)
df = pd.DataFrame({'x': np.tile(np.arange(npoints), npoints), 'y': np.repeat(np.arange(npoints), npoints)})

df['z'] = 0
df.loc[df['x'] == df['y'], 'z'] = df.loc[df['x'] == df['y'], 'x']
df['c'] = np.random.choice(np.arange(3) + 1, df.shape[0])
df.loc[df['x'] != df['y'], 'c'] = 0

fig, ax = plt.subplots()
qp = sns.heatmap(df[['x', 'y', 'z']].set_index(['x', 'y'])['z'].unstack(), ax=ax)
# df.plot(x='x', y='y', ax=ax, color=df['c'], kind='scatter')
ax.scatter(df['x']+0.5,df['y']+0.5,c=df['c'])

ax.set_title('Heatmap and scatter points - scales problem')

plt.show()

result:

answered Nov 4, 2016 at 18:19

Kennet Celeste

4,8113 gold badges28 silver badges34 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Python: Sharing scale between matplotlib and seaborn

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related