No output of xticklabels when plotting to gridspec subplots

[JWarmenhoven]

What seems to be the problem here?
The xticklabels for the first axes defined in a gridspec are not visible when plotting with pandas. The corresponding plot using pyplot.plot does have the labels.

pandas 0.16
matplotlib 1.4.3

import pandas as pd
import numpy as np
import matplotlib as mpl
import matplotlib.pyplot as plt

%matplotlib inline
df = pd.DataFrame(np.random.randn(8, 4), index=np.arange(8), columns=['A', 'B', 'C', 'D'])

Matplotlib

gs = mpl.gridspec.GridSpec(2, 4)
ax1 = plt.subplot(gs[0,:-1])
ax2 = plt.subplot(gs[0,-1])
ax3 = plt.subplot(gs[1,:-2])
ax4 = plt.subplot(gs[1,-2:])

ax1.plot(df.index.values, df.A)
ax2.plot(df.index.values, df.B)
ax3.plot(df.index.values, df.C)
ax4.plot(df.index.values, df.D)

gs.tight_layout(plt.gcf())

pandas

gs = mpl.gridspec.GridSpec(2, 4)
ax1 = plt.subplot(gs[0,:-1])
ax2 = plt.subplot(gs[0,-1])
ax3 = plt.subplot(gs[1,:-2])
ax4 = plt.subplot(gs[1,-2:])

df.A.plot(ax=ax1)
df.B.plot(ax=ax2)
df.C.plot(ax=ax3)
df.D.plot(ax=ax4)

gs.tight_layout(plt.gcf())

It seems that xticklabels are absent when plotting with pandas to the first axis generated in a gridspec. Switching ax1 and ax2 definition:

pandas

gs = mpl.gridspec.GridSpec(2, 4)
ax2 = plt.subplot(gs[0,-1])
ax1 = plt.subplot(gs[0,:-1])
ax3 = plt.subplot(gs[1,:-2])
ax4 = plt.subplot(gs[1,-2:])

df.A.plot(ax=ax1)
df.B.plot(ax=ax2)
df.C.plot(ax=ax3)
df.D.plot(ax=ax4)

gs.tight_layout(plt.gcf())

Related?
http://stackoverflow.com/questions/29552382/legend-overriding-x-label-and-x-ticks

[TomAugspurger]

Could you give it a shot on a newer version of pandas? All 3 are the same for me (master pandas, matplotlib 1.4.3)

[JWarmenhoven]

Thanks! That seems to have solved the problem...

pandas 0.16.1
matplotlib 1.4.3

gs = mpl.gridspec.GridSpec(2, 4)
ax1 = plt.subplot(gs[0,:-1])
ax2 = plt.subplot(gs[0,-1])
ax3 = plt.subplot(gs[1,:-2])
ax4 = plt.subplot(gs[1,-2:])

df.A.plot(ax=ax1)
df.B.plot(ax=ax2)
df.C.plot(ax=ax3)
df.D.plot(ax=ax4)

gs.tight_layout(plt.gcf())

[jorisvandenbossche]

Maybe we should add a test for this, to ensure this keeps working correctly in the future.
@JWarmenhoven Are you interested to do this?

[JWarmenhoven]

Interested, yes. But since this is not my daily bread...it might take a while :-)
For starters, I wonder what bugfix in 0.16.1 solved this problem.
#9158 perhaps?

[jorisvandenbossche]

That's possible.
If you have any questions how to get started with this, just ask! (here or on https://gitter.im/pydata/pandas)