nunique Error when called by groupby from an empty dataframe

[zhangyingmath]

Initialize a dataframe: (I am using pandas version 0.17.1)

import pandas as pd
a=pd.DataFrame({'f':[1,2,3,4]})

First if I call nunique from an empty dataframe, I get what I expect:

a[a['f']>10]['f'].nunique()
0

But If I call nunique from groupby, I get an error:

a[a['f']>10].groupby('f')['f'].nunique()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/opt/ci1/plus/tc10b/sles11sp2_gcc-4.8.3_x86-64/PythonModules-2.7.8/20160310/lib/python/pandas/core/groupby.py", line 2695, in nunique
    return Series(out if ids[0] != -1 else out[1:],
IndexError: index 0 is out of bounds for axis 0 with size 0

On the other hand, the 'unique' function doesn't have this issue:

a[a['f']>10]['f'].unique()
array([], dtype=int64)

a[a['f']>10].groupby('f')['f'].unique()
Series([], Name: f, dtype: float64)

I would like the 'nunique' function to behave like 'unique' when handling an empty dataframe + groupby. (Or please let me know if this issue is already fixed in later versions.) Thanks!

[jreback]

this is a dupe of #12553