ENH/VIS: Plotting DataFrame/Series with matplotlib.table

[sinhrks]

closes #4803

Related to #4803, I prepared some codes to add a matplotlib.Table to DataFrame and Series plot using table keyword. table can take followings as an input:

• False: Do nothing (default).
• True: Draw a table using the data called plot method. Data will be transposed to meet the matplotlib's default layout.
• DataFrame or Series: Draw matplotlib.table using the passed data. The data will be drawn as displayed in print method (not transposed automatically).

Also, helper function pandas.tools.plotting.table is added to create a table from DataFrame and Series, and add it to the matplotlib.Axes.

Example:

fig, axes = plt.subplots(1, 3, figsize=(14, 4))
plt.subplots_adjust(top=0.97, bottom=0.2, left=0.05, right=0.97, hspace=0.2)

df.plot(ax=axes[0], table=True, legend=False)
axes[0].get_xaxis().set_visible(False)

df.plot(ax=axes[1], table=np.round(df.T, 2), legend=False)
axes[1].get_xaxis().set_visible(False)

df.plot(ax=axes[2], legend=False)

import pandas.tools.plotting as plotting
plotting.table(axes[2], np.round(df.describe(), 2),
               loc='upper right', colWidths=[0.2, 0.2, 0.2])

plt.show()

Outputs:

[jreback]

@TomAugspurger pls have a look at this when you get a chance

[jreback]

pls add the release note to v0.14.0 as well

[jreback]

@sinhrks can you rebase this?

[jreback]

also, can you add an example in plotting.rst using this?

[TomAugspurger]

Could we get rid of the transpose argument? It's probably easier for people to do table(ax, data.T) rather than table(ax, data, transpose=True). Fewer arguments too. That should also go along better with the rowLabels and colLabels arguments. If we have transpose=True and rowLabels=None is it the index of the original data or the transposed data? If we get rid of transpose then they pass data.T it's clear what rowLabels refers to.

[TomAugspurger]

Also change rowLables -> row_labels and colLabels -> col_labels for PEP8 style

[sinhrks]

Yes, transpose can be removed. It was for draw horizontal table by series which doesn't have transpose little easily, but it can be done once casted to DataFrame.

rowLabels and colLabels are derived from matplotlib. Thus I think it shouldn't be changed. I'll modify the comment to indicate that these are mpl keywords.

[TomAugspurger]

When you put the example in plotting.rst, maybe note that you can get the table from the axes object with ax.tables. That way they can easily tweak the formatting.

I don't think that we should do much try to automatically format the tables well. I'm guessing this is something that doesn't have a great general solution.

[TomAugspurger]

Let me know when you get those changes and rebased and I'll merge. Thanks and sorry for missing this when you submitted the PR.

[sinhrks]

@TomAugspurger Thanks for your review and comments.

I've added examples to plotting.rst and modified based on your comments except keyword name. As answered in above, these are defined in MPL so I think it is better to use as it is, clarifying these are from MPL.

[TomAugspurger]

Agreed about sticking with the MPL keywords. Good call. Thanks @sinhrks!

[jreback]

hmm....the first example in the docs is not printing the table
is this a version issue? 2.7 builds with matplotlib 1.1.1 (is this too old)?

http://pandas-docs.github.io/pandas-docs-travis/visualization.html#plotting-with-table

[TomAugspurger]

It actually may be a figsize thing. It looks like it is plotting the table but it gets cut off.

[TomAugspurger]

I'm looking into this btw.

[jreback]

@TomAugspurger when you update can you add a link in v0.14.0 to the docs for table plotting? thanks

[TomAugspurger]

I'm not able to reproduce this locally, so I haven't found out what's causing the problem. The good news is that it's not mol 1.1.1 related. I was unable to reproduce it with that also.

These are built by travis right? So there's no way to get access to the machine as its building?

[jreback]

I think you could in your branch push something up with the flag to turn doc builds on and the n have it dump whatever output you need (e.g. just use print statements)