Labels are cut off

Question

Hi I am using the following code in order to plot a 3D chart:

%matplotlib inline


fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
ax.scatter(rf_pd['max_depth'], rf_pd['n_estimators'], [1] - rf_pd['mean_train_score'], c='red', s=60)
ax.view_init(50, 320)
ax.set(xlabel='Max depth', ylabel='Nº estimators', zlabel='Error: 1 - F1 Score')
plt.title('Random Forest - F1 error al entrenar por parámetros')
fig.tight_layout()
plt.show()
fig.savefig('a.png')

And I am getting the following image:

The questions is how can fix the fact of the labels (z) and ticks in X and Y are cut off?

Thanks

and if you add some padding, e.g. `fig.tight_layout(pad=5.0)`? — Stef, Jun 26 '20 at 13:30
With that the preview is worst because the ticks are too near — Tlaloc-ES, Jun 26 '20 at 15:47

score 2 · Accepted Answer · answered Jun 27 '20 at 00:44

If you have Matplotlib >= 3.1 you can do

fig = plt.figure(constrained_layout=True)

and remove fig.tight_layout()

To match the behavior of the image shown inline do

fig.savefig('a.png', bbox_inches='tight')

which will "shrink wrap" the figure to make sure all of the artists are visible. The down side is you lose control of the final size of the final saved figure.

If you want to keep control of the size of the final output and have an older version of Matplotlib removing fig.tight_layout() and manually adjust the location of the Axes until it looks good via ax.set_position (https://matplotlib.org/3.3.0/api/_as_gen/matplotlib.axes.Axes.set_position.html)

This output will be slightly different in mpl 3.3 (see https://matplotlib.org/3.3.0/users/whats_new.html#axes3d-no-longer-distorts-the-3d-plot-to-match-the-2d-aspect-ratio)

Does this mean that if I am using mpl>=3.1, I will have to compromise between a cut off 3d plot with my desired figsize or a 3d plot with all artist visible but with a random figsize? — JaffXXI, Jan 18 '22 at 10:18

Labels are cut off

1 Answers1

Linked