Matplotlib（GridSpec）-子图轴标签被截断了吗？ 尝试过`tight_layout`

Question

I'm using GridSpec to organise subplots. I have a shared colorbar for all the plots.

All suggestions online seem to point out that tight_layout() is the way to fix issues with axis labels cutting off, however this doesn't seem to be working here (unless it comes in another form which I am unaware of).

I have also tried using the rect parameter of tight_layout for fig , plt , and gs .

import matplotlib
import matplotlib.pyplot as plt
import numpy as np 
from pylab import *
import matplotlib.gridspec as gridspec
import matplotlib.colors
from mpl_toolkits.mplot3d import Axes3D

gs = gridspec.GridSpec(1,7,hspace=0.05,wspace=0.5, width_ratios=[1,1,1,1,1,1,0.1])
figure(num=None, figsize=(18, 2), dpi=80, facecolor='w', edgecolor='k')

data = np.random.rand(3,6,224,5)

for i in range(6):
    ax = plt.subplot(gs[0, i], projection='3d')
    p = ax.scatter(data[0,i,:,0], data[0,i,:,1], data[0,i,:,2], c=data[0,i,:,4], marker='o')
    title("Case " + str(i+1))
    ax.set_xlabel('Batch Size', linespacing=3)
    ax.set_ylabel('Window Size', linespacing=3)
    ax.set_zlabel('Neurons', linespacing=3)
    ax.xaxis.labelpad=20
    ax.yaxis.labelpad=20
    ax.zaxis.labelpad=10

cbar = plt.subplot(gs[0,6])
colorbar(p, cax=cbar, label='RMSE')

plt.show()

This generates the image below.

Answer 1

As commented, setting the bottom parameter to a larger value, eg bottom=0.3 would give you more space to accomodate the axes decorators.

In addition it may be useful to make the figure a little bit taller (eg 3 inch instead of 2) in order not to shrink the plots too much.

import matplotlib.pyplot as plt
import numpy as np 
import matplotlib.gridspec as gridspec
from mpl_toolkits.mplot3d import Axes3D

gs = gridspec.GridSpec(1,7,hspace=0.05,wspace=0.5, bottom=0.3,
                       left=0.02, right=0.95, width_ratios=[1,1,1,1,1,1,0.1])
fig = plt.figure(figsize=(18, 3), dpi=80, facecolor='w', edgecolor='k')

data = np.random.rand(3,6,224,5)

for i in range(6):
    ax = plt.subplot(gs[0, i], projection='3d')
    p = ax.scatter(data[0,i,:,0], data[0,i,:,1], data[0,i,:,2], 
                   c=data[0,i,:,4], marker='o')
    ax.set_title("Case " + str(i+1))
    ax.set_xlabel('Batch Size', linespacing=3)
    ax.set_ylabel('Window Size', linespacing=3)
    ax.set_zlabel('Neurons', linespacing=3)
    ax.xaxis.labelpad=20
    ax.yaxis.labelpad=20
    ax.zaxis.labelpad=10

cbar = plt.subplot(gs[0,6])
fig.colorbar(p, cax=cbar, label='RMSE')

# This is only needed for jupyter
fig.add_axes([0,0,1,1]).axis("off")

plt.show()

Unfortunately in jupyter the %matplotlib inline backend always creates its images with a the bbox_inches = "tight" setting. So a workaround is to create some element in the figure that ensures that the "tight" area is large enough. Here, an option is to use fig.add_axes([0,0,1,1]).axis("off") .