BUG: DataFrame.groupby() on tuple column works only when column name is "key"

[dragonator4]

This is the weirdest bug I have seen in Pandas. But I am guessing (hoping) the fix will not be too difficult.

Code Sample

Consider the following two code blocks:

Block 1: key column is called "k"

>>> df1 = pd.DataFrame({'x' : [1,2,3,4,5]*3,
                        'y' : [10,20,30,40,50]*3,
                        'z' : [100,200,300,400,500]*3})
>>> df1['k'] = [(0,0,1),(0,1,0),(1,0,0)]*5

Block 2: key column is called "key"

>>> df2 = pd.DataFrame({'x' : [1,2,3,4,5]*3,
                        'y' : [10,20,30,40,50]*3,
                        'z' : [100,200,300,400,500]*3})
>>> df2['key'] = [(0,0,1),(0,1,0),(1,0,0)]*5

Note that the same, static data is used, so that nothing else may be different, and hence culpable.

Problem description

Running a simple .groupby().describe() operation produces the following results:

>>> df1.groupby('k').describe()
# No Result

>>> df2.groupby('key').describe()
			x		y		z
key				
(0, 0, 1)	count	5.000000	5.000000	5.000000
         	mean	3.000000	30.000000	300.000000
         	std	1.581139	15.811388	158.113883
         	min	1.000000	10.000000	100.000000
         	25%	2.000000	20.000000	200.000000
         	50%	3.000000	30.000000	300.000000
         	75%	4.000000	40.000000	400.000000
         	max	5.000000	50.000000	500.000000
(0, 1, 0)	count	5.000000	5.000000	5.000000
         	mean	3.000000	30.000000	300.000000
         	std	1.581139	15.811388	158.113883
         	min	1.000000	10.000000	100.000000
         	25%	2.000000	20.000000	200.000000
         	50%	3.000000	30.000000	300.000000
         	75%	4.000000	40.000000	400.000000
         	max	5.000000	50.000000	500.000000
(1, 0, 0)	count	5.000000	5.000000	5.000000
         	mean	3.000000	30.000000	300.000000
         	std	1.581139	15.811388	158.113883
         	min	1.000000	10.000000	100.000000
         	25%	2.000000	20.000000	200.000000
         	50%	3.000000	30.000000	300.000000
         	75%	4.000000	40.000000	400.000000
         	max	5.000000	50.000000	500.000000

Note that groupby().mean(), sum(), and a few others work fine. describe() is the only one I think is causing the problem.

Expected Output

Obviously, the expected output for df1.groupby('k').describe() should be the same as df2.groupby('key').describe().

Output of pd.show_versions()

INSTALLED VERSIONS ------------------ commit: None python: 3.5.2.final.0 python-bits: 64 OS: Linux OS-release: 4.4.0-53-generic machine: x86_64 processor: x86_64 byteorder: little LC_ALL: None LANG: en_US.UTF-8 LOCALE: en_US.UTF-8

pandas: 0.19.1
nose: None
pip: 9.0.1
setuptools: 27.2.0
Cython: 0.25.1
numpy: 1.11.2
scipy: 0.18.1
statsmodels: 0.6.1
xarray: 0.8.2
IPython: 5.1.0
sphinx: None
patsy: 0.4.1
dateutil: 2.6.0
pytz: 2016.7
blosc: None
bottleneck: 1.1.0
tables: 3.3.0
numexpr: 2.6.1
matplotlib: 1.5.3
openpyxl: None
xlrd: None
xlwt: None
xlsxwriter: 0.9.4
lxml: None
bs4: None
html5lib: None
httplib2: None
apiclient: None
sqlalchemy: None
pymysql: None
psycopg2: None
jinja2: 2.8
boto: None
pandas_datareader: None

[jorisvandenbossche]

Tuples in columns in not that well supported/tested, but this looks indeed like a bug. Welcome to look into it!

[dragonator4]

It was working perfect in 0.19.0. Updating to 0.19.1 broke my code and this turned out to be the cause. I usually don't work with tuples in columns, but my current project called for it. I had the choice of using three columns to store the three values in the tuple, but since tuples are immutable, I chose to use them.

I have workarounds, so it is not high priority for me.

[Dr-Irv]

There's a correlation between the length of the column name and the number of items in the tuples. So 'key' worked in the example, and I think any other 3 letter name would as well because the tuples were of that length. I will try to fix.