seaborn 配对图旁边的颜色条

Question

from sklearn.datasets import make_blobs
import matplotlib.pyplot as plt
from pandas import DataFrame

# generate 2d classification dataset
X, y = make_blobs(n_samples=100, centers=3, n_features=2)
# scatter plot, dots colored by class value
df1 = DataFrame(dict(x=X[:,0], y=X[:,1], label=y))


norm = plt.Normalize(df1.x.min(), df1.x.max())
sm = plt.cm.ScalarMappable(cmap=sns.cubehelix_palette(df1['x'].max(), start=.5, rot=-.75,as_cmap=True), norm=norm)
sm.set_array([])

ax=sns.pairplot(df1,vars=['x','y'], 
            hue='label',
            palette=sns.cubehelix_palette(df1['x'].max(), start=.5, rot=-.75),
            diag_kind=None,plot_kws={"s": 50})

ax._legend.remove()

# ax.set_ylabel('WT04: Pairplot for features')
cbar=ax.fig.colorbar(sm)
m0=int(np.floor(df1.x.min()))            # colorbar min value
m4=int(np.ceil(df1.x.max()))             # colorbar max value
m1=int(1*(m4-m0)/4.0 + m0)               # colorbar mid value 1
m2=int(2*(m4-m0)/4.0 + m0)               # colorbar mid value 2
m3=int(3*(m4-m0)/4.0 + m0)               # colorbar mid value 3
cbar.set_ticks([m0,m1,m2,m3,m4])
cbar.set_ticklabels([m0,m1,m2,m3,m4])

ax.fig.suptitle('WT04: Pairplot for features',y=0.99)
plt.subplots_adjust(top=0.94)
plt.show()

plt.savefig('result.png')

Generates this plot

However, I would like to have the colorbar to the right of the pairplot, and not the bottom right. How can this be done? Some of my pairplots have other dimensions, so something scalable would be nice.

Answer 1

You can try adding the colorbar with plt :

ax._legend.remove()

# remove this colorbar
# cbar=ax.fig.colorbar(sm)

m0=int(np.floor(df1.x.min()))            # colorbar min value
m4=int(np.ceil(df1.x.max()))             # colorbar max value
m1=int(1*(m4-m0)/4.0 + m0)               # colorbar mid value 1
m2=int(2*(m4-m0)/4.0 + m0)               # colorbar mid value 2
m3=int(3*(m4-m0)/4.0 + m0)               # colorbar mid value 3

ax.fig.suptitle('WT04: Pairplot for features',y=0.99)

# use this colorbar
cbar = plt.colorbar(sm, ax=ax.axes)

cbar.set_ticks([m0,m1,m2,m3,m4])
cbar.set_ticklabels([m0,m1,m2,m3,m4])

Output: