DOC: Suppress warnings in visualization.rst #8877

sinhrks · 2014-11-22T04:28:48Z

Closes #8234. Sorry to take a long.

I understand there are 2 warnings:

Warning in /Users/sin/Documents/Git/pandas/doc/source/visualization.rst at block ending on line 1275
This is raised from mpl as plotting empty axes for demonstration purpose. Thus simply suppressed.
Warning in /Users/sin/Documents/Git/pandas/doc/source/visualization.rst at block ending on line 1312
This is a warning displayed when passing multiple axes. Added explanations.

jreback · 2014-11-22T16:50:03Z

jorisvandenbossche · 2014-11-24T09:54:06Z

doc/source/visualization.rst


   @savefig frame_plot_subplots_layout.png
-   df.plot(subplots=True, layout=(3, 2), figsize=(6, 6), sharex=False);
+   df.plot(subplots=True, layout=(2, 3), figsize=(6, 6), sharex=False);


Is this change so you see better the empty axes?

The explanation on line 1293 is now no longer correct.

Yes, fixed the text and example.

jorisvandenbossche · 2014-11-30T13:31:05Z

@sinhrks these warnings also appear when running the tests, so probably there are also some of these calls that can be fixed (see https://travis-ci.org/pydata/pandas/jobs/42314766)

jreback · 2014-12-03T22:46:17Z

@sinhrks @jorisvandenbossche this ready to merge?

jorisvandenbossche · 2014-12-03T22:58:55Z

just the tests, it would be good to take a look at that as well

jreback · 2014-12-06T17:12:44Z

@sinhrks can you see if you can fix the test warnings as well?

jreback · 2014-12-08T21:34:14Z

@jorisvandenbossche @sinhrks ?

jreback · 2014-12-11T13:34:25Z

@jorisvandenbossche ?

jorisvandenbossche · 2014-12-11T13:36:50Z

bumbed to 0.16, as you don't visually see it in the docs (warning is not shown by sphinx), so no blocker

sinhrks · 2014-12-20T01:34:49Z

Sure, will fix tests also.

sinhrks · 2015-01-17T03:33:42Z

Updated, and prepared a separate fix #9278 for a program change.

sinhrks · 2015-01-24T01:52:57Z

@jorisvandenbossche OK, now all the warnings should disapear. Pls check.

jorisvandenbossche · 2015-01-26T00:15:11Z

Just looking at the travis log, I still see some warnings. Eg (in the log of the third job: https://travis-ci.org/pydata/pandas/jobs/48128130):

..................S.................................................../home/travis/build/pydata/pandas/pandas/tools/plotting.py:3225: UserWarning: To output multiple subplots, the figure containing the passed axes is being cleared

"is being cleared", UserWarning)

........../home/travis/build/pydata/pandas/pandas/tools/plotting.py:3208: UserWarning: When passing multiple axes, sharex and sharey are ignored.These settings must be specified when creating axes

"These settings must be specified when creating axes", UserWarning)

./home/travis/miniconda/envs/pandas/lib/python2.7/site-packages/matplotlib/axes.py:4747: UserWarning: No labeled objects found. Use label='...' kwarg on individual plots.

warnings.warn("No labeled objects found. "

It is not clear to me if you can see which test triggered the warnings.

jorisvandenbossche · 2015-01-26T00:20:27Z

pandas/tests/test_graphics.py

-
-        returned = df.plot(subplots=True, ax=axes, layout=(-1, 2))
+        with warnings.catch_warnings():
+            warnings.simplefilter('ignore')


Why is the 'catch warnings' needed here? I thought that explicitly stating sharex=False, sharey=False (as you do in this test) is enough to suppress the warning? (that is what you added in the docs?)

Required, because the test specifies layout keyword which can't be effective when ax keyword is passed. I understand these tests was added in #8297, and intend to test layout is ignored properly.

@TomAugspurger ?

ah, that makes sense. Maybe explicitely test that the warning is triggered instead of ignoring it?

@sinhrks that sounds about right, but I could have added both the axes and layout kws without realizing that the layout would be ignored.

jorisvandenbossche · 2015-01-26T00:26:21Z

@sinhrks add some questions

@TomAugspurger can you also give this a brief look?

TomAugspurger · 2015-01-26T01:44:03Z

pandas/tests/test_groupby.py

@@ -4594,9 +4594,6 @@ def test_frame_groupby_plot_boxplot(self):
        _skip_if_mpl_not_installed()

        import matplotlib.pyplot as plt
-        import matplotlib as mpl


Was this causing warnings as well? I think I've seen warnings about trying to set the backend twice...

Yes it did, thus removed. Do you know the reason why these are added originally?

UserWarning: This call to matplotlib.use() has no effect because the the backend has already been chosen; matplotlib.use() must be called *before* pylab, matplotlib.pyplot, or matplotlib.backends is

The reason it causes a warning is because it is called after import matplotlib.pyplot as plt.
I was looking at the test_graphics, and there they also set the matplotlib backend, but in the test to skip it (https://github.com/sinhrks/pandas/blob/plot_doc/pandas/util/testing.py#L180). So you can also put it in the _skip_if_mpl_not_installed here. Only you have to import matplotlib in that test, and not matplotlib.pyplot

There is _skip_if_mpl_not_installed already. And I think matplotlib.pyplot must be installed here, because continuous test case accesses to pyplot.

So current fix (remove problematic part) looks OK if there is no specific reason to keep mpl.use.

