df.plot(ax=ax) AttributeError: 'Axes' object has no attribute 'is_first_col'

[dragoljub]

When attempting to use the Pandas 0.17.0 plot method to target plotting on an external axes: AttributeError: 'Axes' object has no attribute 'is_first_col'.

Pandas: 0.17.0
Numpy: 1.9.2
Matplotlib: 1.5.0

import pandas as pd
import numpy as np
from matplotlib import pyplot as plt

df = pd.DataFrame({'A' : ['foo', 'bar', 'foo', 'bar',
                          'foo', 'bar', 'foo', 'foo'],
                    'B' : ['one', 'one', 'two', 'three',
                          'two', 'two', 'one', 'three'],
                    'C' : np.random.randn(8),
                    'D' : np.random.randn(8)})

fig = plt.figure()
ax = fig.add_axes((0,0,1,1))

df.plot(kind='scatter', ax=ax, x='C', y='D', c=df.C, s=100, linewidth=0, colormap='hsv')
---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-21-b38f872a3fc3> in <module>()
----> 1 df.plot(kind='scatter', ax=ax, x='C', y='D', c=df.C, s=100, linewidth=0)

D:\Python27\lib\site-packages\pandas\tools\plotting.pyc in __call__(self, x, y, kind, ax, subplots, sharex, sharey, layout, figsize, use_index, title, grid, legend, style, logx, logy, loglog, xticks, yticks, xlim, ylim, rot, fontsize, colormap, table, yerr, xerr, secondary_y, sort_columns, **kwds)
   3665                           fontsize=fontsize, colormap=colormap, table=table,
   3666                           yerr=yerr, xerr=xerr, secondary_y=secondary_y,
-> 3667                           sort_columns=sort_columns, **kwds)
   3668     __call__.__doc__ = plot_frame.__doc__
   3669 

D:\Python27\lib\site-packages\pandas\tools\plotting.pyc in plot_frame(data, x, y, kind, ax, subplots, sharex, sharey, layout, figsize, use_index, title, grid, legend, style, logx, logy, loglog, xticks, yticks, xlim, ylim, rot, fontsize, colormap, table, yerr, xerr, secondary_y, sort_columns, **kwds)
   2550                  yerr=yerr, xerr=xerr,
   2551                  secondary_y=secondary_y, sort_columns=sort_columns,
-> 2552                  **kwds)
   2553 
   2554 

D:\Python27\lib\site-packages\pandas\tools\plotting.pyc in _plot(data, x, y, subplots, ax, kind, **kwds)
   2378         plot_obj = klass(data, subplots=subplots, ax=ax, kind=kind, **kwds)
   2379 
-> 2380     plot_obj.generate()
   2381     plot_obj.draw()
   2382     return plot_obj.result

D:\Python27\lib\site-packages\pandas\tools\plotting.pyc in generate(self)
    990             self._post_plot_logic_common(ax, self.data)
    991             self._post_plot_logic(ax, self.data)
--> 992         self._adorn_subplots()
    993 
    994     def _args_adjust(self):

D:\Python27\lib\site-packages\pandas\tools\plotting.pyc in _adorn_subplots(self)
   1141                                 naxes=nrows * ncols, nrows=nrows,
   1142                                 ncols=ncols, sharex=self.sharex,
-> 1143                                 sharey=self.sharey)
   1144 
   1145         for ax in self.axes:

D:\Python27\lib\site-packages\pandas\tools\plotting.pyc in _handle_shared_axes(axarr, nplots, naxes, nrows, ncols, sharex, sharey)
   3403                 # as we only have labels in teh first column and we always have a subplot there,
   3404                 # we can skip the layout test
-> 3405                 if ax.is_first_col():
   3406                     continue
   3407                 if sharey or len(ax.get_shared_y_axes().get_siblings(ax)) > 1:

AttributeError: 'Axes' object has no attribute 'is_first_col'

[jorisvandenbossche]

@dragoljub Is there a specific reason that you use add_axes instead of add_subplot?
Because pandas seems to assume that the ax argument is an AxesSubplot and not a plain axes (the is_first_col method is only available on the former)

However, I can't reproduce this issue (using pandas master and matplotlib 1.5). Are you sure the above is a reproducible example? As the place where it fails should only be reached if you have multiple axes (and your example has only one)

[jorisvandenbossche]

But that said, we maybe shouldn't rely on the assumption it is a Subplot ?

@TomAugspurger @sinhrks

[TomAugspurger]

Couldn't reproduce it either. (pandas master, matplotlib 1.5, OS X).

[dragoljub]

@jorisvandenbossche I'm specifically using ax = fig.add_axes((0,0,1,1)) because I'm creating 1000's of plots that must completely fill the plotting area (edge to edge) when png images are generated.

I am able to reproduction it Jupiter Notebook 4.0 (all Windows X64, builds from Christoph Gohlke's UCI repository). I'll do some more digging to confirm this on other machines.

