ENH/CLN: add HistPlot class inheriting MPLPlot

[sinhrks]

Because hist and boxplot are separated from normal plot, there are some inconsistencies with these functions. Looks better to include them to MPLPlot framework.

Maybe scatter and hist can be deprecated in 0.15 if MPLPlot can offer better GroupBy plot (plan to do in separate PR).

Example

This allows to use kind='hist in DataFrame.plot and Series.plot. (No changes for DataFrame.hist)

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

df = pd.DataFrame(np.random.randn(1000, 5))
df.plot(kind='hist', subplots=True)

df.plot(kind='hist', stacked=True)

Remaining Items

• Add a release note in API section detailing this change/enhancement
• Modify doc
• Add tests (both histogram and kde)
  • Add support for rot and fontsize (depending on orientation kw) (rely on BUG/VIS: rot and fontsize are not applied to timeseries plots #7844)
  • Add tests for xticklabels and yticklabels
  • Add tests for colors
  • Handling nan
• Check stacked=True can be supported (DataFrame.hist doesn't support it though..).
  • Add tests for stacking

[jreback]

does this change anything? or is just an internal refactor?

what is more consistent?

[jreback]

this independent of #7717 ?

[shoyer]

It looks like the main API change is that as of this PR you can make a histogram with df.plot(kind='hist'), as well as df.hist(). Seems like a win for consistency!

[jreback]

needs to add this to the doc strings for plot_series/frame, no?

[jreback]

@sinhrks also let's add a release note in API section detailing this change/enhancement

[sinhrks]

Will do. Added example and remaining items on top.

[jreback]

is DataFrame.hist now just a call to df.plot(kind='hist')?

[jreback]

change the title on this as well (the PR), no longer a WIP, rigth?

[sinhrks]

Going to leave hist as it is (maybe hist can be deprecate in future). There are some items should be done / considered. Once finished, I'll change the title.

[sinhrks]

I think this can be reviewed (though rot and fontsize processing relies on #7844)

[jreback]

so #7844 should be merged first? or does this include as well?

[sinhrks]

Yes, #7844 should be first.

[jreback]

this looked ok, @sinhrks can you rebase

@TomAugspurger have a look?

[sinhrks]

OK, rebased.

[TomAugspurger]

I just gave it a quick look (didn't test, computer was stolen!) and I think it looks good.

Only comment was maybe setting a random seed in the doc example, very minor.

Nice work!

On Aug 10, 2014, at 16:21, Sinhrks [email protected] wrote:

OK, rebased.

—
Reply to this email directly or view it on GitHub.

[jreback]

@sinhrks hmm, needs rebasing again, then ok to merge

[jreback]

ping when pushed

[sinhrks]

@jreback Rebased and got green.

[jreback]

thanks!