Erroneous AssertionError: Index length did not match values on plot

[fonnesbeck]

I have what appears to be a valid Series with a time series index as follows:

In [41]: inactive
Out[41]: 
2010-06-10 20:43:07    37.2
2010-06-21 06:37:28     0.0
2009-07-20 06:53:38     0.0
2012-05-17 01:50:13    27.4
2009-07-27 21:09:15     0.0
2010-05-04 09:06:54     0.0
2010-05-06 03:38:54    32.5
2010-05-09 08:33:16    56.9
2010-05-26 18:58:42     0.0
2010-05-31 18:40:41     0.0
2010-06-13 08:46:17     0.0
2010-06-16 18:27:51     0.0
2010-06-19 11:59:48     0.0
2010-06-24 15:24:26    33.4
2010-06-28 10:54:27     0.0
...
2008-11-02 07:35:22     71.2
2008-11-05 17:33:39     97.6
2009-10-22 23:15:39     75.7
2009-10-27 20:55:07     67.5
2009-06-02 13:57:15     35.8
2009-06-06 14:15:41     25.9
2008-11-07 01:22:46      0.0
2009-10-12 17:08:20    100.0
2010-09-19 10:21:39      0.0
2010-09-24 15:46:50      0.0
2010-09-30 16:48:09      0.0
2010-10-08 21:43:10      0.0
2010-10-15 19:27:11      0.0
2010-10-28 19:10:42      0.0
2010-06-13 10:32:45    100.0
Name: pdgt10, Length: 13536, dtype: float64

Note also:

In [43]: inactive.isnull().sum()
Out[43]: 0

In [45]: inactive.index.size
Out[45]: 13536

In [46]: inactive.values.size
Out[46]: 13536

However, when I try calling inactive.plot() in order to get a time series plot, I receive a puzzling AssertionError:

In [40]: inactive.plot()
---------------------------------------------------------------------------
AssertionError                            Traceback (most recent call last)
<ipython-input-40-cdcfa4d871e5> in <module>()
----> 1 inactive.plot()

/Library/Python/2.7/site-packages/pandas-0.12.0_80_gfcaf9a6_20130729-py2.7-macosx-10.8-intel.egg/pandas/tools/plotting.pyc in plot_series(series, label, kind, use_index, rot, xticks, yticks, xlim, ylim, ax, style, grid, legend, logx, logy, secondary_y, **kwds)
   1730                      secondary_y=secondary_y, **kwds)
   1731 
-> 1732     plot_obj.generate()
   1733     plot_obj.draw()
   1734 

/Library/Python/2.7/site-packages/pandas-0.12.0_80_gfcaf9a6_20130729-py2.7-macosx-10.8-intel.egg/pandas/tools/plotting.pyc in generate(self)
    856         self._compute_plot_data()
    857         self._setup_subplots()
--> 858         self._make_plot()
    859         self._post_plot_logic()
    860         self._adorn_subplots()

/Library/Python/2.7/site-packages/pandas-0.12.0_80_gfcaf9a6_20130729-py2.7-macosx-10.8-intel.egg/pandas/tools/plotting.pyc in _make_plot(self)
   1244             lines = []
   1245             labels = []
-> 1246             x = self._get_xticks(convert_period=True)
   1247 
   1248             plotf = self._get_plot_function()

/Library/Python/2.7/site-packages/pandas-0.12.0_80_gfcaf9a6_20130729-py2.7-macosx-10.8-intel.egg/pandas/tools/plotting.pyc in _get_xticks(self, convert_period)
   1034                 x = index._mpl_repr()
   1035             elif is_datetype:
-> 1036                 self.data = self.data.reindex(index=index.order())
   1037                 x = self.data.index._mpl_repr()
   1038             else:

/Library/Python/2.7/site-packages/pandas-0.12.0_80_gfcaf9a6_20130729-py2.7-macosx-10.8-intel.egg/pandas/core/series.pyc in reindex(self, index, method, level, fill_value, limit, copy, takeable)
   2649 
   2650         # GH4246 (dispatch to a common method with frame to handle possibly duplicate index)
-> 2651         return self._reindex_with_indexers(new_index, indexer, copy=copy, fill_value=fill_value)
   2652 
   2653     def _reindex_with_indexers(self, index, indexer, copy, fill_value):

/Library/Python/2.7/site-packages/pandas-0.12.0_80_gfcaf9a6_20130729-py2.7-macosx-10.8-intel.egg/pandas/core/series.pyc in _reindex_with_indexers(self, index, indexer, copy, fill_value)
   2653     def _reindex_with_indexers(self, index, indexer, copy, fill_value):
   2654         new_values = com.take_1d(self.values, indexer, fill_value=fill_value)
-> 2655         return Series(new_values, index=index, name=self.name)
   2656 
   2657     def reindex_axis(self, labels, axis=0, **kwargs):

/Library/Python/2.7/site-packages/pandas-0.12.0_80_gfcaf9a6_20130729-py2.7-macosx-10.8-intel.egg/pandas/core/series.pyc in __new__(cls, data, index, dtype, name, copy)
    491         else:
    492             subarr = subarr.view(Series)
--> 493         subarr.index = index
    494         subarr.name = name
    495 

/Library/Python/2.7/site-packages/pandas-0.12.0_80_gfcaf9a6_20130729-py2.7-macosx-10.8-intel.egg/pandas/lib.so in pandas.lib.SeriesIndex.__set__ (pandas/lib.c:29092)()

AssertionError: Index length did not match values

As I have shown above, my index length and my values are the same length.

Running a recent (<2 weeks) build from master on OS X 10.8.4 and Python 2.7.2.

[cpcloud]

This has shown up in my attempts to fix some of the period index joining ops.. I'll see if I can track it down

[cpcloud]

i can't reproduce this with the following code

def randts(start, stop, size=1, klass=Series, name=None):
    scalar_mapper = pd.Timestamp
    start, stop = scalar_mapper(start), scalar_mapper(stop)
    mapper = np.vectorize(scalar_mapper)
    values = np.random.random_integers(start.asm8.view(int),
                                       stop.asm8.view(int), size=size)
    return klass(mapper(values), name=name or rands(10))

n = 13536
index = tm.randts(Timestamp('1/1/2009'), Timestamp('1/1/2012'), size=n,
                    name='pdgt10', klass=Index)
values = randn(n)
s = Series(values, index=index)
s.plot()

I get the following plot:

[cpcloud]

I'll try with the commit you're on

[cpcloud]

hm it works on that commit...

[fonnesbeck]

Your example works here. Let me put together an example where it fails.

[cpcloud]

sorry that plot is not showing up my connection is weird.....public wifi is not always so reliable

[fonnesbeck]

OK, here is a failing example.

[cpcloud]

I get all NaN in foo...is that the issue? Just that it should give a better error message?

[cpcloud]

ahhh nevermind

it should be

foo = Series(inactive.pdgt10.values, inactive.st_time.values)

[cpcloud]

i can repro with that

thanks @fonnesbeck

[fonnesbeck]

Right, sorry!

[cpcloud]

strangely (or maybe not) if i sort the index then i get a plot

[cpcloud]

maybe the dupe index value are throwing it off

[cpcloud]

either way i'll fix

[cpcloud]

yep it's the dupe index that is the issue...investigating

[cpcloud]

simple reproducible example (using tm.rands()) from above

n = 10
index = tm.randts(1, n, size=n * 2, klass=Index)
values = randn(index.size)
s = Series(values, index=index)
s.plot()