[dragoljub]

@jorisvandenbossche @TomAugspurger I just reproduced it on OSX, Pandas 0.17.0, Matplotlib 1.4.3

Looks like you have to add the cmap= or colormap= argument to see the issue:

df.plot(kind='scatter', ax=ax, x='C', y='D', c=df.C, s=100, linewidth=0, colormap='hsv')

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-3-76da9d6c62ce> in <module>()
----> 1 df.plot(kind='scatter', ax=ax, x='C', y='D', c=df.C, s=100, linewidth=0, colormap='hsv')

/Users/gagi/anaconda/lib/python2.7/site-packages/pandas/tools/plotting.pyc in __call__(self, x, y, kind, ax, subplots, sharex, sharey, layout, figsize, use_index, title, grid, legend, style, logx, logy, loglog, xticks, yticks, xlim, ylim, rot, fontsize, colormap, table, yerr, xerr, secondary_y, sort_columns, **kwds)
   3665                           fontsize=fontsize, colormap=colormap, table=table,
   3666                           yerr=yerr, xerr=xerr, secondary_y=secondary_y,
-> 3667                           sort_columns=sort_columns, **kwds)
   3668     __call__.__doc__ = plot_frame.__doc__
   3669 

/Users/gagi/anaconda/lib/python2.7/site-packages/pandas/tools/plotting.pyc in plot_frame(data, x, y, kind, ax, subplots, sharex, sharey, layout, figsize, use_index, title, grid, legend, style, logx, logy, loglog, xticks, yticks, xlim, ylim, rot, fontsize, colormap, table, yerr, xerr, secondary_y, sort_columns, **kwds)
   2550                  yerr=yerr, xerr=xerr,
   2551                  secondary_y=secondary_y, sort_columns=sort_columns,
-> 2552                  **kwds)
   2553 
   2554 

/Users/gagi/anaconda/lib/python2.7/site-packages/pandas/tools/plotting.pyc in _plot(data, x, y, subplots, ax, kind, **kwds)
   2378         plot_obj = klass(data, subplots=subplots, ax=ax, kind=kind, **kwds)
   2379 
-> 2380     plot_obj.generate()
   2381     plot_obj.draw()
   2382     return plot_obj.result

/Users/gagi/anaconda/lib/python2.7/site-packages/pandas/tools/plotting.pyc in generate(self)
    990             self._post_plot_logic_common(ax, self.data)
    991             self._post_plot_logic(ax, self.data)
--> 992         self._adorn_subplots()
    993 
    994     def _args_adjust(self):

/Users/gagi/anaconda/lib/python2.7/site-packages/pandas/tools/plotting.pyc in _adorn_subplots(self)
   1141                                 naxes=nrows * ncols, nrows=nrows,
   1142                                 ncols=ncols, sharex=self.sharex,
-> 1143                                 sharey=self.sharey)
   1144 
   1145         for ax in self.axes:

/Users/gagi/anaconda/lib/python2.7/site-packages/pandas/tools/plotting.pyc in _handle_shared_axes(axarr, nplots, naxes, nrows, ncols, sharex, sharey)
   3403                 # as we only have labels in teh first column and we always have a subplot there,
   3404                 # we can skip the layout test
-> 3405                 if ax.is_first_col():
   3406                     continue
   3407                 if sharey or len(ax.get_shared_y_axes().get_siblings(ax)) > 1:

AttributeError: 'Axes' object has no attribute 'is_first_col'

[TomAugspurger]

Ok that one failed fails for me, thanks.

Needing colormap is suggestive. I suspect that we think we're only dealing with one axes since that's what is passed in. Then we add a secondary axes to draw the colormap on, so we reach that codepath which assumes there's multiple axes.

[jorisvandenbossche]

Yes, now I can also reproduce it. @dragoljub To get around the bug for now, you can always specify colorbar=False

[dragoljub]

@TomAugspurger @jorisvandenbossche Yes, I just noticed by default Pandas adds the colorbar to the plot even though the argument is optional. In my case I do not want the colorbar but would like to customize the colormap. In that case colorbar=False works.

Thanks!

[sinhrks]

There seems to be some cases which ax doesn't have is_first_col and is_last_col (tested in 1.4.3). Nice to add test cases for them.

import matplotlib.pyplot

fig = plt.figure()
ax = fig.add_subplot(1, 1, 1)
hasattr(ax, 'is_first_col')
# True

# divider
from mpl_toolkits.axes_grid1 import make_axes_locatable
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.05)
hasattr(cax, 'is_first_col')
# False

# insert_ax
from mpl_toolkits.axes_grid.inset_locator import inset_axes
iax = inset_axes(ax, width="30%", height=1., loc=3)
hasattr(iax, 'is_first_col')
# False

# add_axes
aax = fig.add_axes()
hasattr(aax, 'is_first_col')
# False