API: consider deprecating DataFrame.hist in favor of DataFrame.plot.hist

[shoyer]

This came up in the PR adding plot submethods like .plot.hist: #9321 (comment)

As long as API differences can be reconciled, I think it would be a nice idea to deprecate (and eventually remove) DataFrame.hist, DataFrame.boxplot and Series.hist in favor of the accessor mtehods.

This old issue discusses API differences (possibly no longer accurate/relevant?): #413

[jreback]

@shoyer @TomAugspurger either one of you for 0.18.0?

[ResidentMario]

Excuse my naivete, but what is to be done here? I did a quick test and all of the DataFrame.plot arguments seem to work when passed into sub-methods (granted, I did not test this extensively, but...).

[jreback]

DataFrame.hist, DataFrame.boxplot and Series.hist in favor of

DataFrame.plot.hist etc

this issue is about emitting a deprecation for these (the replace is already there)

[ResidentMario]

Is this as simple as tossing a bunch of deprecate decorators onto things, then?

[jreback]

@ResidentMario

I think these are actually defined in pandas.tools.plotting.hist_series as a real method. I think that should be removed and instead direct to the plotting accessor.

@TomAugspurger @jorisvandenbossche

[ResidentMario]

Well, the hist_series code does include this comment:

# hack until the plotting interface is a bit more unified

Not a typical comment in pandas source code. 😄 So it does indeed sound like things are a bit more complicated than that.

PS: I think that closing this should also correspond with closing one of the oldest issues still open, #413.

[jreback]

yeah, plotting is like Alice-in-wonderland..........

[jbrockmendel]

Out of curiosity, why are plot, hist, and boxplot defined at the bottom of frames.py instead of included in the class definition? Same goes for Series.plot and Series.hist.

[jreback]

@jbrockmendel I think these are all historical. No reason they should not be deprecated / fixed. (refering to.hist and .boxplot); .plot is by-definition the namespace

[jreback]

anyone want to do this deprecation?

[mangecoeur]

I've noticed that .hist() and .plot.hist() give different outputs (pandas 0.22, matplotlib 2.2) - I opened issue #21048 . I don't know if this happens in all cases but I guess this should be addressed before hist() is replaced by plot.hist()

[TomAugspurger]

Right, part of the deprecation would be providing a way to achieve the behavior of DataFrame.hist within .plot.hist. Copying over your example here.

---

from @mangecoeur in #21048 (comment)

Code Sample, a copy-pastable example if possible

cols = ['heat_total', 'area_density', 'mean_n_floors',
       'n_buildings', 'area_demand_density_grid', 'floor_demand_density']

plt.figure()
df[cols].plot.hist(
    bins=20,
    alpha=0.5
)

plt.figure()
df[cols].hist(
    bins=20,
    alpha=0.5
)

Problem description

df.hist() and df.plot.hist() give different plots: .hist() plots a grid of figures while .plot.hist() plots overlapping distributions.

Result of df.hist()

Result df.plot.hist()

Expected Output

.hist() and .plot.hist() should give exactly the same results

Output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.6.3.final.0
python-bits: 64
OS: Windows
OS-release: 10
machine: AMD64
processor: Intel64 Family 6 Model 94 Stepping 3, GenuineIntel
byteorder: little
LC_ALL: None
LANG: None
LOCALE: None.None

pandas: 0.22.0
pytest: 3.2.1
pip: 9.0.1
setuptools: 36.5.0.post20170921
Cython: 0.26.1
numpy: 1.14.2
scipy: 0.19.1
pyarrow: None
xarray: 0.10.0
IPython: 6.2.1
sphinx: 1.6.3
patsy: 0.4.1
dateutil: 2.6.1
pytz: 2017.2
blosc: None
bottleneck: 1.2.1
tables: 3.4.2
numexpr: 2.6.4
feather: None
matplotlib: 2.2.2
openpyxl: 2.4.9
xlrd: 1.1.0
xlwt: 1.3.0
xlsxwriter: 1.0.2
lxml: 4.1.0
bs4: 4.6.0
html5lib: 0.999999999
sqlalchemy: 1.1.13
pymysql: None
psycopg2: 2.7.4 (dt dec pq3 ext lo64)
jinja2: 2.9.6
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: None