ValueError on groupby boxplot

[fonnesbeck]

I have a pretty simple data frame of two columns, containing dates and velocities

    velocity    date
0    83.9    2007-08-04
1    89.6    2007-08-04
2    88.2    2007-08-04
3    88.9    2007-08-04
4    89.4    2007-08-04

Yet, when I try to make a box plot of velocities grouped by date, I get a ValueError:

pitches.boxplot(column='velocity', by='date')

ValueError: Wrong number of items passed 109, indices imply 108 

I've confirmed that velocity are floats and date are np.datetime64.

[cpcloud]

Tried to repro this with

In [1]: df = DataFrame({'velocity': normal(88, 0.5, size=10), 
                        'date': [pd.Timestamp('20070804')] * 5 + [pd.Timestamp('2008-08-09')] * 5})

In [2]: df
Out[2]:
        date   velocity
0 2007-08-04  87.622146
1 2007-08-04  87.817566
2 2007-08-04  87.878030
3 2007-08-04  87.823837
4 2007-08-04  88.603917
5 2008-08-09  88.038840
6 2008-08-09  88.591845
7 2008-08-09  87.881469
8 2008-08-09  87.510463
9 2008-08-09  87.527890

[10 rows x 2 columns]

but that works. Can you post ci/print_versions.py?

[fonnesbeck]

Here is print_versions output:

INSTALLED VERSIONS
------------------
commit: None
python: 2.7.5.final.0
python-bits: 64
OS: Darwin
OS-release: 13.0.0
machine: x86_64
processor: i386
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8

pandas: 0.13.0-395-gef55e60
Cython: 0.19.2
numpy: 1.9.0.dev-78e29a3
scipy: 0.14.0.dev-40896a5
statsmodels: 0.6.0.dev-a0fc6d1
IPython: 2.0.0-dev
sphinx: 1.1.3
patsy: None
scikits.timeseries: None
dateutil: 2.2
pytz: 2012d
bottleneck: None
tables: None
numexpr: 2.1
matplotlib: 1.4.x
openpyxl: 1.6.1
xlrd: 0.9.2
xlwt: None
xlsxwriter: None
sqlalchemy: 0.8.2
lxml: None
bs4: None
html5lib: None
bq: None
apiclient: None

FWIW, the boxplot worked using Seaborn:

[cpcloud]

Okay I'll try out an Env with those versions and see if I can reproduce.

[jreback]

maybe a matplotlib 1.4 change?

[jreback]

this was fixed by #6659