I mean you can put the line in the existing _skip_if_mpl_not_installed. So adapt _skip_if_mpl_not_installed to import matplotlib and call mpl.use('Agg'), and then import pyplot here in the testµµ

I also don't exactly know why it is needed (maybe to run the tests where no graphical toolkit is avaliable?), but since all existing tests for plotting do this, I would keep it that way.

@jorisvandenbossche Understood what you mean. How about moving them to test_graphics.py to simplify.

Euhm, I would keep it here, as it really tests the plotting method of the GroupBy object. But maybe you could see if you can eg put the _skip_if_mpl_not_installed in util.testing for reuse.

Currently Groupby.plot is tested mostly under test_graphics.py, and plotting tests in test_groupby.py looks not maintained. Moving remaining to test_graphics is better for maintenance.
https://github.com/pydata/pandas/blob/master/pandas/tests/test_graphics.py#L2015

Separate issue is test_graphics gets larger though.

Ah I didn't know you had already a TestDataFrameGroupByPlots in test_graphics. Yep, certainly OK then to move the tests there.

But is seems a lot of the tests in TestDataFrameGroupByPlots don't really belong there (as it are just Series.plot tests)

Yeah, better to re-organize test_graphics. I'll issue separate one.

sinhrks · 2015-02-04T13:45:08Z

@TomAugspurger Thanks to confirm. I think everything done now.

TomAugspurger · 2015-02-04T14:16:48Z

OK. I've only been able to give it a quick glance, but from that everything looks fine.

jorisvandenbossche · 2015-02-04T14:41:10Z

Regarding my comment above (#8877 (comment)) the No labeled objects found. Use label='...' kwarg on individual plots. is still present in the tests, but this is more a leftover not addressed in #9278, so see if you can look at it

sinhrks · 2015-02-04T15:12:43Z

Yes, will do. I originally misunderstood you're referring to different line.

jorisvandenbossche · 2015-02-11T12:56:38Z

@sinhrks This is ready to merge for you?

In any case, thanks for your endurance, this has been a long one ..

sinhrks · 2015-02-12T14:15:08Z

@jorisvandenbossche Yes. I expect all warnings should be disappears after this and #9464, otherwise will implement a fix in #9464.

DOC: Suppress warnings in visualization.rst

jorisvandenbossche · 2015-02-12T14:23:35Z

Thanks!
Maybe then rebase #9464 to be sure

jreback added Docs Visualization plotting labels Nov 22, 2014

jreback added this to the 0.15.2 milestone Nov 22, 2014

jorisvandenbossche reviewed Nov 24, 2014
View reviewed changes

sinhrks force-pushed the plot_doc branch from 4056efb to 6ff83e8 Compare November 27, 2014 14:57

jorisvandenbossche modified the milestones: 0.16.0, 0.15.2 Dec 11, 2014

sinhrks force-pushed the plot_doc branch from 6ff83e8 to 6326d70 Compare January 17, 2015 01:52

sinhrks mentioned this pull request Jan 17, 2015

BUG: Subplotting boxplot shows unnecessary warnings #9278

Merged

sinhrks force-pushed the plot_doc branch from 6326d70 to c3c048d Compare January 17, 2015 03:32

sinhrks force-pushed the plot_doc branch 2 times, most recently from 36ee23b to 0b75ac9 Compare January 24, 2015 01:13

jorisvandenbossche reviewed Jan 26, 2015
View reviewed changes

TomAugspurger reviewed Jan 26, 2015
View reviewed changes

sinhrks force-pushed the plot_doc branch 2 times, most recently from a6dff05 to f99a8a7 Compare February 3, 2015 12:41

DOC: Suppress warnings in visualization.rst

82744b9

sinhrks force-pushed the plot_doc branch from f99a8a7 to 82744b9 Compare February 11, 2015 10:47

sinhrks mentioned this pull request Feb 11, 2015

BUG: subplots with layout kw may show unnecessary warning #9464

Merged

jorisvandenbossche added a commit that referenced this pull request Feb 12, 2015

Merge pull request #8877 from sinhrks/plot_doc

2f41991

DOC: Suppress warnings in visualization.rst

jorisvandenbossche merged commit 2f41991 into pandas-dev:master Feb 12, 2015

sinhrks deleted the plot_doc branch February 15, 2015 01:33

DOC: Suppress warnings in visualization.rst #8877

DOC: Suppress warnings in visualization.rst #8877

Conversation

sinhrks commented Nov 22, 2014

jreback commented Nov 22, 2014

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jorisvandenbossche commented Nov 30, 2014

jreback commented Dec 3, 2014

jorisvandenbossche commented Dec 3, 2014

jreback commented Dec 6, 2014

jreback commented Dec 8, 2014

jreback commented Dec 11, 2014

jorisvandenbossche commented Dec 11, 2014

sinhrks commented Dec 20, 2014

sinhrks commented Jan 17, 2015

sinhrks commented Jan 24, 2015

jorisvandenbossche commented Jan 26, 2015

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jorisvandenbossche commented Jan 26, 2015

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sinhrks commented Feb 4, 2015

TomAugspurger commented Feb 4, 2015

jorisvandenbossche commented Feb 4, 2015

sinhrks commented Feb 4, 2015

jorisvandenbossche commented Feb 11, 2015

sinhrks commented Feb 12, 2015

jorisvandenbossche commented Feb 12, 2015